Hameatologica, Volume 107, Issue 1 by Haematologica

haematologica Journal of the Ferrata Storti Foundation

Editor-in-Chief Jacob M. Rowe (Jerusalem)

Deputy Editors Carlo Balduini (Pavia), Jerry Radich (Seattle)

Associate Editors Hélène Cavé (Paris), Monika Engelhardt (Freiburg), Steve Lane (Brisbane), Pier Mannuccio Mannucci (Milan), Pavan Reddy (Ann Arbor), David C. Rees (London), Francesco Rodeghiero (Vicenza), Gilles Salles (New York), Kerry Savage (Vancouver), Aaron Schimmer (Toronto), Richard F. Schlenk (Heidelberg), Sonali Smith (Chicago)

Statistical Consultant Catherine Klersy (Pavia)

Editorial Board Walter Ageno (Varese), Sarit Assouline (Montreal), Andrea Bacigalupo (Roma), Taman Bakchoul (Tübingen), Pablo Bartolucci (Créteil), Katherine Borden (Montreal), Marco Cattaneo (Milan), Corey Cutler (Boston), Kate Cwynarski (London), Mary Eapen (Milwaukee), Francesca Gay (Torino), Ajay Gopal (Seattle), Alex Herrera (Duarte), Shai Izraeli (Ramat Gan), Martin Kaiser (London), Marina Konopleva (Houston), Johanna A. Kremer Hovinga (Bern), Nicolaus Kröger (Hamburg), Austin Kulasekararaj (London), Shaji Kumar (Rochester), Ann LaCasce (Boston), Anthony R. Mato (New York), Neha Mehta-Shah (St. Louis), Alison Moskowitz (New York), Yishai Ofran (Haifa), Farhad Ravandi (Houston), John W. Semple (Lund), Liran Shlush (Toronto), Sara Tasian (Philadelphia), Pieter van Vlieberghe (Ghent), Ofir Wolach (Haifa), Loic Ysebaert (Toulouse)

Managing Director Antonio Majocchi (Pavia)

Editorial Office Lorella Ripari (Office & Peer Review Manager), Simona Giri (Production & Marketing Manager), Paola Cariati (Graphic Designer), Giulia Carlini (Graphic Designer), Igor Poletti (Graphic Designer), Marta Fossati (Peer Review), Diana Serena Ravera (Peer Review) , Laura Sterza (Account Administrator)

Assistant Editors Britta Dorst (English Editor), Rachel Stenner (English Editor), Bertie Vitry (English Editor), Massimo Senna (Information technology), Idoya Lahortiga (Graphic artist)

haematologica Journal of the Ferrata Storti Foundation

Brief information on Haematologica Haematologica (print edition, pISSN 0390-6078, eISSN 1592-8721) publishes peer-reviewed papers on all areas of experimental and clinical hematology. The journal is owned by a non-profit organization, the Ferrata Storti Foundation, and serves the scientific community following the recommendations of the World Association of Medical Editors (www.wame.org) and the International Committee of Medical Journal Editors (www.icmje.org). Haematologica publishes Editorials, Original articles, Review articles, Perspective articles, Editorials, Guideline articles, Letters to the Editor, Case reports & Case series and Comments. Manuscripts should be prepared according to our guidelines (www.haematologica.org/information-for-authors), and the Uniform Requirements for Manuscripts Submitted to Biomedical Journals, prepared by the International Committee of Medical Journal Editors (www.icmje.org). Manuscripts should be submitted online at http://www.haematologica.org/. Conflict of interests. According to the International Committee of Medical Journal Editors (http://www.icmje.org/#conflicts), “Public trust in the peer review process and the credibility of published articles depend in part on how well conflict of interest is handled during writing, peer review, and editorial decision making”. The ad hoc journal’s policy is reported in detail at www.haematologica.org/content/policies. Transfer of Copyright and Permission to Reproduce Parts of Published Papers. Authors will grant copyright of their articles to the Ferrata Storti Foundation. No formal permission will be required to reproduce parts (tables or illustrations) of published papers, provided the source is quoted appropriately and reproduction has no commercial intent. Reproductions with commercial intent will require written permission and payment of royalties. Subscription. Detailed information about subscriptions is available at www.haematologica.org. Haematologica is an open access journal and access to the online journal is free. For subscriptions to the printed issue of the journal, please contact: Haematologica Office, via Giuseppe Belli 4, 27100 Pavia, Italy (phone +39.0382.27129, fax +39.0382.394705, E-mail: info@haematologica.org). Rates of the printed edition for the year 2021 are as following: Institutional: Euro 700 Personal: Euro 170 Advertisements. Contact the Advertising Manager, Haematologica Office, via Giuseppe Belli 4, 27100 Pavia, Italy (phone +39.0382.27129, fax +39.0382.394705, e-mail: marketing@haematologica.org). Disclaimer. Whilst every effort is made by the publishers and the editorial board to see that no inaccurate or misleading data, opinion or statement appears in this journal, they wish to make it clear that the data and opinions appearing in the articles or advertisements herein are the responsibility of the contributor or advisor concerned. Accordingly, the publisher, the editorial board and their respective employees, officers and agents accept no liability whatsoever for the consequences of any inaccurate or misleading data, opinion or statement. Whilst all due care is taken to ensure that drug doses and other quantities are presented accurately, readers are advised that new methods and techniques involving drug usage, and described within this journal, should only be followed in conjunction with the drug manufacturer’s own published literature.

Direttore responsabile: Prof. Carlo Balduini; Autorizzazione del Tribunale di Pavia n. 63 del 5 marzo 1955. Printing: Press Up, zona Via Cassia Km 36, 300 Zona Ind.le Settevene - 01036 Nepi (VT)

Associated with USPI, Unione Stampa Periodica Italiana. Premiato per l’alto valore culturale dal Ministero dei Beni Culturali ed Ambientali

haematologica Journal of the Ferrata Storti Foundation

Table of Contents Volume 107, Issue 1: January 2022 About the Cover 1

Images from the Haematologica Atlas of Hematologic Cytology: Sézary syndrome Rosangela Invernizzi

https://doi.org/10.3324/haematol.2021.280262

Editorial 2

Toward further excellence in Haematologica Jacob M. Rowe et al.

https://doi.org/10.3324/haematol.2021.280017

Landmark Papers in Hematology 3

The “7+3” regimen in acute myeloid leukemia Jacob M. Rowe

https://doi.org/10.3324/haematol.2021.280161

Review Series 4

Indolent lymphomas: introduction to a series highlighting progress and ongoing challenges Sonali M. Smith and Gilles Salles

https://doi.org/10.3324/haematol.2021.280218

Time for an individualized approach to first-line management of follicular lymphoma Guillaume Cartron and Judith Trotman https://doi.org/10.3324/haematol.2021.278766

Prospects in the management of patients with follicular lymphoma beyond first-line therapy David Qualls and Gilles Salles

https://doi.org/10.3324/haematol.2021.278717

Marginal zone lymphoma: present status and future perspectives Chan Y. Cheah et al.

https://doi.org/10.3324/haematol.2021.278755

Editorials 44

Fusion genes in acute myeloid leukemia: do acute myeloid leukemia diagnostics need to fuse with RNA-sequencing? Felicitas Thol

https://doi.org/10.3324/haematol.2021.278983

Adding eltrombopag to immunosuppression: the importance of predicting outcome André Tichelli et al.

https://doi.org/10.3324/haematol.2021.278761

Articles Acute Lymphoblastic Leukemia 49 Pre-existing antibodies against polyethylene glycol reduce asparaginase activities on first administration of pegylated E. coli asparaginase in children with acute lymphocytic leukemia Alaeddin Khalil et al.

https://doi.org/10.3324/haematol.2020.258525

Haematologica 2022; vol. 107 no. 1 - January 2022 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

Acute Myeloid Leukemia 58 Targeting MCL-1 dysregulates cell metabolism and leukemia-stroma interactions and re-sensitizes acute myeloid leukemia to BCL-2 inhibition Bing Z. Carter et al.

https://doi.org/10.3324/haematol.2020.260331

A genome-wide CRISPR screen identifies regulators of MAPK and MTOR pathways that mediate resistance to sorafenib in acute myeloid leukemia Alisa Damnernsawad et al.

https://doi.org/10.3324/haematol.2020.257964

Epigenetic changes in human model KMT2A leukemias highlight early events during leukemogenesis Thomas Milan et al.

https://doi.org/10.3324/haematol.2020.271619

100

Fusion gene detection by RNA-sequencing complements diagnostics of acute myeloid leukemia and identifies recurring NRIP1-MIR99AHG rearrangements Paul Kerbs et al.

https://doi.org/10.3324/haematol.2021.278436

Blood Transfusion 112 Beta thalassemia minor is a beneficial determinant of red blood cell storage lesion Vassilis L. Tzounakas et al.

https://doi.org/10.3324/haematol.2020.273946

Bone Marrone Failure 126 Predicting response of severe aplastic anemia to immunosuppression combined with eltrombopag Yoshitaka Zaimoku et al.

https://doi.org/10.3324/haematol.2021.278413

Chronic Lymphocytic Leukemia 134 The impact of early discontinuation/dose modification of venetoclax on outcomes in patients with relapsed/refractory chronic lymphocytic leukemia: post-hoc analyses from the phase III MURANO study Anthony R. Mato et al.

https://doi.org/10.3324/haematol.2020.266486

143

The miR-200c/141-ZEB2-TGFβ axis is aberrant in human T-cell prolymphocytic leukemia Stefan J. Erkeland et al.

https://doi.org/10.3324/haematol.2020.263756

Hematopoiesis 154 Loss of Nupr1 promotes engraftment by tuning the quiescence threshold of hematopoietic stem cells via regulation of the p53-checkpoint pathway Tongjie Wang et al.

https://doi.org/10.3324/haematol.2019.239186

167

Human erythroid differentiation requires VDAC1-mediated mitochondrial clearance Martina Moras et al.

https://doi.org/10.3324/haematol.2020.257121

Myelomonocytic Leukemia 178 Molecular and phenotypic diversity of CBL-mutated juvenile myelomonocytic leukemia Anna Hecht et al.

https://doi.org/10.3324/haematol.2020.270595

Non Hodgkin-Lymphoma 187 Micro-RNA networks in T-cell prolymphocytic leukemia reflect T-cell activation and shape DNA damage response and survival pathways Till Braun et al.

https://doi.org/10.3324/haematol.2020.267500

Haematologica 2022; vol. 107 no. 1 - January 2022 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

201

Phenogenomic heterogeneity of post-transplant plasmablastic lymphomas Rebecca J. Leeman-Neill et al.

https://doi.org/10.3324/haematol.2020.267294

211

Shallow-depth sequencing of cell-free DNA for Hodgkin and diffuse large B-cell lymphoma (differential) diagnosis: a standardized approach with underappreciated potential Lennart Raman et al.

https://doi.org/10.3324/haematol.2020.268813

221

Baseline SUVmax is related to tumor cell proliferation and patient outcome in follicular lymphoma Cédric Rossi et al.

https://doi.org/10.3324/haematol.2020.263194

Plasma Cell Disorders 231 Autologous stem cell transplantation is safe and effective for fit, older myeloma patients: exploratory results from the Myeloma XI trial Charlotte Pawlyn et al.

https://doi.org/10.3324/haematol.2020.262360

Platelet Biology & its Disorders 243 Post-translational polymodification of β1-tubulin regulates motor protein localization in platelet production and function Abdullah O. Khan et al.

https://doi.org/10.3324/haematol.2020.270793

260

Dysregulation of oncogenic factors by GFI1B p32: investigation of a novel GFI1B germline mutation Michela Faleschini et al.

https://doi.org/10.3324/haematol.2020.267328

Red Cell Biology & its Disorders 268 The epigenetic regulator RINF (CXXC5) maintains SMAD7 expression in human immature erythroid cells and sustains red blood cell expansion Audrey Astori et al.

https://doi.org/10.3324/haematol.2020.263558

Letters to the Editor 284

Diabetes mellitus and risk of plasma cell and lymphoproliferative disorders in 94,579 cases and 368,348 matched controls Urvi A. Shah et al.

https://doi.org/10.3324/haematol.2021.278772

287

https://doi.org/10.3324/haematol.2021.279389

292

Vecabrutinib inhibits B-cell receptor signal transduction in chronic lymphocytic leukemia cell types with wild-type or mutant Bruton tyrosine kinase Burcu Aslan et al.

https://doi.org/10.3324/haematol.2021.279158

298

Stored blood has compromised oxygen unloading kinetics that can be normalized with rejuvenation and predicted from corpuscular side-scatter Killian Donovan et al.

https://doi.org/10.3324/haematol.2021.279296

303

Elastic compression stockings for prevention of the post-thrombotic syndrome in patients with and without residual vein thrombosis and/or popliteal valve reflux Paolo Prandoni et al.

https://doi.org/10.3324/haematol.2021.279680

307

Targeting Wnt signaling in acute myeloid leukemia stem cells Felice Pepe et al.

https://doi.org/10.3324/haematol.2020.266155

Haematologica 2022; vol. 107 no. 1 - January 2022 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

312

In utero thirdhand smoke exposure modulates platelet function in a sex-dependent manner Hamdy E. A. Ali et al.

https://doi.org/10.3324/haematol.2021.279388

317

Ikaros deficiency is associated with aggressive BCR-ABL1 B-cell precursor acute lymphoblastic leukemia independent of the lineage and developmental origin Célestine Simand et al.

https://doi.org/10.3324/haematol.2021.279125

321

Planned withdrawal of dexamethasone after pomalidomide low-dose dexamethasone induction for lenalidomide-refractory multiple myeloma (ALLG MM14) Anna Kalff et al.

https://doi.org/10.3324/haematol.2021.278655

326

Signal transduction pathway involved in platelet activation in immune thrombotic thrombocytopenia after COVID-19 vaccination Roberta Misasi et al.

https://doi.org/10.3324/haematol.2021.279729

330

First report of inherited protein S deficiency caused by paternal PROS1 mosaicism Satomi Nagaya et al.

https://doi.org/10.3324/haematol.2021.278527

334

Treatment with ibrutinib does not induce a TP53 clonal evolution in chronic lymphocytic leukemia Luciana Cafforio et al.

https://doi.org/10.3324/haematol.2020.263715

338

A novel index to evaluate ineffective erythropoiesis in hematological diseases offers insights into sickle cell disease John Brewin et al.

https://doi.org/10.3324/haematol.2021.2021.279623

342

Tumor suppressor function of WT1 in acute promyelocytic leukemia Matthew J. Christopher, et al.

https://doi.org/10.3324/haematol.2021.279601

Case Report 347

Increased double-negative αβ+ T cells reveal adult-onset autoimmune lymphoproliferative syndrome in a patient with IgG4-related disease Nivaz Brar et al.

https://doi.org/10.3324/haematol.2021.279297

Haematologica 2022; vol. 107 no. 1 - January 2022 http://www.haematologica.org/

ABOUT THE COVER Images from the Haematologica Atlas of Hematologic Cytology: Sézary syndrome Rosangela Invernizzi University of Pavia, Pavia, Italy E-mail: ROSANGELA INVERNIZZI - rosangela.invernizzi@unipv.it doi:10.3324/haematol.2021.280262

ézary syndrome is the leukemic variant of a cutaneous mature T-cell lymphoma, characterized by achronic course, and by the triad of diffuse erythroderma, lymphadenopathy, and neoplastic T-lymphocytes in the blood without significant involvement of the bone marrow. Sézary cells show typically striking nuclear convolutions which may give the nucleus a cerebriform appearance, condensed chromatin and sometimes evident nucleoli; in some cells the agranular cytoplasm contains a ring of vacuoles (Figure A-D). The cytoplasmic vacuoles are strongly positive for periodic acid Schiff stain (Figure E, F). Sézary cells show focal reactivity for acid hydrolases, a mature T-cell, CD4+, often aberrant phenotype and monoclonal rearrangement of TCR genes. In addition to the above reported triad, one or more of the following criteria are required for diagnosis: an absolute Sézary cell count ≥1,000/mL, an expanded CD4+ T-cell population with a CD4:CD8 ratio of ≥10, and loss of one or more T-cell antigens.1 Disclosures No conflicts of interest to disclose

Reference 1. Invernizzi R. Mature T- and NK-cell neoplasms. Haematologica. 2020; 105(Suppl 1):162-170.

haematologica | 2022; 107(1)

EDITORIALS Toward further excellence in Haematologica Jacob M. Rowe,1 Carlo Balduini2 and Jerry Radich3 1

Shaare Zedek Medical Center, Jerusalem, Israel; 2Fondazione Ferrata Storti, Pavia, Italy and 3Fred Hutchinson Cancer Research Center, Seattle, WA, USA E-mail: JACOB M. ROWE - rowe@jimmy.harvard.edu doi:10.3324/haematol.2021.280017

s we enter 2022, having come through another difficult year with COVID-19, we mourn the loss of colleagues, as well as many others in the healthcare field. The pandemic has had consequences for us in many ways, affecting all aspects of the journal, including authors, reviewers and editors. At the same time we are fortunate to acknowledge that Haematologica has been blessed by an outstanding group of associate editors as well as a totally dedicated editorial team, even allowing us to look back at this period with some humble pride. As a leading journal in hematology, we continue to strive for excellence and underscore this in the quality of our publications. Our priority remains publishing high quality manuscripts with an emphasis on originality and impact in the international community. In the clinical arena, we prioritize prospectively designed studies addressing current needs, although exceptional retrospective studies or meta-analyses are also considered. At the same time, we continue to encourage publication of advanced basic research in both malignant and benign hematology. As all considered submissions undergo rigorous prioritization and peer review, unfortunately the current reality is that some very good papers may not be published. In addition to regular articles, the journal publishes reviews, perspectives and editorials covering impactful areas of current interest. Most of these are invited manuscripts, although unsolicited submissions can exceptionally be considered. Letters to the journal are also encouraged. These are high quality manuscripts of novel or preliminary reports and undergo the same rigorous standard of peer review. Several innovations were introduced in 2021, all with the hope of further stimulating interest in Haematologica. In this issue we start an exceptional review series led by Drs. Gilles Salles and Sonali Smith, which will include three important articles covering the major aspects of indolent lymphomas. The plan is for this to be followed in the first part of 2022 by another review series comprising classic contributions on the treatment of different thrombocytopenias. This will be headed by Drs. Carlo Balduini and Francesco Rodeghiero. Towards the middle of the year, Dr. Jerry Radich will be lead-

ing another review series on measurable residual disease in acute myeloid leukemia, acute lymphoblastic leukemia, chronic myeloid leukemia and chronic lymphoblastic leukemia, with several outstanding articles. Finally, down the line, toward the end of the year, we expect to have a major review series on acute myeloid leukemia, which will be led by Dr. Richard Schlenk. As is fitting for the oldest existing hematology journal, we are also introducing an innovation paying homage to classic publications from previous generations. Starting with this issue, each month we will be publishing a short report entitled "Landmark Papers in Hematology" which will describe and highlight historic contributions made by past authors. An editor or an editorial board member will be responsible for each submission. On the subject of the editorial board, we have completely replaced our previous board with more than 30 new members who actively assist in reviewing papers, suggesting topics for publication, and promoting the quality of the journal. Furthermore, to help our editors and reviewers, and in fairness to the authors, we are fortunate to have recruited a highly qualified statistical consultant. Finally, since the journal’s format does matter, we hope that you will like our new graphic design, expected in the next few months, which will facilitate the assimilation of the scientific data by readers. Over the past year, we have had considerable success in reducing the backlog of papers that, although published online as 'early view' within days of acceptance, had been waiting for up to nine months or even longer to enter the printed journal. This time gap has now been drastically reduced by over one third. Haematologica faces many challenges, not least of which is the abundance of new academic outlets, both digital and various printed products. While recognizing the magnitude of these challenges, we are prepared to meet them head-on with enthusiasm and provide the highest quality hematology journal for our dedicated community of readers and tireless authors, editors and reviewers. The editors will always welcome constructive ideas or feedback from our readers.

haematologica | 2022; 107(1)

LANDMARK PAPERS IN HEMATOLOGY The “7+3” regimen in acute myeloid leukemia Jacob M. Rowe Shaare Zedek Medical Center, Jerusalem, Israel E-mail: rowe@jimmy.harvard.edu doi:10.3324/haematol.2021.280161

TITLE

Cytosine arabinoside (NSC-63878) and daunorubicin (NSC-83142) therapy in acute nonlymphocytic leukemia

AUTHORS

Yates JW, Wallace HJ Jr, Ellison RR, Holland JF

JOURNAL

Cancer Chemother Rep. 1973;57(4):485-8. PMID: 4586956

owadays, the majority of patients with acute myeloid leukemia (AML) achieve a complete remission and many are cured. This remarkable achievement takes place with a standardized induction regimen consisting of an anthracycline and cytarabine. The landmark paper in 1973 – from the prestigious USA group of James Holland at the Roswell Park Memorial Institute in Buffalo, New York – reported on an astounding complete remission rate of 63% among patients with both previously-treated and previouslyuntreated AML. The drugs used were the anthracycline, daunorubicin (formerly rubidomycin in France and daunomycin in the USA), and the pyrimidine antimetabolite, cytosine arabinoside. For a decade prior to this work, each of these drugs was shown to have potent anti-leukemic activity as single agents1,2 and even combinations of these two drugs had been reported.3 In all cases only a small number of remissions

were achieved and they were of very short duration. The ultimate breakthrough came, not with the addition of new drugs, but after longer exposure of the leukemic cells to the combination of daunorubicin and cytarabine, as cytosine arabinoside is now known. This breakthrough combination and schedule of drugs for AML, widely known as the “7 and 3” regimen, dramatically changed the prognosis of patients, has remained the backbone of standard therapy for close to five decades, and, amazingly, is still going strong. This remarkable development has been at the core of the current constantly improving outlook for newly-diagnosed patients with AML. From having an incurable disease in the early 1970s, approximately 35-40% of young adults can now expect to be long-term survivors – after starting with “7+3” induction therapy. Furthermore, and quite unthinkably even two decades ago, patients tolerate this combination well into their seventies.

References 1. Ellison RR, Holland JF, Weil M, et al. Arabinosyl cytosine: a useful agent in the treatment of acute leukemia in adults. Blood. 1968;32(4):507-523. 2. Bernard J, Weil M, Boiron M, Jacquillat C, Flandrin G, Gemon MF. Acute promyelocytic leukemia: results of treatment by daunorubicin. Blood. 1973;41(4):489-496. 3. Crowther D, Powles RL, Bateman CJ, et al. Management of adult acute myelogenous leukaemia. Br Med J. 1973;1(5846):131-137.

haematologica | 2022; 107(1)

REVIEW SERIES Ferrata Storti Foundation

Indolent lymphomas: introduction to a series highlighting progress and ongoing challenges Sonali M. Smith1 and Gilles Salles2 1 Section of Hematology/Oncology, The University of Chicago, Chicago, IL and 2Memorial Sloan Kettering Cancer Center, New York, NY, USA

Haematologica 2022 Volume 107(1):4-6

Introduction Indolent lymphomas collectively comprise up to one-third of all non-Hodgkin lymphomas, with an ever-increasing number of survivors. Despite the indolent course, patients are subject to repeated relapses and remissions, and exposed to a recurrent need for treatment and intervention. The ultimate challenge with these lymphomas is that cure remains elusive, and the impact of the disease on quality and quantity of life is significant; a portion of patients will die from either disease or treatment-associated toxicity, and lymphoma-related mortality remains the most common cause of death.1 Furthermore, transformation to a more aggressive histology can occur, and drastically shifts the management approach and prognosis. Historically, the treatment approach was limited to chemotherapy or radiotherapy, and median survival times were less than a decade, likely reflecting intrinsic resistance of indolent biology to agents requiring high cell turnover for optimal effect. The past few years have witnessed significant advances in biological insights, targeted therapies, and prolonged survival for most patients. This issue of Haematologica contains a timely series on indolent lymphomas, with three comprehensive papers2-4 presented by experts in the field; the series is intended to provide an update on where the field is going in terms of new data and new perspectives, while also outlining challenges that remain.

Follicular lymphoma

Correspondence: SONALI M. SMITH smsmith@medicine.bsd.uchicago.edu Received: October 19, 2021. Accepted: October 19, 2021. https://doi.org/10.3324/haematol.2021.280218

©2022 Ferrata Storti Foundation Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

As the first two papers highlight, sequential advances have dramatically improved survival for patients with follicular lymphoma (FL) which now borders on several decades for most individuals. We have finally come to a point at which most FL patients will die with lymphoma rather than from lymphoma. Unfortunately, determining individual prognosis at initial diagnosis is not yet feasible; as Cartron and Trotman2 discuss, prognostic indices at baseline (FLIPI, FLIPI2, M7-FLIPI, PRIMA-PI) are better suited to groups of patients rather than to individual patients. Furthermore, the type of treatment influences the robustness of these indices and further limits their applications. The authors also raise the limitations of the Groupe d'Etude des Lymphomes Folliculaires (GELF) criteria, and suggest opportunities for functional imaging with positron emission tomography and computed tomography for decision-making. Overall, we are unable to advise patients at the time of diagnosis on predicted outcome. Correcting this deficit is important for two reasons: we must identify those destined to do well, for whom minimal exposure to therapy is sufficient, and also identify patients with a shortened median survival for whom new or aggressive treatments are not only justified, but badly needed. For the former group, primum non nocere comes to mind, and watch and wait (or, as the authors elegantly state: “therapeutic abstention with dynamic observation”) remains the standard of care. At the other extreme are patients with rapid kinetic failure who are only identified after relapse, with two recently defined high-risk subgroups: early progression (defined as relapse within either 12 months or 24 months of initial chemoimmunotherapy) and double-refractory disease (typically defined as relapse within 6 months despite treatment with both alkylating agents and anti-CD20 monoclonal antibodies). These groups are appropriately the focus for several clinical trials, with the US Intergroup S1608 being one example of a trial specifically comparing chemo-immunotherapy against two biological doublets in patients with progression of disease within 24 months of initial chemo-immunotherapy (NCT03269669). Adding to the complexity, there is a middle group of patients,

haematologica | 2022; 107(1)

Introduction to series on indolent lymphomas

described by Qualls and Salles,3 in whom relapse is neither early nor yet refractory. The challenge here is that we have a myriad of options ranging from anti-CD20 antibodies alone, chemo-immunotherapy, immunomodulatory agents, phosphoinositide-3 kinase inhibitors, an EZH2 inhibitor, transplantation and even cellular therapy. We have no data on sequencing, minimal comparative data, and no inclusion of precision approaches with the exception of better outcomes for patients with EZH2mutated FL treated with tazemetostat (as compared to patients without the EZH2 mutation). As the authors saliently note, this plethora of agents in the second line and beyond is largely marked by phase II data, gained from studies with heterogeneous inclusion criteria, and no indication on whether one treatment might preclude the use of another later on. Furthermore, although Haematologica has an international readership, it is important to acknowledge that many of the options presented will not be available to patients in the majority of the world and, even within wealthy nations, not all patients will have equal access. Qualls and Salles also highlight that newer options should not mean that the old ones have no place; as discussed in their review, radiotherapy and/or chemotherapy in the relapsed/refractory setting, while not typically preferred when alternatives are available, is valuable. An additional challenge is that many of these treatment options were designed for indefinite treatment and there is no guidance on when it is appropriate to stop treatment or consider a treatment holiday. Extending this dilemma of an expanding toolbox without data on how to best sequence or select a specific treatment is the challenge presented by cellular therapy, or chimeric antigen receptor T-cell therapy (CAR-T). This exciting modality engineers autologous T cells to recognize CD19 on tumor cells with subsequent T-cell activation, expansion, and tumor eradication. The attendant toxicities include cytokine release syndrome and neurotoxicity, and the “financial toxicity” is considerable. Nevertheless, CAR-T has dramatically changed the algorithm for aggressive lymphomas and offers the potential for durable remission and possibly cure; its precise role in indolent lymphomas is unknown, but CAR-T with axicabtagene ciloleucel is now approved for FL. As discussed by Qualls and Salles, in the ZUMA-5 trial the response rate to CAR-T was 94% (with a complete response rate of 80%) in heavily pretreated FL patients although the follow up is still less than 2 years. A careful review of the inclusion criteria in both the ZUMA-5 and ELARA trials shows that many patients had early progression, double refractory disease, refractory disease to the most recent treatment, and a high baseline FLIPI score.5,6 Clinicians will be asking, when should a patient with indolent lymphoma be referred for this therapy? To date, there is no consensus on this topic, but following the inclusion criteria of the published studies is reasonable. An important aspect to furthering precision medicine in indolent lymphomas is understanding the biology and pathogenesis of the diseases, with an aim of identifying therapeutic vulnerabilities. FL is a complex and biologically heterogeneous disease, but recent insights provide valuable opportunities with several emerging themes.7-9 First, the t(14;18) rearrangement, long known as a hallmark lesion, is insufficient for FL and is present in a significant portion of the normal population. Second, FL progression occurs via divergent evolution from a common haematologica | 2022; 107(1)

progenitor cell. This critical observation may underlie the inability of our current tools to eradicate (or cure!) the disease. A reservoir population of common progenitor cells may account for the eventual resurgence of disease, even when there is clinical evidence of a complete remission. Third, sequencing of large databases of FL patients has identified epigenetic deregulation as a consistent and early event, with mutations involving KMT2D, CREBBP, H1 histones, EZH2, and EP300 collectively occurring in the majority of patients. Finally, FL biology and clinical behavior appear heavily reliant on the tumor microenvironment and immune composition. While these areas of FL biology under scrutiny are listed here, there are clearly other areas in active development, and the field is in an exciting phase as we contemplate the next generation of therapies.

Marginal zone lymphoma The third paper in this series focuses on marginal zone lymphomas (MZL), a complex and heterogeneous group of diseases. MZL is a paradigmatic disease to study the role of antigenic stimuli, inflammation, and immune “irritation” leading to the development of cancer; the prototypic antigen-driven lymphoma is Helicobacter pylori-associated extranodal MZL of gastric lymphoid tissue. As described in the article by Cheah and colleagues,4 the three main clinicopathological entities have both common and distinct features. For example, deregulation of epigenetic pathways and NOTCH2 signaling are seen across multiple MZL subsets, while mutations such as PTPRD are restricted to nodal MZL.10 The treatment for MZL remains largely based on the site of clinical involvement, and is often influenced by issues of local control. A major gap in MZL is the relative paucity of disease-specific therapeutic trials; most treatment data for disseminated disease are derived from trials of FL patients. However, as many trials suggest, MZL is biologically distinct and often has disparate outcomes compared to FL; this argues for disease-specific trials and not simply lumping MZL into other trials of indolent lymphomas. An example of this observation is quite clear in cellular therapy trials in which it appears that the CAR-T approach leading to regulatory approval for FL had minimal activity in MZL.6 Despite these caveats, outcomes are generally favorable with current guidelines and a focus on short- and long-term toxicity is an essential element of the therapeutic approach. Ideally, the emerging mutational landscape and insights into the often inflammatory microenvironment will lead to more targeted therapies in the future.

Concluding remarks When reading these reviews side-by-side, a number of common themes and struggles emerge. On a positive note, there has been dramatic progress in terms of survival and a general shift away from cytotoxic agents. A relatively new concept is that not all relapses are the same. However, several common challenges persist: inability to determine individual prognosis at the time of diagnosis, lack of data on optimal sequencing, no clear biologically driven indication for treatment selection, lim5

S.M. Smith and G. Salles

ited guidance on surveillance strategies that balance risks and benefits of repeated radiological imaging, and, of course, cure remaining elusive. A clear modern-day obstacle is the ongoing COVID-19 pandemic which must be considered within the decision-making process. The negative impact of anti-lymphoma therapy on immunity puts our patients at higher risk of morbidity and mortality. There are now emerging data that rituximab or antiCD20 antibody therapy interferes with a robust antibody response to the vaccine for approximately 12 months,8 and the use of agents such as bendamustine suppresses

References 1. Sarkozy C, Maurer MJ, Link BK, et al. Cause of death in follicular lymphoma in the first decade of the rituximab era: a pooled analysis of French and US cohorts. J Clin Oncol. 2019;37(2):144-152. 2. Cartron G, Trotman J. Time for an individualized approach to first-line management of follicular lymphoma. Haematologica. 2022;107(1):7-18. 3. Qualls D, Salles G. Prospects in the management of patients with follicular lymphoma beyond first-line therapy. Haematologica. 2022;107(1):19-34. 4. Cheah CY, Zucca E, Rossi D, Habermann TM. Marginal zone lymphoma: present status and future perspectives. Haematologica. 2022;107(1):35-43

both B- and T-cell function. The discussion on risks versus benefits of treatment and treatment selection has never been so crucial. This first series on indolent lymphomas artfully presents the progress to date in detail, with discussion of the many challenges that remain. Disclosures No conflicts of interest to disclose. Contributions The authors contributed equally.

5. Schuster SJ, Dickinson MJ, Dreyling M, et al. Efficacy and safety of tisagenlecleucel (Tisa-Cel) in adult patients (Pts) with relapsed/refractory follicular lymphoma (r/r FL): primary analysis of the phase 2 Elara trial. J Clin Oncol. 2021;39(15_ Suppl):7508. 6. Jacobson CA, Chavez JC, Sehgal AR, et al. Primary analysis of Zuma-5: a phase 2 study of axicabtagene ciloleucel (Axi-Cel) in patients with relapsed/refractory (R/R) indolent non-Hodgkin lymphoma (iNHL). Blood. 2020;136(Suppl_1):40-41. 7. Roulland S, Faroudi M, Mamessier E, Sungalee S, Salles G, Nadel B. Early steps of follicular lymphoma pathogenesis. Adv Immunol. 2011;111:1-46. 8. Pasqualucci L. Molecular pathogenesis of germinal center-derived B cell lymphomas.

Immunol Rev. 2019;288(1):240-261. 9. Kumar E, Pickard L, Okosun J. Pathogenesis of follicular lymphoma: genetics to the microenvironment to clinical translation. Br J Haematol. 2021;194(5):810-821. 10. Vela V, Juskevicius D, Dirnhofer S, Menter T, Tzankov A. Mutational landscape of marginal zone B-cell lymphomas of various origin: organotypic alterations and diagnostic potential for assignment of organ origin. Virchows Arch. 2021;Sep 8. [Epub ahead of print]. 11. Gurion R, Rozovski U, Itchaki G, et al. Humoral serologic response to the BNT162b2 vaccine is abrogated in lymphoma patients within the first 12 months following treatment with anti-CD2O antibodies. Haematologica. 2021;Jul 29. [Epub ahead of print].

haematologica | 2022; 107(1)

REVIEW SERIES

Time for an individualized approach to first-line management of follicular lymphoma

Ferrata Storti Foundation

Guillaume Cartron1 and Judith Trotman2 Department of Hematology, University Hospital, Montpellier, France and 2Department of Hematology, Concord Hospital, Sydney, Australia

ABSTRACT

ollicular lymphoma is a heterogeneous B-cell lymphoma both in presentation and at progression. For most patients it is a chronic, relapsing indolent disease with overall survival expectations now potentially beyond 20 years. However, in a significant minority (~20%) who experience early progression or histological transformation after treatment, the disease no longer has an indolent behavior. This review looks at the development of prognostic indices, staging and therapies for follicular lymphoma, identifying where the data can, and cannot, guide the multidisciplinary team to determine an individualized approach to first-line therapy. A nuanced patient- and disease-specific approach is necessary to maximize disease response and survival while minimizing therapeutic toxicity.

Haematologica 2022 Volume 107(1):7-18

Introduction Follicular lymphoma (FL) is an indolent lymphoid neoplasm derived from germinal center B cells. Rapid therapeutic initiation is rarely required. With the oftenadvanced age of patients at diagnosis and the indolent nature of the disease, some patients will never need treatment. Almost half of patients with asymptomatic disease undergoing an accepted “watch and wait strategy” will not have commenced treatment 31 months after diagnosis.1 For symptomatic patients, the combination of a monoclonal antibody against CD20 (rituximab, obinutuzumab) with chemotherapy results in 80% survival at 10 years.2 Many of these patients will not experience disease relapse or die from their disease, even though FL remains the most common cause of death.3 The challenge for hematologists is to identify patients who need treatment and to define the most appropriate treatment for them considering their age, comorbidities and likelihood of subsequent relapse, to optimize both longevity and quality of life. While most patients are destined to have a prolonged survival, described as a “functional cure”, approximately 20% of patients will progress within the first 2 years of initiating treatment. These early failures, which frequently correspond with histological transformation into diffuse large B-cell lymphoma, constitute the true evolutionary turning point of this disease, accounting for more than 50% of patinets’ deaths in the first decade.3 Faced with difficulty in predicting an individual patient’s outcome before treatment initiation, it is important to implement a dynamic strategy that ensures the early identification of those patients who experience histological transformation in order to offer them innovative therapeutic strategies. The nuanced management of FL is now well charted. With our comprehensive knowledge of the disease, its evolution and patient-related factors, we are now able to discuss different therapeutic approaches with our patients. The recent arrival of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) challenges certain therapeutic choices. Here again, our evolving knowledge and a balanced and honest discussion with the patient should enable informed decision-making adapted to the patient’s life priorities.

Epidemiology and environmental factors

Correspondence: GUILLAUME CARTRON g-cartron@chu-montpellier.fr Received: June 23, 2021. Accepted: October 1, 2021. https://doi.org/10.3324/haematol.2021.278766

FL is the most frequent indolent lymphoma, representing around 20% of all adult lymphomas in Europe, 30% in the USA and 10% in Asia and developing

haematologica | 2022; 107(1)

G. Cartron and J. Trotman

countries The median age at diagnosis is 63 years and the male/female ratio is around 1.2.4 The incidence is 2.7 per 100,000 people, such that there are approximately 13,200 new cases of FL in the USA each year. Most patients (54.8%) are aged 55 - 75 years whereas patients younger than 40 and older than 85 represent a minority (2.4 and 4.8%, respectively). Epidemiological data from outside the USA remain sparse, but a French report charted a similar incidence and median age.5 A family history of non-Hodgkin lymphoma (odds ratio [OR]: 1.99), high body mass index in young adulthood (OR: 1.15) and Sjögren syndrome in young women (OR: 3.37) increase the risk of FL.6 Among environmental factors, smoking (especially heavy smoking in women, OR: 1.41) also correlated with FL.7 Prolonged agriculture exposure to pesticides increases the risk of FL8 (OR: 2.95.0 according to the type of pesticide), a situation also associated with both circulating B cells harboring the initiating t(14;18) and genetic alterations linked to malignant progression.9

Clinical features and diagnostic approach Most patients with FL present with asymptomatic disease and the diagnosis is often suspected incidentally during a clinical or radiological examination performed for symptoms unrelated to FL. When symptomatic, the clinical presentation is dominated by lymphadenopathy or less frequently disease-related complications, such as abdominal pain (mesenteric lymphadenopathy) or respiratory signs (cough, shortness of breath, pleural effusion) in the event of thoracic lymphadenopathy. The general “B” symptoms (recurrent fever >38°C, drenching sweats, weight loss >10%) associated with lymphoma only occur in about 20% of patients and should raise the suspicion of histological transformation. Laboratory investigations look for the existence of cytopenia, particularly anemia. While bone marrow infiltration is the most common cause, the possibility of autoimmune cytopenia should not be underestimated. The presence of circulating lymphoma cells, found in fewer than 10% of cases, is associated with a shorter progression-free survival (PFS).10 The lactate dehydrogenase and b -microglobulin levels are important prognostic markers, and serum electrophoresis should be systematically requested in the search for an associated monoclonal protein and viral serology (hepatitis B and C, human immunodeficiency virus) should be determined before any treatment using anti-CD20 monoclonal antibody. The diagnosis of FL is made on a lymph node or, in the uncommon extranodal FL, organ biopsy. Initial diagnosis by fine needle aspiration or needle biopsy is not recommended due to the importance of architectural examination in diagnosis and the risk of missing histological transformation. Multiple core needle biopsies, guided by ultrasonography or a computerized tomography (CT) scan, using a 14- or 16-gauge needle, may be informative but should be reserved in the context of disease to abdominal or extranodal sites. If it remains debated, when clinical findings suggest a possible hidden histological transformation, a 18F-fluorodeoxyglucose (FDG) positron emission tomography, coupled with low-dose CT (PET) may assist in identifying sites where a repeat biopsy could identify such. There are however no prospective data to 2

support re-biopsy on the basis of baseline quantitative PET metrics alone. Indeed, a large body of data from the GALLIUM study suggests no association of maximum standardized glucose uptake value (SUVmax) with either histological transformation11 or PFS.12 There is marked heterogeneity in FDG uptake within patients which reflects metabolic activity of the malignant B cells as well as that of the microenvironment.13 Likewise, the isolated analysis of bone marrow or blood infiltration is not appropriate for the diagnosis of FL.

Histopathology FL is an entity well-defined in the World Health Organization (WHO) classification.14 The macroscopic appearance of lymph nodes involved by FL is usually vaguely nodular with a predominantly follicular organization microscopically. Thus, tumors frequently maintain a normal germinal center architecture forming enlarged lymphoid follicles with centroblasts and centrocytes randomly distributed with a loss of the polarization usually seen in reactive lymph nodes. On immunochemistry, FL cells typically express a germinal center pattern (CD10, BCL6), B-markers (CD19, CD20, SIg) and BCL2. Although absent in 10 to 15% of cases, BCL2 overexpression, secondary to t(14;18), is the hallmark of FL and can be useful in distinguishing FL from reactive follicles. It is important to note that t(14;18) (q32;q21), placing BCL2 under the transcriptional regulation of IGH regulatory regions leading to the overexpression of this anti-apoptotic protein, is a lymphoma-initiating event. In cases lacking BCL2 expression, the usual BCL6 and CD10 expression helps to distinguish FL from other lymphoproliferative diseases. Present at low level in 70% of healthy individuals, lymphoid cells harboring t(14;18) require additional genetic events for the development of FL. A grading system based on counting the absolute number of centroblasts in ten neoplastic follicles per high power field is used to characterize FL into grades 1 and 2 (grade 1: 0-5 centroblasts per high power field, grade 2: 615) which represent the majority of cases (80%). Grade 3 FL has >15 centroblasts per high power field and is further subdivided into FL 3A (centrocytes still present) and FL 3B (composed exclusively of centroblasts). FL grade 3B is now managed as an aggressive lymphoma similar to diffuse large B-cell lymphoma. Considering the therapeutic impact of distinguishing between FL3A/3B, an expert review of histological classification may be useful. Rare entities including testicular FL, duodenal-type FL, pediatric FL and in situ follicular neoplasia are also characterized in the WHO classification. These entities are not discussed in this review.

Staging At diagnosis, staging is mandatory to determine the distribution and burden of the lymphoma. Staging is an important factor in guiding treatment decisions - influencing both choice and timing of therapy- and re-staging is important to assess disease response after treatment and at times of progression requiring treatment. The current staging procedures include physical examination, blood investigations (full blood count, renal function and liver function haematologica | 2022; 107(1)

Individualized first-line management of FL

Table 1. Revised staging system for primary lymphoma (adapted from Cheson et al.76).

Stage Limited I II I or II bulky* Advanced III IV

Nodal involvement

Extranodal involvement (E)

One or several nodes of one group Two or more nodal groups on the same side of the diaphragm I or II as above with bulky disease

Single extranodal Stage I or II with contiguous and limited extranodal involvement

Nodes on both sides of the diaphragm or nodes above the diaphragm with spleen involvement Additional non-contiguous or extended extranodal involvement

Not applicable

*‘Bulky’ disease has been defined as any nodal mass of 10cm or greater than one third of the transthoracic diameter at any level of thoracic vertebrae. Tonsils, Waldeyer ring and spleen are considered nodal sites.

Table 2. Criteria for initiation of treatment in patients with follicular lymphoma.

GELF criteria27

BNLI criteria28

Presence of at least one of the following criteria: Any B symptom Involvement of ≥3 nodal sites, each with a diameter ≥ 3 cm Tumor mass ≥7 cm Symptomatic splenomegaly Pleural effusion or ascites Organ compression Serum LDH or b M above upper limit of normal

Presence of at least one of the following criteria: Pruritus or B symptom(s) Rapid generalized disease progression in the preceding 3 months Life-endangering organ involvement Significant bone marrow infiltration Localized bone lesions detected on X-ray or isotope scan Renal infiltration ‘Macroscopic’ as opposed to ‘microscopic’ liver involvement

GELF: Groupe d'Etude des Lymphomes Folliculaires; BNLI: British National Lymphoma Investigation; LDH: lactate dehydrogenase; b M: beta-2 microglobulin. 2

tests, lactate dehydrogenase and b -microglobulin), CTscan of the neck, thorax, abdomen and pelvis and PET. PET is now the recommended gold-standard imaging investigation at initial staging. FL has a universally, albeit not uniformly, glucose-avid histology, with FDG uptake in 98% of patients.15 PET is more sensitive than standard contrast-enhanced CT at detecting extranodal disease.16 The PET report should indicate nodal and extranodal 18FFDG uptake due to lymphoma as well as the lesion with the highest standardized glucose uptake value (SUVmax). The extent of the disease is then classified according to the Lugano classification (Table 1). PET scanning resulted in up-staging of as many as 62% of localized (I/II) cases into advanced stage (III/IV) cases.17 The utility of PET has been demonstrated by identifying disseminated disease ultimately associated with a poorer outcome if treated as localized on the basis of CT-based staging.17 The prognostic value of total metabolic tumor volume at diagnosis in FL remains debated12,18 and the predictive value of automated software solutions for measuring total metabolic tumor volume19 needs to be validated in prospective studies before being utilized in a standard manner in clinical practice. The prognostic value of end-of-induction PET response after first-line immunochemotherapy for FL20,21,22 is now clear and provides a platform for PET-adapted therapies in current clinical trials.23,24 While focal bone marrow involvement can be identified on PET, it is generally less sensitive than bone marrow biopsy because it is often diffuse and low volume. Thus, when identification of bone marrow involvement is necessary (cytopenia, clinical trial, confirmation of localized disease), bone marrow biopsy (including both aspirate and trephine) is required to complete marrow assessment when the PET is negative. Identification of marrow involvement does not change treatment in dis2

haematologica | 2022; 107(1)

seminated disease and its prognostic value is debated, with prognostic merit identified in the PRIMA study25 but not in the GALLIUM study.26

Treatment-initiation criteria Tumor burden is usually considered an important prognostic factor in FL. This observation and the commonly indolent nature of FL led the Groupe d'Etude des Lymphomes Folliculaires (GELF)27 and the British National Lymphoma Investigation (BNLI)28 to propose criteria for initiation of treatment (Table 2). These are mainly clinical criteria, empirically defined in the 1990s. They are still used to guide initial management in both clinical practice and trials and have shown consistent prognostic value.29 Notwithstanding the value of such criteria, in the patient with a slowly waxing and waning disease, it may not be appropriate to commence treatment in routine clinical practice just because they have a mildly elevated lactate dehydrogenase or an asymptomatic abdominal mass that has taken years to reach 7 cm in longest dimension. Conversely, some young patients who do not strictly meet GELF/BNLI criteria may warrant initiation of therapy for rapidly progressing disease.

Prognostic indices In 2004, the Follicular Lymphoma International Prognostic Index (FLIPI) was the first prognostic index dedicated to FL30 (Table 3). FLIPI separates three groups of patients (approximately one third each) with significant differences in overall survival (OS) (Table 4). Although originally designed in the pre-rituximab era, FLIPI has 9

G. Cartron and J. Trotman Table 3. Adverse prognostic factors according to FLIPI, FLIPI-2 and PRIMA-PI.

FLIPI30

FLIPI-231

Age > 60 years Ann Arbor stage III/IV LDH > normal Hemoglobin < 120 g/L Nodal sites involved > 4

Age > 60 years Marrow involvement b -microglobulin > normal Hemoglobin < 120 g/L Tumor mass > 6 cm 2

PRIMA-PI20 Marrow involvement b -microglobulin > 3 mg/L 2

FLIPI: Follicular Lymphoma International: Prognostic Index; PRIMA-PI; PRIMA Prognostic Index; LDH: lactate dehydrogenase.

been validated in studies using immunochemotherapy and is now widely used in daily practice and clinical trials. FLIPI-2 was developed in a population of rituximab-treated patients for whom b -microglobulin was available31 and is also predictive of PFS and OS. The PRIMA prognostic index (PRIMA-PI) was developed from a retrospective analysis of a population of patients receiving immunochemotherapy followed by rituximab maintenance25 in a large randomized clinical trial. This prognostic index is based on b -microglobulin and marrow infiltration, two easily available parameters, albeit with the requirement of an invasive bone marrow biopsy. Recently, with the description of recurrent gene mutations in FL, some groups have designed new prognostic indices, including both clinical factors and mutational status.32,33 The use of these indices is uncommon in routine practice due to the limited availability of mutational analysis, and they have only been validated as prognostic in the FL population treated with R-CHOP (rituximab plus cyclophosphamide, doxorubicin, vincristine, prednisone). 2

Natural course of the disease FL is still considered as an indolent but usually incurable disease, although the age- and gender-matched survival in patients who have remained event-free for 2 years after initial treatment challenges our concepts of cure.34 FL is characterized by a heterogeneous presentation and outcome reflecting its biological heterogeneity. Patients with advanced stage indolent disease are manageable with a “wait and watch” strategy for many years whereas others may experience short OS related to transformation into diffuse large B-cell lymphoma. In the former population, representing one third of patients, an indolent presentation does not require therapeutic intervention.35 For such asymptomatic patients, 50% and 20% will not require treatment at 3 and 10 years, respectively, after diagnosis and their estimated OS is probably higher than the median 80% charted at 10 years after rituximab chemotherapy.1,2,28 The pattern of clinical evolution seems to be the same for FL grades 1,2 and 3A whereas, grade 3B FL is now recognized to be genetically closer to diffuse large B-cell lymphoma with a more aggressive course that requires anthracycline-containing immunochemotherapy at diagnosis.36 For patients requiring treatment, the use of combined immunochemotherapy results in a 10-year OS of around 80%.2 However, the lack of a plateau in the PFS curve suggests that we have probably only extended the natural history of this indolent disease with repeated relapses at increasingly shorter intervals. Despite recent therapeutic progress and improved OS, FL remains the leading cause of death in patients with a cumulative incidence of 10% at 10 years.3 Nonetheless, even after first 10

relapse/progression the survival data are promising, with a median OS for patients who received second-line treatment beyond 10 years.37 Patients experiencing disease progression within 24 months after treatment initiation (POD24) represent a particular group of need, with a 5year survival of only 50%.38 However, it should be noted that this datum was derived in an era that predated PETbased staging and in one large retrospective populationbased study the prognostic impact of POD24 is not as powerful in the modern era.39 Histological transformation is probably a turning point for patients’ outcome with a shorter survival if it occurs after FL treatment. De novo histological transformation (i.e., diffuse large B-cell lymphoma histology with pathological findings showing existence of a FL component at the time of diagnosis) has a better prognosis.40 The median survival after histological transformation is around 4 to 5 years and probably shorter if the transformation occurs within the first year.41 The annual incidence of biopsydocumented histological transformation is <3%, with a plateau after the first 2 years following treatment initiation.11,41 One study of more than 8,000 patients identified a lower incidence of histological transformation in the rituximab era. The 10-year cumulative hazard of histological transformation was 5.2% (95% CI: 4.5-6.2) in patients who received rituximab and 8.7% in those who did not (hazard ratio=0.73, 95% CI: 0.58-0.90; P=0.004).42 Histological transformation is probably under-reported given that biopsy, while recommended, is not uniformly performed at relapse/progression.

Initial treatment of follicular lymphoma After distinguishing localized from disseminated disease applying the Lugano 2014 classification, the GELF/BNLI28,29 (Tables 1 and 2) treatment-initiation indices help to identify patients with symptomatic FL requiring treatment. A combination of staging and GELF/BNLI characteristics distinguishes three clinical situations that require different therapeutic approaches: (i) patients with localized FL; (ii) patients with disseminated FL not meeting treatment-initiation criteria (also called “low-tumor burden”); and (iii) patients with disseminated FL who meet treatment-initiation criteria (also called “high-tumor burden”).

Localized follicular lymphoma Fewer than 20% of patients present with stage I or II FL at diagnosis.37 Such patients are identified after a comprehensive assessment, including both PET and bone marrow biopsy to confirm the localized nature of the disease. Stage I disease, with removal of the only pathological node (stage I0) is an infrequent clinical situation that can benefit from therapeutic abstention. While data are sparse, additional treatment (such as radiotherapy) does not appear to haematologica | 2022; 107(1)

Individualized first-line management of FL

Table 4. Survival of patients with follicular lymphoma according to FLIPI, FLIPI-2 and PRIMA-PI.

Risk group FLIPI30 Low Intermediate High FLIPI-231 Low Intermediate High PRIMA-PI20 Low Intermediate High

N. of risk factors

Patient distribution (%)

Survival

0-1 2 3-5

36 37 27

0 1-2 3-5

20 53 27

5-year OS (%) 91 78 52 3-year PFS (%) 91 69 51

b M ≤ 3 mg/L and BM (-) b M ≤ 3 mg/L and BM (+) b M > 3 mg/L

21 36 43

N/A N/A N/A

10-year OS (%) 71 51 35.5 5-year PFS (%) 79.5 51 19 69 51 37

FLIPI: Follicular Lymphoma International :Prognostic Index; PRIMA-PI; PRIMA Prognostic Index; OS: overall survival; PFS: progression-free survival; b M: beta 2-microglobulin; BM: bone marrow. 2

modify the evolution of these patients whose relapse profile is disseminated disease in half of the cases.43 For patients with stage I or II FL with persistent lymphadenopathy after diagnostic biopsy, there is no consensus on appropriate therapy. Alternatives to discuss with the patient include abstention (“wait and watch”), radiotherapy, and rituximab and chemotherapy followed by radiotherapy or immunochemotherapy alone. None of these strategies has demonstrated superiority with regard to OS even if immunochemotherapy,44 chemotherapy followed by radiation therapy45 or radiation followed by chemotherapy46 results in improved PFS. If radiotherapy is chosen, a dose of 24 Grays (Gy) in 12 fractions gives similar results to a dose of 40-45 Gy.47 Low-dose radiotherapy (4 Gy in 2 fractions) may be preferred for some disease locations to minimize toxicity, particularly when radiotherapy is of palliative intent, but is associated with an inferior PFS.48 Thus, for most patients, the physician could recommend either treatment abstention, radiotherapy (24 Gy) or rituximab (4 weekly infusions).44,45 Some patients with localized FL have poor prognostic characteristics according to GELF/BNLI criteria or FLIPI and FLIPI-2 scores. For these patients, immunochemotherapy as discussed later for advanced stage symptomatic FL could be considered. These patients include subjects with bulky disease (>7 cm according to GELF, or 6 cm according to FLIPI-2), particularly abdominal disease, or a lymphomarelated increase of lactate dehydrogenase or b -microglobulin levels. 2

Advanced stage follicular lymphoma without treatment-initiation criteria (low-tumor burden) For advanced stage FL without criteria for treatment initiation (asymptomatic disease), the recommendation is usually to propose therapeutic abstention with dynamic observation. There is no demonstrated OS benefit from initiating early treatment.27,28 The observation that 15 to 20% of patients have still not received treatment 10 years after diagnosis, and that 12% will observe a spontaneous reduction of their disease with an estimated 6% obtaining a complete response, also argues for a “watch and wait” strategy. These patients must, however, be carefully monitored because the median time to starting treatment is around 31 months.1 These clinical observations and the unease that some patients and physicians have with this “watch and wait” strategy led to investigations on the use haematologica | 2022; 107(1)

of rituximab in asymptomatic and disseminated FL. Rituximab monotherapy was examined in two large prospective multicenter trials.1,49 In these trials, rituximab was administered in four weekly infusions followed by maintenance rituximab (infusions every 2 months for 2 years or until progression)1 or with re-treatment according to the same schedule (4 weekly infusions) at progression.49 These two prospective trials demonstrated that after a long follow-up, rituximab monotherapy had no significant side effects. Rituximab increased both PFS and time to the next treatment (TTNT) without having an impact on OS. Ongoing treatment with rituximab maintenance augments the PFS and the TTNT advantage but also the treatment duration (2 years) and increases the rate of grade 1-2 infections. The USA study also demonstrated that there is no benefit from maintenance treatment compared to re-treatment at progression, with four weekly rituximab infusions, leading to a lower utilization of resources for a similar result. If “watch and wait” is an appropriate strategy for a given patient with asymptomatic disease, the physician must ensure that the objectives of this strategy are well understood and accepted by the patient or risk undermining the patient’s confidence. The patient needs to be aware that this approach does not affect OS and is accompanied by a median time to treatment of 2.5 years.1 The physician who proposes this abstention should not underestimate the psychological consequences for the patient and his or her family of the announcement of a malignant disease, not justifying treatment at the moment but being able to justify it in the near future. The delay in initiating treatment is often seen by patients as time lost in the “fight against the disease” and the physician should not underrate the potential negative impact of this strategy. Instead, with psychologist support, the time can be positioned as a period in which to optimize patients’ health and address social issues that will affect their fitness and tolerance of therapy, such as weight management, physical fitness and smoking cessation and arranging leave from employment and/or securing extra support from caregivers. Efforts must be made to help the patient understand the proposed lymphoma strategy and to reciprocally understand and acknowledge the patient’s experience of their disease. Administration of four weekly injections of rituximab can be an alternative to therapeutic abstention for some patients, for whom compliance with a “watch and wait” proposal appears difficult. The impact of such a strategy on subsequent access to promising first11

G. Cartron and J. Trotman

Figure 1. Results of immunochemotherapy with or without rituximab maintenance as first-line treatment in patients with follicular lymphoma. (A-C) KaplanMeier estimates of progression-free survival (A); time to next anti-lymphoma treatment (B) and overall survival (C) from randomization. OS: overall survival; PFS: progression-free survival; TTNLT: time to next anti-lymphoma treatment; HR: hazard ratio; 95% CI: 95% confidence interval. Adapted from Bachy E, et al.2.

haematologica | 2022; 107(1)

Individualized first-line management of FL

line treatment options in clinical trials needs to be understood. If maintenance rituximab is provided, a short schema (M3, M5, M7 and M9) is preferred because these give similar results to those of a prolonged schedule.50 Because such a recommendation results in complete response in 50 to 70%, without significant side effects, it allows “time without disease and chemotherapy” (option preferred by the patient) instead of “time to wait for treatment” (option preferred by the physician). With rituximab treatment, approximately 25% of patients will need retreatment within 5 years1. In patients for whom “wait and watch” has been chosen, close clinical monitoring (every 3 months) should be carried out during the first year of surveillance, before moving to less intensive observation (every 6 months) in the absence of early signs of progression. Patients must be assured that they will be seen promptly in the event of any concerns about progression. Annual CT scanning may be appropriate to offer and reassure the patient with isolated abdominal disease. The frequency of scanning is ultimately often reduced by those concerned about radiation exposure but for many with stable or only very slowly progressive disease annual imaging provides reassurance that they can continue with work or other life plans without expectation of interruption. The use of PET to monitor FL during the “watch and wait” period is discouraged.

Advanced stage follicular lymphoma with treatment initiation criteria (high-tumor burden) Immunochemotherapy improves overall survival For disseminated FL that meets criteria for treatment initiation, the benefit of combining rituximab with chemotherapy has been clearly demonstrated in four prospective randomized trials51-54 (Table 5.). A Cochrane analysis55 including these studies showed a 37% reduction in the risk of death (HR=0.63, 95% CI: 0.51 - 0.79) when rituximab is used in combination with chemotherapy as first-line treatment. Because of the significant decrease of toxicity, rituximab alone has been proposed. This approach cannot, however, be considered as standard treatment for symptomatic patients with FL considering the reduced PFS and TTNT compared to those achieved with immunochemotherapy.56 Rituximab maintenance improves progression-free survival but not overall survival The PFS benefit of maintenance treatment using rituximab was demonstrated in the PRIMA trial.2 After induction using immunochemotherapy, with R-CHOP, R-CVP (rituximab plus cyclophosphamide. vincristine, prednisone) or R-FCM (rituximab plus fludarabine, cyclophos-

phamide, mitoxantrone), responding patients were randomized to observation or maintenance treatment with rituximab every 2 months for 2 years. With 10 years of follow-up, this trial showed a significant increase in PFS for patients receiving maintenance treatment with a median PFS of 10.5 years versus 4.1 years for those not given maintenance. The median time to the next treatment was not reached in patients receiving maintenance treatment compared to 6.6 years in the control arm. However, this trial did not show an OS benefit for maintenance, with 10-year OS being 80% in both arms, raising questions about the merits of maintenance, especially in elderly patients or those with respiratory problems in whom toxicity may be less acceptable. The PFS benefit of maintenance treatment with rituximab after immunochemotherapy with bendamustine-rituximab was suggested in an ad hoc analysis in the BRIGHT study.57 The best chemotherapy regimen is likely patient-specific A number of chemotherapy regimens have been enhanced by their combination with an anti-CD20 antibody (Table 5). One study (Foll05) prospectively compared three different chemotherapy regimens,58 R-CVP, R-CHOP and R-FC (rituximab plus fludarabine and cyclophosphamide), showing that while both R-CHOP and R-FC produced superior PFS and TTNT, in the long-term, the immunosuppressive toxicity of fludarabine precluded it as a first-line therapy. In two other studies R-CHOP was compared with rituximab-bendamustine.59,60 Although both studies suffered from methodological issues (nonexclusively FL population, abnormally low results of the RCHOP arm, R-CHOP/R-CVP aggregation) they showed a superior PFS of the rituximab-bendamustine combination. The recent GALLIUM study,61 albeit with a non-randomized choice of chemotherapy, was arguably the only study to systematically collect comprehensive toxicity data after bendamustine use. As a counterpoint to an emphasis on PFS as the key consideration when choosing bendamustine as the chemotherapy backbone, the rate of fatal adverse events in GALLIUM was 5.3% (36/676) in patients assigned to bendamustine as part of induction compared to 1.8% (9/513) in those who received CHOP or CVP after a median follow-up of 41 months. The difference was particularly notable in older patients. The choice of chemotherapy backbone for a given patient is strongly influenced by the patient’s age, co-morbidities and treatment preferences (Table 6). Notwithstanding the debate over the potential trade-offs between toxicity and efficacy of CHOP and bendamustine, in patients with a contraindication to anthracyclines, the preference may be for bendamustine in younger patients while the CVP combination may be preferred in elderly patients because of its lower

Table 5. Randomized trials demonstrating the benefit of chemo-immunotherapy on overall survival.

Trial M390251 GLSG54 M3902353 FL200052

Treatment

N. of patients

Follow-up (months)

Control

8 x CVP vs 8 x R-CVP 6-8 x CHOP vs 6-8 x R-CHOP 8 x MCP vs 8 x R-MCP 6 x CHVP-I vs 6 x R-CHVP-I

318 428 201 360

53 mo 60 mo 47 mo 100 mo

77 84 74 70

OS (%)

Rituximab

83 90 87 78

0.029 0.016 0.0096 0.076

OS: overall aurvival; CVP: 750 mg/m2 cyclophosphamide 750 mg/m2 intravenously (iv) day 1; vincristine 1.4 mg/m2 (maximum 2 mg) iv day 1; prednisone 40 mg/m2 orally (po) days 1-5. CHOP: cyclophosphamide 750 mg/m2 iv day 1; doxorubicin 50 mg/m2 iv day 1; vincristine 1.4 mg/m2 (max, 2.0 mg) iv day 1; prednisone 100 mg/m2 po days 1-5. MCP : mitoxantrone 8 mg/m2 iv day 1; chlorambucil 3x3 mg/m2 po days 1-5; prednisolone 25 mg/m2 po days 1-5. CHVP : cyclophosphamide 600 mg/m2 iv day 1; doxorubicin 25 mg/m2 iv day 1; etoposide 100 mg/m2 iv day 1; prednisolone 40 mg/m2 po days 1-5. I: interferon-a: 3 x/week for 18 months, 4.5 MUI < 70 years ; 3 MUI > 70 years. R: rituximab: 375 mg/m2 iv

haematologica | 2022; 107(1)

G. Cartron and J. Trotman

Table 6. Summary of chemotherapy regimen: advantages and disadvantages. Bendamustine

CHOP

CVP

Advantages

Disadvantages

No alopecia Less febrile neutropenia No cardiomyopathy or peripheral neuropathy PFS TTNT Lower rate of POD24 Effective in HT/ 3B disease Lower rates of stem cell toxicity and persistent immunodeficiency Lower rates of subsequent HT Effectiveness in 3A disease Lower infectious toxicity Less fatigue Minimal alopecia

Long-term infectious risk and mortality in patients > 70 years Early and prolonged CD4+ depletion ➝ Increased risk of severe COVID-19 ➝ Increased risk of lymphocyte harvest failure for CAR T-cell manufacturing Increased risk of cardiac toxicity Peripheral neuropathy Steroid side effects Alopecia Lower PFS/TTNT Steroid side effects Peripheral neuropathy

COVID-19; coronavirus-2019 disease; CAR: chimeric antigen receptor; PFS: progression-free survival; TTNT: time to next treatment; POD24: progression of disease by 24 months; HT: histological transformation; CHOP: cyclophosphamide, doxorubicin, vincristine, prednisone; CVP: cyclophosphamide, vincristine, prednisone.

hematologic toxicity. The RELEVANCE study aimed to demonstrate superiority of an immunotherapy combining rituximab and lenalidomide over conventional immunochemotherapy.62 The lack of any advantage on both response rates and PFS has dampened hopes of a chemotherapy-free standard-of-care first-line treatment for FL. Obinutuzumab offers improved progression-free survival but not overall survival Obinutuzumab is an anti-CD20 monoclonal antibody modified to improve direct cytotoxicity and antibodydependent cell cytotoxicity. This antibody has been compared to rituximab in a phase III trial (GALLIUM)61 combined in induction with different chemotherapy regimens (CHOP, bendamustine, CVP). This immunochemotherapy was followed in responding patients by maintenance treatment with rituximab or obinutuzumab every 2 months for 2 years. It is noted that the dose and administration schedule of the two anti-CD20 antibodies were different, the pharmacokinetic studies carried out during the development of obinutuzumab making it possible to optimize its administration. This study showed that obinutuzumab provides a significant improvement in PFS and TTNT for patients. After a follow-up of 57 months, this translated into risk reductions of 27% (HR=0.73, 95% CI: 0.59- 0.90) for progression and 30% (HR=0.70, 95% CI: 0.54, 0.90) for TTNT.63 The advantage of obinutuzumab on PFS and TTNT was especially marked in patients receiving bendamustine.64 Nonetheless, mindful that TTNT is arguably a more important endpoint in routine clinical practice than in clinical trials, there was no significant TTNT difference at 3 years across all obinutuzumab-chemotherapy arms: obinutuzumab-bendamustine 87%, (83-91%), obinutuzumab-CHOP 87% (82-91%), obinutuzumab-CVP 87% (75 - 93%). When using obinutuzumab patients need to be educated about the high probability (59%) of an infusionrelated reaction. Reassurance about prompt management of such reactions is important to allay patients’ fears and improve their experience with this monoclonal antibody. The preferred first-line immunochemotherapy approach is patient-specific. Choosing between combination immunochemotherapy options requires an individualized, nuanced approach 14

for each patient, with risk:benefit calculations made both before induction and during treatment. While there is no direct randomized comparison between induction with bendamustine and CHOP for young fit patients in the modern era, accumulated data suggest that where PFS is the overriding priority this may be best achieved with the use of obinutuzumab-bendamustine followed by continued therapy with obinutuzumab maintenance. Where the anticipated bendamustine toxicity (including that which may impair T-cell function, affecting both immune fitness and potential chimeric antigen receptor T-cell collection) creates a preference for CHOP (particularly where concerns exist regarding occult transformation) or even CVP, their combination with obinutuzumab results in excellent PFS and TTNT with acceptable toxicity. Nonetheless, the GALLIUM study data are early data, and 5 years of follow-up are inadequate for a disease with a median OS beyond 15-20 years. Clinicians are advised to keep this in mind when scheduling first-line therapies for the management of high-tumor burden FL. Mindful of the increased risk of infections related to the use of maintenance with anti-CD20 therapy, ongoing use of such maintenance needs careful reconsideration in the event of any infection and meticulous adherence to antimicrobial prophylaxis is encouraged. Because of increased toxicity related to obinutuzumab (first dose infusion-related reactions, cytopenia and infections), the lack of a demonstrated OS improvement, and cost considerations, rituximab may remain the preferred anti-CD20 antibody, particularly in very frail elderly patients for whom PFS and TTNT are not the overriding clinical objectives. In this setting, the combination of rituximab with CVP offers the lowest toxicity, albeit with a markedly inferior PFS compared to other immunochemotherapy options. Conversely, where chemotherapy toxicity is of concern but PFS remains a priority, pairing less toxic CVP with obinutuzumab with careful management of infusion-related reactions may be a preferred approach for these elderly patients. Surveillance during first remission in follicular lymphoma Current guidelines from the European Society for Medical Oncology advise clinical and imaging surveillance every 6 months for 2 years after induction and, optionally, annually up to 5 years.65 Surveillance imaging recommendations were mostly based on data from studhaematologica | 2022; 107(1)

Individualized first-line management of FL

ies of aggressive lymphoma whereas FL has a different disease biology and pattern of relapse. For post-induction PET-negative patients in first remission destined to have a long survival and for whom most relapses will occur after more than 3 years, the cumulative radiation exposure and the increase of life-time incidence of cancer attributable to radiation exposure66 do not argue for prolonged imaging surveillance, aside from considerations of increased expense and heightened anxiety. Thus, the limited value of surveillance imaging in detecting asymptomatic nonHodgkin lymphoma relapse earlier, relative to symptomdriven investigations is now generally acknowledged even if not investigated in a randomized controlled manner. A recent retrospective study analyzing the effect of surveillance imaging on relapse detection and OS concluded that the majority of relapses after first remission are initially suggested by clinical signs and symptoms and relapse detection through routine surveillance imaging in asymptomatic patients carries no additional survival advantage over imaging in the case of clinical concerns.67 Although PET-CT is more sensitive than CT-scan for asymptomatic surveillance imaging in FL, it has a lower specificity compared to CT-scan, leading to invasive biopsies and heightening anxiety and cost burden. There is now evidence for recommendations limiting imaging surveillance to examinations that significantly change treatment and/or demonstrate prognostic value. Thus PET-CT at the end of induction and CT scanning every 6 months during the 2 years following induction therapy enable the detection of POD24 patients at risk of early FLrelated death. After that, imaging investigations should be limited to confirmation of relapse in the presence of concerning clinical signs and symptoms. Molecular monitoring for minimal residual disease is not recommended outside of clinical trials.

Ongoing unmet needs in follicular lymphoma The estimated prolonged survival of most patients diagnosed with FL in 2021 represents one of the greatest oncology advances in the new millennium. However, this advance is based primarily on the introduction of combined immunochemotherapy induction and maintenance and/or re-treatment with a failure to demonstrate superiority of a lenalidomide-rituximab “chemo-free” approach. While collaborative efforts have studied imaging and molecular markers of response, we still struggle to clearly identify at diagnosis the individual patient destined to have an inferior response for whom therapies used in relapse warrant testing in the first-line setting. With perhaps the future exception of EZH2 inhibitors, a greater understanding of the mutational landscape has not eventuated in the hoped for precision medicine in FL. The biological plausibility of efficacy of BCL-2 and BTK inhibitors has not translated into meaningful clinical benefit and the phophoinositidine-3 kinase inhibitors have remained firmly as therapies for later relapse. Rather, we are currently focused on the potential to harness the power of chimeric antigen receptor T cells and bi-specific antibody therapies in patients with multiply relapsed/refractory disease with hope that success with these cellular therapies will warrant their study in the management of early relapses. Our current focus on the haematologica | 2022; 107(1)

unmet need for the early progressing (POD24) population is appropriate. Patients who have remained event free for 24 months after immunochemotherapy have an age- and gender-matched survival comparable to that of the general population.34 However, for such patients destined to have a “functional cure”, a clear unmet is the development of appropriate trial designs that move beyond PFS as the sole primary endpoint. Perhaps it is time to develop, in parallel, a composite health utility measure that incorporates both efficacy and toxicity of therapy for such patients with FL.

Impact of the SARS-CoV-2 pandemic on the care of patients with follicular lymphoma Coronavirus disease 2019 (COVID-19) has become a global health problem. Despite the arrival of vaccines, SARS-CoV-2 infection disproportionately affects our patients. Those with hematologic malignancies are at higher risk of death from COVID-19.68 Indolent lymphoma was an independent predictor of mortality (HR=2.19, 95%CI: 1.07-4.48) in a retrospective Italian study.69 Initial data support expectations that the rate of SARS-CoV-2 IgG seroconversion after vaccination is lower in patients receiving an anti-CD20 monoclonal antibody.70 The profound impact of rituximab was recently highlighted in patients with chronic lymphocytic leukemia in whom none of 22 patients exposed to rituximab in the 12 months preceding vaccination responded to the mRNA vaccines.71 Circulating rituximab (and obinutuzumab) is found 3 to 6 months after four weekly infusions of the monoclonal antibody and rituximab induces B-cell depletion72 which recovers 9 to 12 months after the last infusion.73 Treatment with bendamustine, which lowers the CD4+ T-lymphocyte count for several months starting as early as after one cycle,64 has also recently been suggested as a determinant of severe COVID-1974 and likely also jeopardizes post-vaccination immune responses. General recommendations for FL patients to continue wearing a face mask and maintain social distancing as well as vaccination of all close contacts remain important. For patients with FL not requiring immediate therapy, rituximab could be withheld at least until there is some confidence that the patients exhibit SARS-Cov-2 post-vaccination immunization, and that their contacts and the community as a whole have likewise received vaccination. For symptomatic patients, vaccination could be attempted before treatment initiation for those able to wait at least 2 to 3 months. For patients requiring treatment, the greater immunosuppression, particularly T-cell depletion, associated with bendamustine suggests that this drug should be avoided. The question of maintenance antibody treatment is challenging since induction immunochemotherapy alone likely results in an inability of patients to become immunized for at least 1 year after the last infusion of the antibody. Maintenance therapy prolongs this problem by at least 2 years. In the COVID era the long-term PRIMA follow-up data warrant a closer look.2 While the median PFS of patients not receiving rituximab maintenance was only 4.1 years there was a median 9.3 years for time to next chemotherapy in this arm. In future years, booster vaccines, herd immunity and monoclonal antibody therapies for acute COVID-19 may make SAR-CoV-2 infection less of a threat, but vac15

G. Cartron and J. Trotman

cine hesitancy and viral evolution globally continue to challenge confidence in achieving a safe community for our patients.

Conclusions FL is an indolent B-cell lymphoma and affected patients now have a very prolonged median OS approaching 20 years. A hallmark of FL is its heterogeneity both at presentation and in the event of relapse. A significant proportion of patients managed with a “watch and wait” approach do not require any therapy, and the use of rituximab monotherapy for treating these patients is likely lower in the current COVID pandemic. With more accurate staging and expansion of therapeutic options in the modern era the treatment of each patient needs to be individualized. Nuanced decisions made in partnership with the patient and their families must account for competing priorities between efficacy and safety as we “play the long game” in lymphoma management. This is particularly important as data make it clear that neither PFS nor POD2475 are surrogates for OS. The small but cumulative risk of histological transformation mandates biopsy at relapse, especially in patients with early symptomatic progression for whom

References 1. Ardeshna KM, Qian W, Smith P, et al. Rituximab versus a watch-and-wait approach in patients with advanced-stage, asymptomatic, non-bulky follicular lymphoma: an open-label randomised phase 3 trial. Lancet Oncol. 2014;15(4):424-435. 2. Bachy E, Seymour JF, Feugier P, et al. Sustained progression-free survival benefit of rituximab maintenance in patients with follicular lymphoma: long-term results of the PRIMA study. J Clin Oncol. 2019;37(31): 2815-2824. 3. Sarkozy C, Maurer MJ, Link BK, et al. Cause of death in follicular lymphoma in the first decade of the rituximab era: a pooled analysis of French and US cohorts. J Clin Oncol. 2019 ; 37(2):144-152. 4. National Cancer Institute Surveillance, Epidemiology, and End Results Program. Cancer stat facts: NHL-follicular lymphoma. Accessed June 11, 2020. https://seer. cancer.gov/statfacts/html/follicular.html 5. Defossez G, Le Guyader-Peyrou S, Uhry Z, et al. Estimations nationales de l'incidence et de la mortalité par cancer en France métropolitaine entre 1990 et 2018 - Volume 1 : Tumeurs solides : Étude à partir des registres des cancers du réseau Francim. https://www.santepubliquefrance.fr/maladies-et-traumatismes/cancers/cancer-dusein/documents/rapport-synthese/estimations-nationales-de-l-incidence-et-de-lamortalite-par-cancer-en-france-metropolitaine-entre-1990-et-2018-volume-1tumeurs-solides-etud 6. Linet MS, Vajdic CM, Morton LM, et al. Medical history, lifestyle, family history, and occupational risk factors for follicular lymphoma: the InterLymph Non-Hodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monographs. 2014;2014(48):26-40. 7. Schollkopf C, Smedby KE, Hjalgrim H, et al. Cigarette smoking and risk of non-

aggressive approaches may mitigate the poorer prognosis of early histological transformation. Conversely, for other patients a prolonged first remission after either radiotherapy for low-volume, localized disease or chemoimmunotherapy for high-tumor burden may be sustained for several years. In the elderly or those with substantial co-morbidities, FL may not relapse in their lifetime. The lack of an obvious preferred option for first-line chemoimmunotherapy for FL is exacerbated by the challenges and uncertainties of the COVID era. This pandemic brings into sharp focus the importance of attention to not just PFS but also to treatment toxicities and quality of life when choosing initial chemo-immunotherapy and making an individualized risk:benefit analysis of continuing therapy with antibody maintenance. Disclosures GC reports advisory fees from Roche and Celgene as well as speaker’s fees and honoraria from Takeda, Roche, Janssen, Celgene, Abbvie, Sanofi, and Gilead. JT reports research funding to her institution from Beigene, Celgene-BMS, Roche, Pharmacyclics, Janssen and Takeda. Contributions The two authors jointly decided the plan and write the paper.

Hodgkin’s lymphoma-a population-based case-control study. Cancer Epidemiol Biomarkers Prev. 2005;14(7):1791-1796. 8. Chiu BC, Dave BJ, Blair A, et al. Agricultural pesticide use and risk of t(14;18)-defined subtypes of non-Hodgkin lymphoma. Blood. 2006;108(4):1363-1369. 9. AgopianJ, NavarroJM, GacAC, et al. Agricultural pesticide exposure and the molecular connection to lymphomagenesis. J Exp Med. 2009;206(7):1473-1483. 10. Sarkozy C, Baseggio L, Feugier P, et al. Peripheral blood involvement in patients with follicular lymphoma: a rare disease manifestation associated with poor prognosis. Br J Haematol. 2014 ;164(5):659-667. 11. Mir F, Barrington SF, Brown H, et al. Baseline SUVmax did not predict histological transformation in follicular lymphoma in the phase 3 GALLIUM study. Blood. 2020;135 (15):1214-1218. 12. Barrington S, Trotman J, Sahin D, et al. Baseline PET-derived metabolic tumor volume metrics did not predict outcomes in follicular lymphoma patients treated with firstline immunochemotherapy and antibody maintenance in the phase III GALLIUM study. Blood. 2018;132(Suppl 1):2882. 13. Nath K, Law SC, Sabdia B, et al. Intratumoral T cells have a differential impact on FDG-PET parameters in follicular lymphoma. Blood Adv. 2021;5(12):26442649. 14. Swerdlow SH, Campo E, Pileri SA, et al. The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood. 2016;127(20):2375-2390. 15. Trotman J, Fournier M, Lamy T, et al. Positron emission tomography-computed tomography (PET-CT) after induction therapy is highly predictive of patient outcome in follicular lymphoma: analysis of PET-CT in a subset of PRIMA trial participants. J Clin Oncol. 2011;29(23):3194-3200. 16. Tychyj-Pinel C, Ricard F, Fulham M, et al.

PET/CT assessment in follicular lymphoma using standardized criteria: central review in the PRIMA study. Eur J Nucl Med Mol Imaging. 2014;41(3):408-415. 17. Luminari S, Biasoli I, Arcaini L, et al. The use of FDG-PET in the initial staging of 142 patients with follicular lymphoma: a retrospective study from the FOLL05 randomized trial of the Fondazione Italiana Linfomi. Ann Oncol. 2013;24(8):2108-2112. 18. Meignan M, Cottereau AS, Versari A, et al. Baseline metabolic tumor volume predicts outcome in high-tumor-burden follicular lymphoma: a pooled analysis of three multicenter studies. J Clin Oncol. 2016;34(30): 3618-3626,. 19. Jemaa S, Fredrickson J, Coimbra A, et al. A fully automated measurement of total metabolic tumor burden in diffuse large B-cell lymphoma and follicular lymphoma. Blood. 2019;134(Suppl_1):4666 20. Bachy E, Maurer MJ, Habermann TM, et al. A simplified scoring system in de novo follicular lymphoma treated initially with immunochemotherapy. Blood. 2018;132 (1):49-58. 21. Rutherford SC, Herold M, Hiddemann W, et al. Impact of bone marrow biopsy on response assessment in immunochemotherapy-treated lymphoma patients in GALLIUM and GOYA. Blood Adv. 2021;4(8):1589-1593. 22. Dupuis J, Berriolo-Riedinger A, Julian A, et al. Impact of [(18)F]fluorodeoxyglucose positron emission tomography response evaluation in patients with high-tumor burden follicular lymphoma treated with immunochemotherapy: a prospective study from the Groupe d'Etudes des Lymphomes de l'Adulte and GOELAMS. J Clin Oncol. 2012;30(35):4317-4322. 23. Trotman J, Luminari S, Boussetta S, et al. Prognostic value of PET-CT after first-line therapy in patients with follicular lymphoma: a pooled analysis of central scan review in three multicentre studies. Lancet

haematologica | 2022; 107(1)

Individualized first-line management of FL

Haematol. 2014 ;1(1):e17-27. 24. Trotman J, Barrington SF, Belada D, et al. Prognostic value of end-of-induction PET response after first-line immunochemotherapy for follicular lymphoma (GALLIUM): secondary analysis of a randomised, phase 3 trial. Lancet Oncol. 2018;19(11):1530-1542. 25. Federico M, Mannina D, Versari A, et al. Response oriented maintenance therapy in advanced follicular lymphoma. Results of the interim analysis of the FOLL 12 trial conducted by the Fondazione Italiana Linfomi. Hematol Oncol. 2019 ;37(S2):153-154. 26. Pettitt AR, Barrington S, Kalakonda N, et al. NCRI PETReA trial: a phase 3 evaluation of PET-guided, response adapted therapy in patients with previously untreated, advanced stage, high-tumour-burden follicular lymphoma. Hematol Oncol. 2019;37 (S2):67-68. 27. Brice P, Bastion Y, Lepage E, et al. Comparison in low-tumor-burden follicular lymphomas between an initial no-treatment policy, prednimustine, or interferon alfa: a randomized study from the Groupe d’Etude des Lymphomes Folliculaires. Groupe d’Etude des Lymphomes de l’Adulte. J Clin Oncol. 1997;15(3):1110-1117. 28. Ardeshna KM, Smith P, Norton A, et al. Long-term effect of a watch and wait policy versus immediate systemic treatment for asymptomatic advanced-stage non-Hodgkin lymphoma: a randomised controlled trial. Lancet. 2003;362(9383):516-522. 29. Bachy E, Brice P, Delarue R, et al. Long-term follow-up of patients with newly diagnosed follicular lymphoma in the prerituximab era: effect of response quality on survival-a study from the Groupe d’Etude des Lymphomes de l’Adulte. J Clin Oncol. 2009 ;28(5):822-829. 30. Solal-Céligny P, Roy P, Colombat P, et al. Follicular lymphoma international prognostic index. Blood. 2004;104(5):1258-1265. 31. Federico M, Bellei M, Marsheselli L, et al. Follicular Lymphoma International Prognostic Index 2: a new prognostic index for follicular lymphoma developed by the international follicular lymphoma prognostic factor project. J Clin Oncol. 2009;27(27):4555-4562. 32. Pastore A, Jurinovic V, Kridel R, et al. Integration of gene mutations in risk prognostication for patients receiving first-line immunochemotherapy for follicular lymphoma: a retrospective analysis of a prospective clinical trial and validation in a population-based registry. Lancet Oncol. 2015;16(9):1111-1122. 33. Huet S, Tesson B, Jais JP, et al. A geneexpression profiling score for prediction of outcome in patients with follicular lymphoma: a retrospective training and validation analysis in three international cohorts. Lancet Oncol. 2018;19(4):549-561. 34. Maurer MJ, Bachy E, Ghesquières H, et al. 2016 Early event status informs subsequent outcome in newly diagnosed follicular lymphoma. Am J Hematol. 2016;91(11):10961101. 35. Friedberg JW, Taylor MD, Cerhan JR, et al. Follicular lymphoma in the United States: first report of the National LymphoCare Study. J Clin Oncol. 2009;27(8):1202-1208. 36. Barraclough A, Bishton M, Cheah C, et al. The diagnostic and therapeutic challenges of Grade 3B follicular lymphoma. Br J Haematol. 2021;195(1):15-24. 37. Batlevi C, Sha F, Alperovich A, et al. Follicular lymphoma in the modern era: survival, treatment outcome, and identification

haematologica | 2022; 107(1)

of high-risk subgroups. Blood Cancer J. 2020;10(7):74. 38. Casulo C, Byrtek M, Dawson KL, et al. Early relapse of follicular lymphoma after rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone defines patients at high risk for death: an analysis from the National LymphoCare Study. J Clin Oncol. 2015;33(23):2516-2522. 39. Batlevi C, Sha F, Alperovich A, et al. Positron-emission tomography-based staging reduces the prognostic impact of early disease progression in patients with follicular lymphoma. Eur J Cancer. 2020;126:78-90. 40. Wagner-Johnston ND, Link BK, Byrtek M, et al. Outcomes of transformed follicular lymphoma in the modern era: a report from the National LymphoCare Study (NLCS). Blood. 2015;126(7):851-857. 41. Sarkozy C, Trneny M, Xerri L, et al. Risk factors and outcomes for patients with follicular lymphoma who had histologic transformation after response to first-line immunochemotherapy in the PRIMA trial. J Clin Oncol. 2016;34(22):2575-2582. 42. Federico M, Callabero Barrigon MD, Marsheselli L, et al. Rituximab and the risk of transformation of follicular lymphoma: a retrospective pooled analysis. Lancet Haematol. 2018;5(8):359-367. 43. Soubeyran P, Eghbali H, Trojani M, et al. Is there any place for a wait-and-see policy in stage I0 follicular lymphoma? A study of 43 consecutive patients in a single center. Ann Oncol. 1996;7(7):713. 44. Michallet AS, Lebras L, Bauwens D, et al. Early stage follicular lymphoma: what is the clinical impact of the first-line treatment strategy? J Hematol Oncol. 2013;6:45. 45. Friedberg JW, Byrtek M, Link BK, et al. Effectiveness of first-line management strategies for stage I follicular lymphoma: analysis of the National LymphoCare Study. J Clin Oncol. 2012;30(27):3368-3375. 46. MacManus M, Fisher R, Roos D, et al. Randomized trial of systemic therapy after involved-field radiotherapy in patients with early-stage follicular lymphoma: TROG 99.03. J Clin Oncol. 2018;36(29):2918-2925. 47. Lowry L, Smith P, Qian W, et al. Reduced dose radiotherapy for local control in nonHodgkin lymphoma: a randomized phase III trial. Radiother Oncol. 2011;100(1):86-92. 48. Hoskin P, Popova B, Schofield O, et al. 4 Gy versus 24 Gy radiotherapy for follicular and marginal zone lymphoma (FoRT): long-term follow-up of a multicentre, randomised, phase 3, non-inferiority trial. Lancet Oncol. 2021;22(3):332-340. 49. Kahl B, Hong F, Williams ME, et al. Rituximab extended schedule or re-treatment trial for low–tumor burden follicular lymphoma: Eastern Cooperative Oncology Group protocol E4402. J Clin Oncol. 2014;32(28):3096-3102. 50. Ghielmini M, Schmitz SF, Cogliatti SB, et al. Prolonged treatment with rituximab in patients with follicular lymphoma significantly increases event-free survival and response duration compared with the standard weekly x 4 schedule. Blood. 2004;103 (12):4416-4423. 51. Marcus R, Imrie K, Solal-Celigny P, et al. Phase III study of R-CVP compared with cyclophosphamide, vincristine, and prednisone alone in patients with previously untreated advanced follicular lymphoma. J Clin Oncol. 2008;26(28):4579-4586. 52. Bachy E, Houot R, Morschhauser F, et al. Long-term follow-up of the FL2000 study comparing CHVP-interferon plus rituximab

in follicular lymphoma. Haematologica. 2013;98(7):1107-1114. 53. Herold M, Haas A, Srock S, et al. Rituximab added to first-line mitoxantrone, chlorambucil, and prednisolone chemotherapy followed by interferon maintenance prolongs survival in patients with advanced follicular lymphoma: an East German Study Group Hematology and Oncology Study. J Clin Oncol. 2007;25(15):1986-1992. 54. Hiddemann W, Kneba M, Dreyling M, et al. Frontline therapy with rituximab added to the combination of cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) significantly improves the outcome for patients with advanced-stage follicular lymphoma compared with therapy with CHOP alone: results of a prospective randomized study of the German Low-Grade Lymphoma Study Group. Blood. 2005;106 (12):3725-3732. 55. Shulz H, Bohlius J, Skoetz N, et al. Chemotherapy plus rituximab versus chemotherapy alone for B-cell nonHodgkin's lymphoma. Cochrane Database Syst Rev. 2007;(4):CD003805. 56. Lockmer S, Østenstad B, Hagberg H, et al. Chemotherapy-free initial treatment of advanced indolent lymphoma has durable effect with low toxicity: results from two Nordic Lymphoma Group trials with more than 10 years of follow-up. J Clin Oncol. 2018;38(33):3315-3323. 57. Kahl B, Burke J, van der Jagt R, et al. Assessment of maintenance rituximab after first-line bendamustine-rituximab in patients with follicular lymphoma: an analysis from the BRIGHT trial. Blood. 2017;130(Suppl 1):484. 58. Federico M, Luminari S, Dondi A, et al. RCVP versus R-CHOP versus R-FM for the initial treatment of patients with advancedstage follicular lymphoma: results of the FOLL05 trial conducted by the Fondazione Italiana Linfomi. J Clin Oncol. 2013;31(12): 1506-1513. 59. Rummel MJ, Niederle N, Maschmeyer G, et al. Bendamustine plus rituximab versus CHOP plus rituximab as first-line treatment for patients with indolent and mantle-cell lymphomas: an open-label, multicentre, randomised, phase 3. Lancet. 2013;381(9873): 1203-1210. 60. Flinn IW, van der Jagt R, Kahl B, et al. Firstline treatment of patients with indolent nonHodgkin lymphoma or mantle-cell lymphoma with bendamustine plus rituximab versus R-CHOP or R-CVP: results of the BRIGHT 5-year follow-up study. J Clin Oncol. 2019;37(12):984-991. 61. Marcus R, Davies A, Ando K, et al. Obinutuzumab for the first-line treatment of follicular lymphoma. N Engl J Med. 2017; 377(14):1331-1344. 62. Morschhauser F, Fowler N, Bouabdallah R, et al. Rituximab plus lenalidomide in advanced untreated follicular lymphoma. N Engl J Med. 2018;379(10):934-947. 63. Townsend W, Buske C, Cartron G, et al. Obinutuzumab-based immunochemotherapy prolongs progression-free survival and time to next anti-lymphoma treatment in patients with previously untreated follicular lymphoma: four-year results from the phase III GALLIUM study. Blood. 2018;132(Suppl 1):1597. 64. Hiddemann W, Barbui AM, Canales MA, et al. Immunochemotherapy with obinutuzumab or rituximab for previously untreated follicular lymphoma in the GALLIUM study: influence of chemotherapy on

G. Cartron and J. Trotman efficacy and safety. J Clin Oncol. 2018;36 (23):2395-2404. 65. Dreyling M, Ghielmini M, Rule S, et al. Newly diagnosed and relapsed follicular lymphoma: ESMO clinical practice guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2020;32(3):298-308. 66. Shenoy P, Sinha R, Tumeh JW, et al. Surveillance computed tomography scans for patients with lymphoma: is the risk worth the benefits? Clin Lymphoma Myeloma Leuk. 2010;10(4):270-277. 67. Goldman MX, Mao JJ, Strouse CS, et al. Surveillance imaging during first remission in follicular lymphoma does not impact overall survival. Cancer. 2021;127(18):33903402. 68. Bhaskaran K, Bacon S, Evans SJ, et al. Factors associated with deaths due to COVID-19 versus other causes: population-based cohort analysis of UK primary care data and

linked national death registrations within the OpenSAFELY platform. Lancet Reg Health Eur. 2021;6:100109. 69. Passamonti F, Cattaneo C, Arcaini L, et al. Clinical characteristics and risk factors associated with COVID-19 severity in patients with haematological malignancies in Italy: a retrospective, multicentre, cohort study. Lancet Haematol. 2020;7(10):e737-747. 70. Thakkar A , Pradhan K, Shawn Jindal S, et al. Patterns of seroconversion for SARS-CoV-2 IgG in patients with malignant disease and association with anticancer therapy. Nature Cancer. 2021;2:392-399. 71. Herishanu Y, Avivi I, Aharon A, et al. Efficacy of the BNT162b2 mRNA COVID-19 vaccine in patients with chronic lymphocytic leukemia. Blood.2021;137(23):3165-3173. 72. Cartron G, Blasco H, Paintaud G, et al. Pharmacokinetics of rituximab and its clinical use: thought for the best use? Crit Rev

Oncol Hematol. 2007;62(1):43-52. 73. Dunleavy K, Hakim F, Kim Hk, et al. B-cell recovery following rituximab-based therapy is associated with perturbations in stromal derived factor-1 and granulocyte homeostasis. Blood. 2005;106(3):795-802. 74. Lamure S, Dulery R, Di Blasi R, et al. Determinants of outcome in Covid-19 hospitalized patients with lymphoma: a retrospective multicentric cohort study. EclinicalMedicine. 2020;27:100549. 75. Bachy E, Cerhan JR, Salles G. Early progression of disease in follicular lymphoma is a robust correlate but not a surrogate for overall survival. Blood Adv. 2021;5(6):1729-1732. 76. Cheson BD, Fisher RI, Barrington SF, et al. Recommendations for initial evaluation, staging, and response assessment of Hodgkin and non-Hodgkin lymphoma: the Lugano classification. J Clin Oncol. 2014;32 (27):3059-3067.

haematologica | 2022; 107(1)

REVIEW SERIES

Prospects in the management of patients with follicular lymphoma beyond first-line therapy

Ferrata Storti Foundation

David Qualls1 and Gilles Salles1,2 Lymphoma Service, Department of Medicine, Memorial Sloan Kettering Cancer Center and 2Weill Cornell Medicine, New York, NY, USA

ABSTRACT

he management of patients with relapsed or refractory follicular lymphoma has evolved markedly in the last decade, with the availability of new classes of agents (phosphoinositide 3-kinase inhibitors, immunomodulators, epigenetic therapies, and chimeric antigen receptor T cells) supplementing the multiple approaches already available (cytotoxic agents, anti-CD20 antibodies, radiation therapy, radioimmunotherapy, and autologous and allogeneic transplants). The diversity of clinical scenarios, the flood of data derived from phase II studies, and the lack of randomized studies comparing treatment strategies preclude firm recommendations and require personalized decisions. Patients with early progression require specific attention given the risk of histological transformation and their lower response to standard therapies. In sequencing therapies, one must consider prior treatment regimens and the potential need for future lines of therapy. Careful evaluation of risks and expected benefits of available options, which vary depending on location and socioeconomics, should be undertaken, and should incorporate the patient’s goals. Preserving quality of life for these patients is essential, given the likelihood of years to decades of survival and the possibility of multiple lines of therapy. The current landscape is likely to continue evolving rapidly with other effective agents emerging (notably bispecific antibodies and other targeted therapies), and multiple combinations being evaluated. It is hoped that new treatments under development will achieve longer progression-free intervals and minimize toxicity. A better understanding of disease biology and the mechanisms of these different agents should provide further insights to select the optimal therapy at each stage of disease.

Introduction While the median overall survival (OS) of patients with follicular lymphoma (FL) was under 10 years over two decades ago, the vast majority (~80%) of patients diagnosed today are likely to be alive 10 years after their initial diagnosis, and their expected median OS may exceed 20 years.1 This remarkable progress reflects the improved efficacy of first-line therapeutic interventions with the introduction of anti-CD20 antibodies, and our ability to deliver active therapies in subsequent lines for those patients experiencing disease progression. Several important challenges should be considered regarding the treatment of patients with disease progression after first-line therapy, who are still generally considered to have an incurable disease after the lymphoma recurrence. The first challenge is the possibility of histological transformation – the leading cause of death of patients with FL.2 Transformation can occur any time during the disease course, although some data suggest an increased risk in the early years after diagnosis which decreases thereafter.2-4 This feared event should always be anticipated, and a new tumor biopsy performed whenever lymphoma progresses or does not respond to therapy. The use of positron emission tomography/computed tomography to guide the site of biopsy in the nodal area with the highest glucose uptake is recommended. The detailed management of transformed FL has been recently reviewed5 and is beyond the scope of this paper. The second consideration is the optimal sequencing of available therapies.6

haematologica | 2022; 107(1)

Haematologica 2022 Volume 107(1):19-34

Correspondence: GILLES SALLES sallesg@mskcc.org Received: August 31, 2021. Accepted: October 5, 2021. https://doi.org/10.3324/haematol.2021.278717

D. Qualls and G. Salles

While some clinical trials were conducted in specific populations of patients, such as those with disease refractory to rituximab or to rituximab and alkylating agents (“double refractory”), there are few randomized clinical trials to guide our choice of a given therapy at a precise time. This is in part due to the broad array of options available after diagnosis, with a variety of mechanisms of action (Figure 1), and to the heterogeneity of patients entering clinical trials evaluating new agents in the relapsed/refractory setting. However, not all available drugs or regimens have similar clinical efficacy - both in terms of response rate, and more importantly, response duration.

A third key element is the patients’ quality of life, including short-term treatment-related side effects and convenience of therapeutic administration, as well as long-term and cumulative toxicities, which may result in cardiac, hematopoietic, infectious, or neurological comorbidities, or increase the risk of potentially fatal secondary malignancies.7 Attractive oral therapies have emerged, although they have their own toxicity profiles, and are often administered for indefinite periods, presenting unique challenges. Overall, these different elements should be discussed with every patient, with transparency regarding the benefits and risks of each strategy, keep-

Figure 1. Mechanisms of action and opportunities for synergy in follicular lymphoma-directed therapies. From top left, clockwise: Autologous chimeric antigen receptor T cells are engineered to target the CD19 epitope on follicular lymphoma (FL) cells, with co-stimulatory domains activating T-cell anti-tumor responses. Bispecific antibodies bind both CD20 on the lymphoma cells and CD3 on the surface of cytotoxic T cells, activating anti-tumor cytotoxicity. Tazemetostat inhibits EZH2mediated suppression of differentiation genes in FL cells, and inhibits suppression of MHC expression, allowing for greater immune recognition of lymphoma cells. PI3K inhibitors block key molecular signal pathways for the growth and survival of lymphoma cells, and also inhibit T regulatory cell function, which may facilitate immune activation against FL cells. Lenalidomide functions via cereblon-mediated ubiquitination and degradation of transcriptional factors, which is directly cytotoxic to lymphoma cells, and potentiates the immune synapse, improving T-cell- and NK-cell-mediated recognition and killing of lymphoma cells. Lenalidomide is synergistic with monoclonal antibodies (rituximab and tafasitamab), which together promote antibody-dependent cellular phagocytosis and antibody-dependent cellular cytotoxicity. Antibody-drug conjugates bind to the cell expressing the target antigen, are internalized, and deliver their cytotoxic payload directly in the cytoplasm. CD47 on FL interacts with SIRPα on macrophages to inhibit phagocytosis of cancer cells. Anti-CD47 antibodies block this checkpoint interaction, promoting macrophage phagocytosis. CAR: chimeric antibody receptor; TCR: T-cell receptor; IgG: immunoglobulin G; MHC: major histocompatibility complex; BCR: B-cell receptor; Ag: antigen; PI3K: phosphoinositide 3-kinase; TILs: tumor-inflitrating lymphocytes; NK: natural killer.

haematologica | 2022; 107(1)

Recent developments for patients with R/R FL

ing in mind our goals (prolongation of OS while preserving quality of life) as well as patients’ own priorities, which might differ according to their age and their personal history.8

Prognostic factors in patients with relapsed or refractory disease While many discoveries have advanced our understanding of the biology of FL, from patterns of gene mutations to the immune microenvironment, efforts to establish biological parameters that predict individual patients’ outcomes have yielded few reliable indicators to date. The extreme clonal heterogeneity and diversification of FL over time point towards the frequent emergence of divergent clones originating from a common progenitor cell, rather than a selective, directional evolutionary process. One suspects that rapidly resistant and evolutive disease may accumulate alterations accounting for chemoresistance, but such alterations have yet to be identified, aside from genetic changes occurring at the time of histological transformation.9,10 The Follicular Lymphoma International Prognostic Index (FLIPI) has been shown to be predictive of the outcome of patients at the time of first lymphoma progression, and one usually assumes that clinical or biological features identified as prognostic factors at diagnosis may apply at the time of progression, although this has not been properly examined in the last decade in large series.11 The clinical impact of biological characteristics (genomic gain or loss, point mutations, immune and microenvironment changes) present at the time of disease progression has not been evaluated extensively, and represents an area for further investigation, especially given the growing accessibility of therapies with distinct mechanisms of action. In recent years, attention has been drawn to the poor outcome of patients experiencing early disease progression after their initial treatment. This has been well documented after immunochemotherapy, with worse outcomes associated with disease progression within 12 or 24 months of initial treatment, described in shorthand as EFS12 or EFS24/POD24 disease, respectively.12,13 These findings have been reproduced in patients treated with rituximab as a single agent or in combination with lenalidomide, as well as in those having received radiation therapy for localized disease.14,15 Patients under observation for whom an event such as treatment initiation or clinical progression within 12 months occurs were similarly found to have a shorter survival.13 These findings have led some to advocate for more intensive therapy (such as autologous stem cell transplant – ASCT) in patients with POD24 disease.16 However, most of these retrospective studies did not account for cases with histological transformation, which are associated with poor outcomes.17 Notably, recent data indicate that a high proportion of patients with early progression present with histological transformation, with numbers ranging from 20 to 40% in patients having received CHOP (cyclophosphamide, adriamycin, vincristine, prednisone) or CVP (cyclophosphamide, vincristine, prednisone) regimens (containing an anti-CD20 antibody) and from 20 to 75% of those having received bendamustine as a cytotoxic backbone.3,17 haematologica | 2022; 107(1)

The precise evaluation of treatment efficacy in patients with POD24 is hampered by the lack of pre-treatment biopsy requirements in some studies, as well as subtle variations in the definition of POD24 disease in others. However, some new agents or combinations remain active in this setting, suggesting that new approaches might potentially circumvent this adverse disease feature.

Old tools remain useful in the relapsed and refractory settings Repeating chemoimmunotherapy: still a standard of care While promising targeted agents and immunotherapies are now available, treatments incorporating chemotherapy and anti-CD20 monoclonal antibodies remain a mainstay of treatment for the management of relapsed/refractory FL. The choice of agents depends on patients’ initial therapy, with preference for a non-cross-resistant option. We have the most robust clinical data to guide decisionmaking in patients who received first-line therapy with rituximab plus CHOP (R-CHOP) or rituximab plus CVP (R-CVP), wherein bendamustine-based therapy is most frequently utilized. The StiL study demonstrated superior progression-free survival (PFS) with rituximab plus bendamustine (B-R) over rituximab plus fludarabine in a randomized study of relapsed/refractory indolent nonHodgkin lymphoma, with a PFS of 34.2 months for B-R versus 11.7 months for rituximab plus fludarabine, and an OS benefit with B-R.18 In this study, 50% of patients had FL, with the majority having been previously treated with first-line CHOP-based therapy. The GADOLIN phase III clinical trial established the efficacy of obinutuzumab plus bendamustine with obinutuzumab maintenance as another second-line option in patients with rituximabrefractory FL, defined as primary resistance or relapse within 6 months of rituximab-containing therapy. Compared to bendamustine alone, obinutuzumab plus bendamustine and obinutuzumab maintenance demonstrated PFS and OS benefits in the overall cohort of nonHodgkin lymphoma patients and in the subset with FL.19,20 Of note, bendamustine regimens were associated with prolonged T-cell depletion in other studies, and precautions, including anti-microbial prophylaxis, should be considered.21 Important gaps remain in our knowledge regarding second-line therapy. In rituximab-sensitive individuals, rituximab and obinutuzumab have not been compared directly. While GADOLIN investigated rituximab-refractory disease, the lack of rituximab in the control arm also raises questions about whether obinutuzumab is truly superior to rituximab in disease defined as rituximab-refractory. Treatment after first-line bendamustine-based therapy is poorly established as there are no dedicated trials evaluating this population. Large long-term outcome studies in FL have included very few patients receiving first-line B-R, likely representing its more recent adoption.22,23 Convention has been to use non-cross-resistant chemotherapy, such as rituximab or obinutuzumab (R/O) with CHOP or R/O-CVP. One small retrospective study of bendamustine-based re-challenge in patients with chronic lymphocytic leukemia or lymphoma showed a response rate of 78% with B-R in lymphoma, with grade 21

D. Qualls and G. Salles

3-4 hematologic toxicity following 35% of treatment administrations.24 Given the availability of alternative agents, along with concerns regarding cumulative hematologic toxicity and risk for secondary malignancies, we continue to favor R/O-CHOP or R/O-CVP over B-R retreatment if chemotherapy is needed. However, this remains a major information gap in FL therapy. Given the increased use of first-line B-R, more studies in this area are needed. As for B-R, few data are available regarding second-line therapy after the rituximab-lenalidomide (R2) combination has been used as first-line treatment.25 A recent retrospective study evaluated outcomes in patients with relapsed or progressive FL who received R2 as first-line therapy.26 The overall response rate was 78% and median PFS was 38 months after salvage therapy, and the use of chemoimmunotherapy (B-R or R-CHOP) was associated with a significantly longer PFS compared to that achieved with biological agents, supporting the use of immunochemotherapy regimens in this second-line context. Other chemoimmunotherapy regimens have demonstrated activity in relapsed/refractory FL, although they are less commonly implemented. These include fludarabinebased combinations (with cyclophosphamide, mitoxantrone, or both), rarely utilized because of their substantial hematologic toxicities, immunosuppression, and risk of secondary neoplasias.27 Cytarabine-containing regimens (such as R-DHAOX [rituximab, dexamethasone, cytarabine, oxaliplatin] and R-DHAP [rituximab, dexamethasone, cytarabine, cisplatin]) showed encouraging results in a retrospective study in patients with early relapsed or refractory disease, often as a path towards high-dose therapy and ASCT.28 The use of chlorambucil (and less commonly cyclophosphamide) in combination with rituximab represents an alternative in comorbid or frail patients.29,30

The role of anti-CD20 maintenance after second line chemoimmunotherapy Following successful second-line treatment with response, maintenance therapy with rituximab or obinutuzumab is often considered. Two large randomized clinical trials by the European Organisation for Research and Treatment of Cancer and the German Low Grade Lymphoma Study Group demonstrated PFS benefit with maintenance rituximab after second-line chemoimmunotherapy.31,32 A subsequent meta-analysis of individual patients’ data from randomized clinical trials evaluating maintenance rituximab found an OS benefit with the use of maintenance rituximab after second-line induction therapy.33 However, many patients in these trials had not received rituximab during their frontline management (and specifically no anti-CD20 maintenance), and since these trials were completed a number of new treatment options for FL have been approved, which may affect any survival benefit with maintenance rituximab. The decision of whether to pursue maintenance rituximab or obinutuzumab should be based on prior therapy and the most recent progression-free interval might be considered. For patients with rituximab-refractory disease, obinutuzumab-based treatment as per GADOLIN would be preferred, whereas in rituximab-sensitive disease continued rituximab is reasonable. In these studies, maintenance treatment was generally well tolerated, although with an increase in grade 3-4 infections.

Should a therapeutic response in the relapsed or refractory setting be consolidated with transplantation? Consolidation high-dose chemotherapy with stem cell rescue can be considered in the relapsed/refractory setting, particularly in those with high-risk disease. An early randomized study of relapsed/refractory FL noted PFS and OS benefits with high-dose chemotherapy and ASCT after initial response to chemotherapy (predominantly CHOP), as opposed to additional chemotherapy alone, although it is difficult to extrapolate these findings which preceded the routine use of rituximab.34 Other studies have demonstrated durable responses following consolidative autologous transplant (with a median PFS exceeding 5 years in one large study), with a suggestion of greater benefit in patients with early progression.35-38 It should be considered that better disease control in patients presenting with histological transformation may have contributed to some of the benefit seen with ASCT in early progression, given the high rates of such transformation in POD24. One study stratifying outcomes by histological transformation showed a clear OS benefit with ASCT in those with transformation, while in those without, ASCT was not associated with improved survival.39 Overall, the sustained PFS benefit achieved with ASCT for patients with FL should be balanced with the immediate and long-term toxicities of the procedure, particularly the risk of secondary malignancies.40 Indications for ASCT also need to be interpreted in the context of alternative treatment options, such as antibody-based therapies and, more recently, chimeric antigen receptor (CAR) T-cell therapy. Allogeneic transplant with reduced intensity conditioning also remains a viable option with prolonged periods of disease-free survival. However, this must be balanced with the relatively high treatment-related morbidity and mortality, with other options including chemoimmunotherapy, CAR T cells, and other immune-based therapies offering more favorable toxicity profiles. We consider this option for patients with multiply refractory FL, particularly younger, otherwise healthy patients with good performance status and adequate candidate donors.

Radiation and radioimmunotherapy: potent tools for local and systemic treatment FL is a highly radiosensitive disease, and both external radiation and radio-immunotherapies are effective tools in the management of relapsed or refractory FL. While extended-field radiation with high doses might expose patients to long-term sequelae, focal radiotherapy to symptomatic lymph nodes represents an efficacious approach which is easy to administer. The optimal dose of radiation in this scenario is an area of active discussion. A recent study comparing standard 24 Gy dosing with a 2 x 2 (4 Gy) regimen demonstrated significantly superior local control with higher dose therapy (5-year progression-free rates of 89.9 vs. 70.4%, respectively), although no difference in OS was observed, and more acute toxicity was seen in the 24 Gy arm.41 In our hands, a 2-year local control rate of 70% has been achieved with 2 x 2 Gy in patients with recurrent disease, and this option appears remarkably versatile in patients with lesions <6 cm (Imber B et al., Blood Adv, in press). Based on the lack of curative intent in the relapsed/refractory setting, reduced

haematologica | 2022; 107(1)

Recent developments for patients with R/R FL

toxicity, and the opportunity for re-treatment, we favor the 2 x 2 Gy approach for local treatment of symptomatic relapsed/refractory FL. For systemic treatment, radioimmunotherapy was developed in the last two decades with 131I coupled to tositumomab and 90Y-ibritumomab tiuxetan, which produced high rates of response.42 Various logistical and economic considerations account for the lack of widespread use of these agents, and only ibritumomab remains available in North America and Europe. Furthermore, since both agents target the CD20 antigen, their efficacy in patients recently exposed to rituximab is possibly hindered by competition with rituximab for target antigens. To circumvent this, radioimmunoconjugates targeting different cell surface antigens have been developed. Encouraging preliminary results were noted with 177Lulilotomab satetraxetan, a CD37-directed antibody, in patients with refractory FL, and a registration phase II study is currently underway (NCT01796171).43

Targeted therapies: exploiting tumor cell vulnerabilities Benefits and limits of phosphoinositide 3-kinase inhibitors FL and other lymphomas exhibit constitutive overactivation of the phosphoinositide 3-kinase (PI3K)-AKT-mecha-

nistic target of rapamycin (mTOR) pathway, which has important roles in proliferation, growth, and survival of both normal and cancer cells.44 Signaling through the B-cell receptor and other cell-surface receptors, which might be critical for lymphoma survival, also depends on this pathway. As a result, PI3K is now a target of significant interest in treating FL, with a number of targeted inhibitors demonstrating activity in multiply refractory disease, and four agents approved in the USA (Table 1).45-49 Approved PI3K inhibitors have all been evaluated in single-arm phase II studies, limiting comparison between PI3K inhibitors or with other treatment modalities (Table 1). Response rates were between 42% and 66% and the median PFS was between 10 and 13 months for the four agents, which were all evaluated in FL that had relapsed after or had been refractory to at least two prior lines of treatment. Differences in the efficacy and toxicity of these inhibitors are largely related to the specific subunits targeted by the therapy.44 Class 1 PI3K is composed of a heterodimer containing a p110 subunit (a, b, g or d) and a regulatory subunit. The g and d subunits are more specifically expressed in leukocytes, whereas the a and b subunits are widely expressed in normal tissue, although all have been implicated in lymphomagenesis. Idelalisib, the first approved PI3K inhibitor, is d-specific, and was shown to generate high response rates in FL refractory to at least two prior lines of treatment including anti-CD20 and alkylator therapy.45,46 Complete

Table 1. Selected trials and outcomes with targeted therapies in relapsed or refractory follicular lymphoma.

PI3K inhibitors (approved) Idelalisib46 Duvelisib47 Umbralisib48 Copanlisib49 Copanlisib + rituximab53 Rituximab control arm

Phase

Prior lines of therapy

N. of patients with FL

ORR

CRR

PFS, median DoR, median (months) (months)

II II II II III III

≥2 ≥2 ≥2 ≥2 ≥I ≥1

72 83 117 104 184 91

66 42 45 59 85 54

14 1 5 20 37 21

11 10 11 13 22 19

11 10 11 14 20.4* 17.3*

2y: 70% 2y: 60% N/A 2y: 69% 3y: 83% 3y: 81%

II II Ib Ib

≥2 ≥2 ≥1 ≥1

74 22 9 39

72 64 78 79

13.5 13.6 N/A N/A

15.8 N/A N/A N/A

NR N/A N/A N/A

N/A N/A N/A N/A

II Ib Ib Ib

≥2 ≥1 ≥1 ≥1

110 12 13 36

21 33 39 72

11 8 8 39

4.6 N/A N/A 25

19.4 NR NR NR

1y 78% N/A N/A N/A

II II

≥1 ≥1

23 39

61 54

N/A 26

7.2* 12.7

N/A 13.3

median: 29.4 mo 3y: 73%

I II II II

≥1 ≥1 ≥1 ≥1

29 51 51 52

38 84 84 35

17 75 69 17

11 N/A N/A 6.6

27 N/A N/A N/A

N/A N/A N/A N/A

PI3K inhibitors (ongoing studies) Parsaclisib (daily dosing)51 Parsaclisib (weekly dosing)51 Zandelisib + rituximab52 Zandelisib

BTK inhibitors Ibrutinib54 Acalabrutinib56 Acalabrutinib + rituximab56 Zanubrutinib + obinutuzumab57

mTOR inhibitors Everolimus58 Temsirolimus59

BCL2 inhibitors Venetoclax60 Venetoclax - BR61 BR control arm Venetoclax - rituximab61

*Overall population including patients with chronic lymphocytic leukemia/small cell lymphoma, marginal zone lymphoma and follicular lymphoma. Data were collected from published papers and abstracts; cross-trial comparisons are not possible because of different eligibility criteria and other differences. FL: follicular lymphoma; ORR: overall response rate; CRR: complete response rate; PFS: progression-free survival; DoR: duration of response; OS: overall survival; PI3K = phosphoinositide 3-kinase; BTK: Bruton tyrosine kinase; mTOR: mechanistic target of rapamycin; BR: bendamustine and rituximab; NR: not reached; N/A: not available or not reported; y: years; mo: months.

haematologica | 2022; 107(1)

D. Qualls and G. Salles

responses, while rare, appear to be durable, and idelalisib has been shown to have a similar efficacy in patients who experienced early progression after first-line therapy.45,50 Copanlisib, a pan-PI3K inhibitor with specificity for the a and d isoforms, had similar activity in patients who had received at least two lines of therapy but requires weekly intravenous infusions for 3 or 4 weeks.49 Duvelisib is a dand g-specific PI3K inhibitor and umbralisib is a d- and casein kinase-1-specific inhibitor. With the limitation of cross-trial comparisons and a possible learning curve during the development of the different PI3K inhibitors, umbralisib appears to have a better tolerability than that of idelalisib or duvelisib.48 There are major limitations regarding PI3K inhibitor monotherapy, including the limited duration of response with a median PFS of approximately 1 year and toxicities with chronic therapy. Class toxicities, including diarrhea/colitis, transaminase elevations, rash, neutropenia, and rarely pneumonitis, along with infectious complications, limit the utility of these agents. Due to the a isoform specificity of copanlisib, toxicities including hypertension and hyperglycemia are unique to copanlisib, and while transient, represent a potential limitation to its use in certain populations. Other PI3K inhibitors are in development and toxicity profiles may improve with changes in PI3K specificity and dosing regimens.51 Currently approved oral PI3K inhibitors are administered on a daily or twice-daily basis throughout treatment. Zandelisib and parsaclisib, d-specific PI3K inhibitors, are being evaluated using unique dosing schedules intended to mitigate toxicity. Zandelisib was administered on days 1-7 of each 28-day cycle starting in cycle 2 in a phase Ib study.52 A phase II study of parsaclisib is evaluating both daily dosing and weekly dosing after an initial 8-week daily dosing period.51 The use of combination therapies may offer the greatest opportunity to leverage the efficacy of PI3K inhibitors while addressing the limited duration of responses. A randomized phase III trial, CHRONOS-3, evaluated rituximab with or without copanlisib in relapsed indolent nonHodgkin lymphoma (60% FL), and demonstrated a PFS benefit from copanlisib with rituximab versus placebo plus rituximab.53 The relatively modest prolongation in response duration (20.4 vs. 17.3 months) for patients with FL should be weighed with the constraints of frequent infusions. Further areas of exploration include the combination of other targeted therapies such as BCL2 and mTOR inhibitors. However, caution is still required, as several studies evaluating combinations, including idelalisib plus entospletinib, idelalisib with lenalidomide and rituximab, and dactolisib with abiraterone or everolimus have resulted in unacceptable toxicity profiles. There is also potential for synergy with immunotherapies via T regulatory cell inhibition, but caution should be employed, as PI3K inhibitor-associated hepatotoxicity and colitis are thought to be immune-mediated and may worsen with immuneactivating therapies such as checkpoint inhibitors.

patients with double-refractory disease, the overall response rate observed with ibrutinib was only 21%, with a complete response rate of 11%.54 While the duration of response was 19 months, the median PFS was only 5 months (Table 1). These disappointing results might be explained by the biological heterogeneity of FL, the presence of mutations in BTK, or bystander effects of the drug on the microenvironment.55 A large randomized study (NCT01974440) combining ibrutinib with standard immunochemotherapy regimens has completed accrual, and results are anticipated shortly. Other combinations with ibrutinib are being evaluated and other BTK inhibitors (acalabrutinib, zanubrutinib) are also being tested, including a randomized study of obinutuzumab with or without zanubrutinib (NCT03332017).56,57 It is hoped that these studies will clarify the potential role of BTK inhibitors in patients with FL.

Inhibitors of the mTOR pathway: a missed opportunity? It is also noteworthy that the mTOR pathway is dysregulated in a substantial proportion of cases of FL, particularly those with mutations activating RRAGC, ATP6V1B2 and ATP6AP1 and inactivating the upstream regulator SESTRIN1. The oral mTORC1 inhibitor everolimus was found to be active with an overall response rate of 61% (and a median response duration of 11.5 months) (Table 1).58 Temsirolimus, which requires intravenous administration, achieved overall and complete response rates of 54% and 26%, respectively, and the median PFS was 13 months.59 Of note, these clinical results were observed prior to our knowledge of the molecular alterations involving this pathway, and it would be interesting to evaluate whether their presence is associated with higher efficacy of mTOR inhibitors.

The unexpected results of BCL2 inhibitors With persistent BCL2 expression linked to the hallmark t(14:18) translocation, one would expect that BCL2 inhibitors would demonstrate activity in FL. In the initial phase I study, response to single-agent venetoclax was observed in 38% of 29 patients (with 17% reaching a complete response), and the median PFS and duration of response were 11 and 27 months, respectively (Table 1).60 Based on these preliminary results, venetoclax was combined with rituximab (single-arm study) or with B-R (randomized phase II study) to gauge its clinical value. Results demonstrated a lack of significant benefit from the addition of venetoclax to B-R (while toxicity was significantly increased) and only modest benefit when venetoclax was added to rituximab.61 Several hypotheses for this limited activity have been raised, including: (i) the possibility that BCL2-deregulated expression is critical in early steps of the development of FL, but that tumor cells subsequently acquire other genetic alterations and no longer depend on BCL2 for their survival; (ii) the role of alternative actors of the apoptotic machinery (BCLX-L) in protecting these cells from dying; and (iii) mutations in the BCL2 gene that can impair venetoclax binding.62

A challenging path for Bruton tyrosine kinase inhibitors

Epigenetic therapies

FL cells express functional B-cell receptors, and depend on downstream signaling pathways for survival. This provided a rationale to investigate the activity of Bruton tyrosine kinase (BTK) inhibitors in patients. However, in a pivotal single-agent phase II study evaluating 110

Mutations in genes involved in chromatin organization appear as early founding events in germinal center B-cell lymphomagenesis. FL relies on epigenetic dysregulation to allow maintenance of a germinal center phenotype, thereby facilitating cell survival and proliferation.

haematologica | 2022; 107(1)

Recent developments for patients with R/R FL

Epigenetic changes also contribute to immune dysregulation via suppression of major histocompatibility complex expression and tumor infiltrating lymphocyte content. Tazemetostat represents the most successful epigenetic therapy to date in FL.63 Activating mutations in EZH2 occur in about 20% of FL, and result in aberrant silencing of differentiation genes, allowing continued survival and proliferation of FL (Figure 2). Tazemetostat inhibits EZH2, allowing for normal transcription of repressed genes. In phase I and II studies evaluating tazemetostat monotherapy, the overall response rate was 69% in EZH2-mutant FL, with a median PFS of 13.8 months. Tazemetostat also demonstrated activity in EZH2 wildtype FL with an overall response rate of 35% and PFS of 11.1 months. Tazemetostat was well tolerated, with only 4% of patients suffering a serious treatment-related adverse event, a favorable profile for combination therapies. Other EZH2 inhibitors are in early development. Other epigenetically targeted agents are also under investigation. Histone deacetylase inhibitors counteract the epigenetic impact of CREBBP and EP300 mutations (Figure 2), and have demonstrated activity in FL. Four phase II studies have documented response rates of 4764% with abexinostat or vorinostat, but with significant

hematologic and gastrointestinal tract toxicity.64-67 In each of these studies, higher response rates to histone deacetylase inhibition were seen in FL than in other types of nonHodgkin lymphoma, suggesting a particular sensitivity to epigenetic modulation. As yet no studies have evaluated whether mutations in EP300 or CREBBP predict for response in FL, and this may be worth exploring; however, it is interesting to note that in a study of the histone deacetylase inhibitor panobinostat in diffuse large B-cell lymphoma, CREBBP or EP300 mutations were not associated with response.68 Additional in vivo work suggests that inhibitors of the demethylase KDM5 may restore normal histone methylation and gene expression in KMT2Dmutant disease, another gene commonly affected in FL.69

Immunotherapies: early successes and promising potential Single-agent anti-CD20 therapy remains viable for selected patients In patients with relapsed disease and low tumor burden following initial anti CD20-containing therapy, re-challenge with monoclonal antibodies, including rituximab

Figure 2. Epigenetic dynamics in follicular lymphoma. Frequent epigenetic mutations in follicular lymphoma result in repression of mature B-cell differentiation factors and cell regulators, preventing terminal differentiation and exit from the cell cycle. This allows a germinal center phenotype, genomic instability, and cellular proliferation to persist. (A) Activating mutations in EZH2 result in aberrant trimethylation of lysine 27 in histone 3 (H3K27me3), resulting in transcriptional silencing and lymphomagenesis. Tazemetostat inhibits EZH2, restoring normal H3K27 methylation. (B) Loss-of-function mutations in the histone acetyltransferases CREBBP and EP300 result in decreased acetylation of histones H3K27 and H3K18, shifting the balance toward histone deacetylation by HDAC3. Histone deacetylase inhibitors prevent this deacetylation. (C) KMT2D is a histone methyltransferase and is frequently inactivated in follicular lymphoma, favoring demethylation of H3K4 and repression of key differentiation genes. Inhibitors of the demethylase KDM5 restore normal H3K4 methylation and gene expression in KMT2D-mutant lymphoma. HDAC: histone deacetylase.

haematologica | 2022; 107(1)

D. Qualls and G. Salles

and obinutuzumab, remains an option. Prior sensitivity to anti-CD20 treatment is essential when considering this option. Sensitivity has classically been defined as having a response to anti-CD20 treatment lasting at least 6 months.70-72 Several studies have evaluated anti-CD20 monotherapy re-challenge in the second or later line of therapy. A phase II trial of rituximab re-challenge in patients with relapsed indolent non-Hodgkin lymphoma (92% FL) demonstrated an overall response rate of 40%, with 11% attaining complete responses and the median time to progression being 17.8 months.73 A case-control study demonstrated response rates of 76% with second-line rituximab monotherapy in FL, and a median PFS of 19.2 months.74 In the RESORT clinical trial, patients with low tumor burden FL with initial response to rituximab monotherapy were randomized to maintenance rituximab or re-treatment with rituximab at progression of disease, with patients in the re-treatment cohort demonstrating response rates of 61% at first re-treatment and 67% at second re-treatment.70 With regard to the choice of anti-CD20 agent, obinutuzumab has not been proven to be significantly superior to rituximab, but it can be used in patients with adverse reactions to rituximab, while ofatumumab appears inferior to rituximab.71,75 We consider single-agent rituximab (or rarely obinutuzumab) a useful option in instances of mild tumor burden where an indication to treat nonetheless exists and when rituximab-based combinations - providing deeper and more durable responses - are not suitable.

Immunomodulators Lenalidomide is an immunomodulatory agent that functions via cereblon-mediated ubiquitination and degradation of target transcriptional factors, resulting in potential direct cellular toxicity to lymphoma cells as well as potentiation of immune effector cells in the tumor microenvironment. Lenalidomide has demonstrated synergy with anti-CD20 antibodies such as obinutuzumab and rituximab, via reconstitution of a functional immune synapse, improved natural killer cell function, and greater antibody-dependent cellular cytotoxicity.76-78 This benefit was demonstrated in the phase III AUGMENT trial, which evaluated rituximab with or without lenalidomide in relapsed rituximab-sensitive FL, and found a significant PFS benefit with the combination of lenalidomide plus rituximab (R2).72 A post-hoc analysis also indicated an OS benefit for the combination, although patients in both study arms had very favorable outcomes (2-year OS of 86% and 95% in the rituximab/placebo and R2 arms, respectively), likely reflecting the select charactersitics of the study population. Of note, infections, neutropenia and skin rashes were more frequently encountered in patients receiving R2. Based on these results, and given the similar efficacy of this combination when compared with classical immunochemotherapy in the first-line setting (as demonstrated in the RELEVANCE study), R2 is now a standard of care for relapsed/refractory, rituximabsensitive FL following anti-CD20 or chemoimmunotherapy.25 A single-arm, phase II study of obinutuzumab and lenalidomide in relapsed/refractory disease, including rituximab-refractory FL, showed similarly high response rates.79 Immunomodulators have intriguing potential in terms 26

of synergy, particularly with other immunotherapies, as already evidenced with rituximab and obinutuzumab. Further studies evaluating the addition of immunotherapies to the R2 or obinutuzumab-lenalidomide backbone are underway. For example, a phase Ib/II study evaluated the use of atezolizumab combined with obinutuzumablenalidomide in patients who had received at least one prior line of immunochemotherapy, with encouraging rates of durable clinical responses (3-year PFS 68.4%, 3year OS 90%).80 Another randomized study is evaluating the combination of R2 with tafasitamab, a CD19-directed Fc-engineered antibody, which had modest activity as a single agent in FL, but produced encouraging response rates in diffuse large B-cell lymphoma when combined with lenalidomide (NCT04680052).81

Cellular therapies Recently, the Food and Drug Administration in the USA approved axicabtagene ciloleucel as the first CAR T-cell therapy for FL, based on results of the ZUMA-5 trial.82 This study evaluated axicabtagene ciloleucel CAR T-cell therapy in patients with indolent non-Hodgkin lymphoma, including 127 with FL, who had relapsed/refractory disease after two or more prior lines of therapy, including anti-CD20 and alkylator-containing therapy. The overall response rate was 94% and complete response rate 79% for FL patients; at 18 months PFS was 69% and OS was 88% (Table 2). Similar response rates, but a decreased PFS rate at 18 months, were noted in patients with POD24 disease compared to those without (55% vs. 84%, respectively).83 The results of this trial have been compared to those of an external propensity scorematched cohort of patients not enrolled in the study who otherwise met criteria for ZUMA-5, showing significant improvement in overall response rate, PFS, time to next treatment, and OS with CAR T-cell therapy compared to alternative therapies.84 Low-grade cytokine release syndrome occurred in the majority of patients with FL, while 6% experienced grade ≥3 cytokine release syndrome; notably, 50% of patients received tocilizumab during therapy, indicating that close monitoring, similar to that necessary with large cell lymphoma, will still be required. For patients with FL, the frequency (56%) and severity (15% with grade ≥3 events) of neurological events, and their duration (median of 14 days), represent a significant hurdle for broader application of this therapy. Interestingly, re-treatment with axicabtagene ciloleucel in patients who relapsed after initial therapy with this drug also generated favorable responses. Thirteen patients (11 with FL) who had undergone initial axicabtagene ciloleucel therapy, as per ZUMA-5, and achieved an initial response followed by progression, were treated with repeated axicabtagene ciloleucel administration after it had been confirmed that CD19 expression persisted on FL cells. The response rate was 100% with 77% achieving a complete response, and 58% being progression-free at 12 months after re-treatment. Evaluation of other CAR T-cell products is also underway. A phase II study of tisagenlecleucel in relapsed/refractory FL after two or more lines of prior therapy is ongoing.85 At the most recent evaluation, there was a high overall response rate of 86% (95% CI: 5675%) with a complete response rate of 66% (95% CI: 7892%) and PFS at 6 months of 76% (95% CI: 65-84%). A favorable toxicity profile was noted with cytokine release haematologica | 2022; 107(1)

Recent developments for patients with R/R FL

Table 2. Selected chimeric antigen receptor T-cell and bispecific antibody trials and outcomes.

CAR T-cell product Axicabtagene ciloleucel82 Tisagenlecleucel85

Bispecific antibody Mosunetuzumab92 Glofitamab93 Epcoritamab94 Odronextamab95

Anti-CD47 antibody Magrolimab97

Phase II II

Phase I I I I

Phase Ib/II

Prior lines N. of patients ORR of treatment with FL (%) ≥2 ≥2

84 94

94 86

Prior lines N. of patients ORR of treatment with FL (%) ≥2 ≥1 ≥1 ≥1

62 44 12 28

68 71 90 93

Prior lines N. of patients ORR of treatment with FL (%) ≥1

66*

CRR PFS DoR (%) (time point, %) (time point, %)

CRS, % (grade III-IV)

ICANS, % (grade III-V)

79 66

18 mo, 69% 6 mo, 76%

18 mo, 69% N/A

78 (6) 48.5 (0)

56 (15) 9.3 (1.0)

CRR (%)

PFS, median (mo)

DoR, median (mo)

CRS, % (grade III-IV)

ICANS, % (grade III-V)

50 48 50 75

11.8 11.8 N/A 12.8

20.4 10.8 N/A 7.7

23 (1.6) 50.3 (3.5)* 59 (0)* 62.2 (7.1)*

45 (0) 3.5 (1.2)* 8 (4)* (3.9)*

CRR (%)

PFS, median (mo)

DoR, median (mo)

Anemia, % (grade III-IV)

24*

N/A

27 (15)

*All patients with non-Hodgkin lymphoma, those with follicular lymphoma not reported individually. Data were collected from published papers and abstracts; comparisons cross trials are not possible because of the different eligibility criteria and other differences. FL: follicular lymphoma; ORR: overall response rate; CRR: complete response rate; PFS: progression-free survival; DoR: duration of response; CRS: cytokine release syndrome; ICANS: immune effector cell associated neurotoxicity; N/A: not available or not reported; mo: months.

syndrome occurring in 49% (all grade 2 or below) and neurotoxicity in 9% (all grade 2 or below, apart from a single grade 4 event which was self-limited). Another study is evaluating the activity of lisocabtagene maraleucel in the same setting (NCT04245839). An array of clinical trials is in development with the aim of improving CAR T-cell therapy in B-cell lymphoma.86 Possible avenues include the use of combination products such as BTK inhibitors and checkpoint inhibitors, bispecific CAR targeting multiple lymphoma epitopes, and armored CAR products intended to overcome the immunosuppressive tumor microenvironment. The current CD19-directed CAR T-cell results are encouraging, although longer-term data will be necessary to define the ultimate value of these treatments. There would be obvious appeal if this single-administration therapy produced prolonged disease-free remissions in a significant proportion of patients. Additionally, mitigation of cytokine release syndrome and neurotoxicity, either via supportive measures or less toxic CAR T-cell products, would broaden the applicability of these therapies, particularly if they can be administered in the outpatient setting. Assuming long-term efficacy is established, and with improvements in the avoidance and management of toxicities, CAR T cells will represent an appealing option in the therapeutic algorithm.

Antibody-drug conjugates Antibody-drug conjugates enable targeted delivery of potent cytotoxic drugs to tumor cells, minimizing systemic toxicity. Such agents, targeting common B-cell antigens, have been developed across the spectrum of B-cell lymphomas, with polatuzumab vedotin and loncastuximab tesirine recently approved for patients with diffuse large B-cell lymphoma.87,88 Polatuzumab vedotin targets the CD79b antigen and has shown promising response rates alone or in combination with rituximab for patients with FL (overall response rate 70%, complete response rate 45%).89 However, high rates of cumulative peripheral neuropathy were noted when prolonged cycles were administered, a significant concern in the context of the chronicity of FL. Furthermore, when combined with B-R in a randomized phase II study, polatuzumab did not improve the complete response rate or PFS for patients haematologica | 2022; 107(1)

with FL, in contrast to the observations made in patients with diffuse large B-cell lymphoma.87 Several combinations of polatuzumab with other agents active in FL are currently being investigated.90 The CD19-directed antibody loncastuximab tesirine also showed activity in 14 patients with FL with an overall response rate of 79% and a complete response rate of 64%.91 Combination studies with rituximab (NCT04998669) and a comparison of this agent versus idelalisib (NCT04699461) are planned.

Bispecific antibodies A novel approach to lymphoma immunotherapy has been the use of bispecific T-cell engager therapies, which incorporate a CD3-binding component that engages T cells and a second tumor antigen-binding domain, most commonly CD20 in B-cell lymphoma. These agents function to induce activation and cytotoxic activity of T cells against CD20-expressing lymphoma cells. Like autologous CAR T-cell therapy, bispecific T-cell engagers recruit the patient’s native T cells to target lymphoma, with the advantage of being an “off-the-shelf” option that bypasses the time and logistical challenges entailed in genetic modification of T cells. Early-phase clinical trials have demonstrated very encouraging results, with high response rates of 68-90% in patients with relapsed/refractory FL, including complete responses in the majority of patients (Table 2).92-95 Limited follow-up is currently available to evaluate response duration, although preliminary data demonstrated durable responses lasting at least 18 months in patients having achieved a complete response after completion of therapy. Toxicities of concern include cytokine release syndrome and neurotoxicity, similar to the toxicities of CAR T-cell therapy, although these events were only observed during the initial one or two infusions, and both rates and severity were lower than those with CAR T-cell therapies. Further mitigation of toxicity may be possible via step-up dosing and/or subcutaneous administration. If durable responses are demonstrated and toxicity proves manageable, this class of agents appears to hold very significant promise in FL. Combination studies with other immune interventions or cytotoxic agents are under development (NCT04712097, NCT04663347, NCT04246086) and are evaluating the use of bispecific therapy in earlier phases of FL treatment. 27

D. Qualls and G. Salles

Phagocytosis checkpoint inhibitors Macrophage-mediated phagocytosis is modulated by a series of checkpoints including the CD47-SIRPa checkpoint, classically known as the “don’t eat me signal.”96 Immunotherapies inhibiting this interaction facilitate macrophage activation and phagocytosis, and show promise in FL. Magrolimab, a humanized IgG4 isotype that binds CD47 and blocks interaction with SIRPa, was evaluated in combination with rituximab in a phase Ib/II study of relapsed/refractory indolent non-Hodgkin lymphoma (28 of 29 patients with FL), with an overall response rate of 66% and a median duration of response not reached (range, 6.2-22.6 months).97 Toxicities included infusion reactions and a class toxicity with first-dose anemia, related to splenic phagocytosis of senescent red blood cells. Other CD47-targeted agents such as ALX148 and TTI-622 have also demonstrated activity in FL.98,99 In the preclinical setting, bispecific antibodies targeting CD47 and CD19 have been designed to increase drug specificity and decrease on-target, off-tumor toxicity.100 Outside of the CD47-SIRPa axis, inhibitors of adenosine receptors have been shown to disinhibit antibody-dependent cellular phagocytosis, and show promise in preclinical studies.101 As with other agents, there is increased interest in combining magrolimab or alternative pro-phagocytic agents with other immune therapies. A study showed that T-cell deficiency in mice abrogated the response to anti-phagocytic agents, and suggested that CD8+ cytotoxic T cells

play an important role in the mechanism of action of prophagocytic therapies.102 CD47 blockade also enhances the ability of dendritic cells to cross-prime T cells.102 Given these findings, there is a mechanistic rationale for the addition of a checkpoint inhibitor or T-cell co-stimulatory agent to anti-CD47 therapy. There is also in vitro evidence that chemotherapy administered prior to phagocytosis checkpoint inhibitors improves tumor response and augments host memory response against relapsing tumors. Further studies of combination therapies are needed, as well as a better basic understanding of the interactions between the innate and adaptive immune responses to lymphoma.102

Checkpoint inhibitors and T-cell co-stimulatory agents The immune infiltrate in FL is enriched in PD-1-positive immune cells, suggesting that immune tolerance plays a key role in lymphomagenesis, and that disruption of these signals via checkpoint blockade would be efficacious.103 Prior studies have investigated mechanisms of Tcell rescue via checkpoint blockers targeting the PD-1/PDL1 axis.104-108 While preliminary data suggested encouraging response rates from this treatment in combination with anti-CD20 monoclonal antibodies, no definitive benefit has been established, and single-agent activity has been disappointing, with response rates of 4-9%. There has also been interest in the utility of co-stimulatory agents in T-cell reconstitution and immunotherapy.

Table 3. Selected pivotal trials in relapsed or refractory follicular lymphoma.

Targeted agents Parsaclisib Zandelisib Ibrutinib

Zanubrutinib

Tazemetostat Abexinostat

Mechanism

Phase

PI3K d inhibitor

III

Drugs

Prior lines of therapy

Parsaclisib + anti-CD20 vs. Anti-CD20 monotherapy PI3K d inhibitor II Zandelisib monotherapy BTK inhibitor III Ibrutinib + R-CHOP or BR vs. R-CHOP or BR III Zandelisib – rituximab vs. BTK inhibitor R-CHOP or BR II Zanubrutinib + obinutuzumab vs. (randomized) Obinutuzumab EZH2 inhibitor III Tazemetostat + R2 vs. R2 pan-HDAC inhibitor II Abexinostat monotherapy

N. of planned participants

Status

NCT #

≥1

416

Not yet NCT04796922 recruiting Recruiting NCT03768505 Active, not recruiting NCT01974440

≥2 ≥1

180 403

≥1

534

Recruiting

NCT04745832

≥2

210

Recruiting

NCT03332017

≥1 ≥3

518 139

Recruiting NCT04224493 Active, not recruiting NCT03600441

Immunotherapies Magrolimab Mosunetuzumab

Tafasitamab Tisagenlecleucel

Anti-CD47 MoAb Anti-CD20 x anti-CD3 bispecific antibody Anti-CD19 MoAb Anti-CD19 CAR T

II III

Magrolimab + rituximab ≥1 Mosunetuzumab + lenalidomide vs. ≥1 Rituximab + lenalidomide

422 400

Recruiting Not yet recruiting

NCT02953509 NCT04712097

III II

Tafasitamab + R2 vs. R2 Phase II study (ELARA)

≥1 ≥2

618 97

Radiolabeled (177 Lutetium) Anti-CD37 MoAb Anti-CD19 antibody drug conjugate

IIb

(177Lu)-Lilotomab satetraxetan

≥2

204

Recruiting

NCT01796171

Loncastuximab tesirine vs. Idelalisib

≥2

150

Recruiting

NCT04699461

Recruiting NCT04680052 Active, not recruiting NCT03568461

Other (177Lu)-Lilotomab satetraxetan Loncastuximab tesirine

This list is not exhaustive but outlines the agents discussed in the review. NCT: National Clinical Trials; PI3K: phosphoinositide 3-kinase; BTK: Bruton tyrosine kinase; R-CHOP: rituximab, cyclophosphamide, doxorubicin, vincristine, prednisone; BR: bendamustine and rituximab; EZH2: enhancer of zeste homolog 2; R2: rituximab and lenalidomide; R: rituximab; HDAC: histone deacetylase; MoAb: monoclonal antibody; CAR T: chimeric antigen receptor T cell.

haematologica | 2022; 107(1)

Recent developments for patients with R/R FL

The co-stimulatory receptor 4-1BB (CD137) is found on many activated immune cell subtypes. Activation of 4-1BB is associated with increased T-cell proliferation, survival, cytokine production, and functional maturation.109 Resident T cells in FL have been shown to express 4-1BB at high levels, providing a biological rationale for 4-1BB agonist therapy.109 However, early clinical trials showed only modest anti-tumor activity, with the therapeutic index limited by dose-related hepatotoxicity.110-112 It is thought that the hepatotoxicity seen in these studies was driven by offtumor Fc-mediated 4-1BB cross-linking by Fc𝛾RIIb expressing liver resident cells.113 While there remains a biological basis for 4-1BB co-stimulation in FL, a more tumor-specific mechanism is probably required. A possible solution is the use of bispecific therapies, which fuse 4-1BBL to a lymphoma-specific target such as CD19, minimizing off-target toxicity. This approach has shown synergistic activity with CD3/CD20-targeted bispecific therapy in vitro and in vivo, and clinical trials are underway (NCT04077723).114

Sequencing therapies in patients with relapsed/refractory follicular lymphoma When and how to utilize the growing array of therapies at our disposal remains a significant challenge. It is important to note that observation in many cases of relapsed or refractory FL remains a viable option if symptom burden is low and no absolute indications for treatment are met, such as the Groupe d'Etude des Lymphomes Folliculaires (GELF) criteria. As shown above, the number of options available for the management of patients requiring second and subsequent lines of therapy has increased markedly in the last decade, and newcomers are emerging. Several pivotal trials are under development, assessing the therapeutic value of single agents or various combinations of active compounds, and emphasizing the

probable expansion of the field (Table 3). However, the lack of head-to-head evaluations of recently approved agents, and the usual pitfalls of cross-trial comparisons, represent obstacles to evidence-based decisions.

Second-line treatment While there is no formal standard of care, the management algorithm (Figure 3) is relatively straightforward, since beside rituximab, alone or in combinations with lenalidomide, immunochemotherapy remains the principal option recommended by the National Comprehensive Cancer Network and European Society for Medical Oncology guidelines in the absence of histological transformation. These options achieve a clinical response in the majority of patients, which can be sustained for years. Choices among the different cytotoxic regimens and anti-CD20 antibodies have already been discussed above, as has the role of consolidation with anti-CD20 maintenance or ASCT. In addition to the AUGMENT trial, the results of the front-line RELEVANCE study, which demonstrated comparable efficacy between R2 and immunochemotherapy, reinforce the efficacy of R2, and this regimen should be strongly considered if immunochemotherapy was used in the first line of treatment.25 Advantages of this approach include using a strategy with a different mode of action and leaving an open space for initiating immunochemotherapy (with eventual consolidation) later in the management of FL. Beyond prior therapy and disease characteristics, the patient’s ability to tolerate therapy must be considered. In patients of advanced age, who are frail, or have comorbidities, options include rituximab as a single agent or in combination with lenalidomide (with the appropriate dose adaptations according to kidney function), alkylating agents or localized radiation therapy. A personalized approach is recommended, and it remains possible to

Figure 3. Schema for treatment choice at the time of first disease progression. After ruling out histological transformation, the choice of agent depends on prior therapy, the timing of relapse, tumor burden, and patients’ ability to tolerate therapy. R: rituximab; O: obinutuzumab; CHOP: cyclophosphamide, adriamycin, vincristine, prednisone; CVP: cyclophosphamide, vincristine, prednisone: ASCT: autologous stem cell transplantation.

haematologica | 2022; 107(1)

D. Qualls and G. Salles

Figure 4. Schema for treatment choice in third-line and subsequent settings. Given the paucity of randomized data and large number of effective therapies in this setting, treatment is highly personalized, dependent on disease characteristics, prior therapies, and patients’ ability to tolerate treatment. Clinical trials should always be strongly considered. ASCT: autologous stem cell transplantation; HCT: hematopoietic cell transplantation; R2: rituximab and lenalidomide; PI3K: phosphoinositide 3-kinase; mut: mutated; WT: wild-type.

offer substantial benefits for these patients while avoiding significant toxicity. Management in early progression In cases of early progression (i.e., EFS12 or POD24 disease), histological transformation should always be considered. A biopsy should be systematically performed in all cases to rule out such transformation; several clinical featues, such as lack of response during chemotherapy or rapid tumor growth, elevated serum lactate dehydrogenase levels, hypercalcemia, appearance of B symptoms, or rapid decline of the performance status are highly suggestive of histological transformation and might be used as surrogates when a biopsy is not feasible.3 If histological transformation is confirmed, in patients not previously exposed to anthracycline, R-CHOP often constitutes the standard of care. Alternative salvage chemotherapy regimens followed by ASCT are recommended for those previously exposed to anthracycline, or in those unresponsive to R-CHOP. Cellular therapy is an important emerging option for patients with histological transformation but unmet needs remain.5 If histological transformation is ruled out, several options should be considered. Standard options such as obinutuzumab-bendamustine and anti-CD20-lenalidomide combinations are often utilized, and their efficacy in POD24 patients has been demonstrated.20,72,79,115 If an adequate response is achieved, some centers advocate for ASCT consolidation in eligible patients, and this remains a viable option, particularly in younger, fit patients in whom durable response is desirable. Aside from chemoimmunotherapy and R2, other strategies appear to be active in patients with POD24, including some PI3K inhibitors, bispecific antibodies (such as mosunetuzumab), and CAR T cells.50,83,92,116 Decisions in this situation can remain challenging, and studies such as SWOG S1608 30

(NCT03269669) comparing obinutuzumab-chemotherapy versus obinutuzumab-lenalidomide or obinutuzumabumbralisib will be helpful to further elucidate the optimal management of these patients.

Third-line setting and beyond In multiply relapsed and refractory disease, a broad array of effective options (Figure 4) remains available, although the choice of therapy is more individualized and less defined by randomized trials. Decisions depend primarily on patients’ prior treatment, response to therapies, disease-related symptoms and tumor burden. The choice of drug is then based on relative efficacy (limited by cross-trial comparisons), toxicity profiles, response rates, and patients’ priorities. Repeating rituximab-based regimens with chemotherapy or lenalidomide is often feasible if a significant progression-free interval of at least 2 years was achieved with the last treatment. In cases of refractory disease or early progression, changing the class of agents is generally preferred. The choice of agent is also dictated by the anticipated depth and duration of response. Oral agents such as PI3K or EZH2 inhibitors are often considered earlier in treatment despite the low complete response rates observed and the median PFS and duration of response close to 12-14 months. These agents may be better positioned when more efficacious strategies have been exhausted, or as temporary solutions before implementing a more effective therapy. Beside the presence of an EZH2 mutation, there are no biological criteria available to guide us toward a given treatment option. If the patient presents with significant tumor bulk or significant disease-related symptoms, it is probably preferable to use regimens allowing a more rapid and sustained response. In fit patients, if adverse features (high FLIPI, bulky disease, B symptoms, short haematologica | 2022; 107(1)

Recent developments for patients with R/R FL

treatment-free interval) are present, and after having ruled out the possibility of histological transformation, strategies offering a higher response rate and prolonged PFS should be considered with the patient. While ASCT or allogeneic transplant were reasonable options in this setting (if not used before), the availability of CAR T cells despite the limited follow-up presently available - represents an important opportunity to consider prior to transplantation. For all patients, the consideration of a clinical trial is encouraged, given the rapid evolution in the field with effective agents and combinations. Overall, the risk/benefit ratio of each option should be assessed in the context of the patients’ overall condition and their individual goals for therapy.

Remaining gaps and perspectives Several questions remain unanswered regarding the optimal sequencing of new agents. Overall, we know little about how prior treatment affects disease biology, and hence influences the response to the next class of agent. It is suspected that the use of T-cell-depleting agents, such as bendamustine or purine analogues, may affect the quality of host T cells and result in suboptimal results of CAR T-cell therapy, and possibly other immunotherapies such as immunomodulators and bispecific antibodies. Further translational and clinical evaluation regarding the impact of prior treatment on the mechanism of action and efficacy of subsequent treatments is needed. Markers able to predict response to therapy are also lacking. The potential influence of specific molecular alterations has not been explored in the relapsed/refractory setting (except for higher response rates to tazemetostat when an EZH2 mutation is present). It is conceivable that alterations in the B-cell receptor signaling pathway or in the metabolic regulatory pathway may affect the efficacy of agents targeting those pathways. Likewise, mutations affecting MHC expression or T-cell subpopulation might potentially influence some immunotherapies. Further investigations in this area, such as circulating tumor DNA assays to characterize the mutational landscape, will be of interest. Another important challenge is the design and clinical execution of more rational therapeutic combinations, and many possible examples have been outlined above. Phase II studies with early and standardized evaluation end-

References 1. Junlén H, Peterson S, Kimby E, et al. Follicular lymphoma in Sweden: nationwide improved survival in the rituximab era, particularly in elderly women: a Swedish Lymphoma Registry study. Leukemia. 2015;29(3):668-676. 2. Sarkozy C, Maurer MJ, Link BK, et al. Cause of death in follicular lymphoma in the first decade of the rituximab era: a pooled analysis of French and US cohorts. J Clin Oncol. 2019;37(2):144-152. 3. Al-Tourah AJ, Gill KK, Chhanabhai M, et al. Population-based analysis of incidence and outcome of transformed non-Hodgkin’s lymphoma. J Clin Oncol. 2008;26(32):51655169.

haematologica | 2022; 107(1)

points, such as positron emission tomography/computed tomography response or circulating tumor DNA evaluation, are desirable. Despite the difficulty of aligning different stakeholders with potential competing interests, conducting drug development in parallel may allow a more direct comparison of the respective benefits of these different strategies. Evaluating patients’ quality of life and assessing their potential preferences when facing options with different routes of administration, treatment durations and toxicity patterns fully deserves additional methodological efforts and investment. Two major therapeutic mechanisms carry hope for future progress. Given the central role of epigenetic alterations in the development of FL, we anticipate that epigenetic modifiers will be able to eradicate – or substantially deplete - the pool of lymphoma clonal precursor cells, providing an avenue toward cure. Tazemetostat represents an important proof of concept, but the development of more efficacious agents remains in its infancy, and specificity and tolerability of epigenetic-directed drugs remain challenging. Immune-based therapies represent another area of significant potential, with promising results observed with cellular therapies and bispecific antibodies. It is likely that various immunotherapies will represent the backbone of future combinations and progress, as shown by multiple studies currently in development. While it remains uncertain whether harnessing immune cells to efficiently eliminate FL cells will let us envision the cure of this disease, the remarkable survival improvements achieved in the last 20 years with antiCD20 monoclonal antibodies allow us to continue to be optimistic. Disclosures In the last 12 months GS has received financial compensations for participating in advisory boards or consulting from : Abbvie, Bayer, Beigene, BMS/Celgene, Epizyme, Genentech/Roche, Genmab, Incyte, Janssen, Kite/Gilead, Loxo, Milteniy, Morphosys, Novartis, Rapt, Regeneron and Takeda. Contributions Both authors collected and analyzed data, wrote and approved the manuscript Funding This research was funded in part through the NIH/NCI Cancer Center support grant P30 CA008748.

4. Federico M, Barrigón MDC, Marcheselli L, et al. Rituximab and the risk of transformation of follicular lymphoma: a retrospective pooled analysis. Lancet Haematol. 2018;5(8):e359-367. 5. Smith S. Transformed lymphoma: what should I do now? Hematol Am Soc Hematol Educ Progr. 2020;2020(1):306-311. 6. Salles G. How do I sequence therapy for follicular lymphoma? Hematology Am Soc Hematol Educ Program. 2020;20(1):287-294. 7. Mozas P, Nadeu F, Rivas-Delgado A, et al. Patterns of change in treatment, response, and outcome in patients with follicular lymphoma over the last four decades: a singlecenter experience. Blood Cancer J. 2020;10 (3):31. 8. Pettengell R, Donatti C, Hoskin P, et al. The

impact of follicular lymphoma on healthrelated quality of life. Ann Oncol. 2008;19 (3):570-576. 9. Kridel R, Chan FC, Mottok A, et al. Histological transformation and progression in follicular lymphoma: a clonal evolution study. PLoS Med. 2016;13(12):e1002197. 10. Pasqualucci L, Khiabanian H, Fangazio M, et al. Genetics of follicular lymphoma transformation. Cell Rep. 2014;6(1):130. 11. Montoto S, López-Guillermo A, Altés A, et al. Predictive value of Follicular Lymphoma International Prognostic Index (FLIPI) in patients with follicular lymphoma at first progression. Ann Oncol. 2004;15(10):14841489. 12. Casulo C, Byrtek M, Dawson KL, et al. Early relapse of follicular lymphoma after ritux-

D. Qualls and G. Salles imab plus cyclophosphamide, doxorubicin, vincristine, and prednisone defines patients at high risk for death: an analysis from the National LymphoCare Study. J Clin Oncol. 2015;33(23):2516-2522. 13. Maurer MJ, Bachy E, Ghesquières H, et al. Early event status informs subsequent outcome in newly diagnosed follicular lymphoma. Am J Hematol. 2016;91(11):10961101. 14. Moccia AA, Schär S, Hayoz S, et al. Prognostic value of POD24 validation in follicular lymphoma patients initially treated with chemotherapy-free regimens in a pooled analysis of three randomized trials of the Swiss Group for Clinical Cancer Research (SAKK). Br J Haematol. 2021;192 (6):1031-1034. 15. Binkley MS, Brady JL, Hajj C, et al. Salvage treatment and survival for relapsed follicular lymphoma following primary radiotherapy: a collaborative study on behalf of ILROG. Int J Radiat Oncol Biol Phys. 2019;104 (3):522. 16. Casulo C, Barr PM. How I treat early-relapsing follicular lymphoma. Blood. 2019;133 (14):1540-1547. 17. Wagner-Johnston ND, Link BK, Byrtek M, et al. Outcomes of transformed follicular lymphoma in the modern era: a report from the National LymphoCare Study (NLCS). Blood. 2015;126(7):851-857. 18. Rummel M, Kaiser U, Balser C, et al. Bendamustine plus rituximab versus fludarabine plus rituximab for patients with relapsed indolent and mantle-cell lymphomas: a multicentre, randomised, openlabel, non-inferiority phase 3 trial. Lancet Oncol. 2016;17(1):57-66. 19. Sehn LH, Chua N, Mayer J, et al. Obinutuzumab plus bendamustine versus bendamustine monotherapy in patients with rituximab-refractory indolent nonHodgkin lymphoma (GADOLIN): a randomised, controlled, open-label, multicentre, phase 3 trial. Lancet Oncol. 2016;17(8):1081-1093. 20. Cheson BD, Chua N, Mayer J, et al. Overall survival benefit in patients with rituximabrefractory indolent non-hodgkin lymphoma who received obinutuzumab plus bendamustine induction and obinutuzumab maintenance in the GADOLIN study. J Clin Oncol. 2018;36(22):2259-2266. 21. Hiddemann W, Barbui AM, Canales MA, et al. Immunochemotherapy with obinutuzumab or rituximab for previously untreated follicular lymphoma in the GALLIUM study: Influence of chemotherapy on efficacy and safety. J Clin Oncol. 201836(23):2395-2404. 22. Link BK, Day B-M, Zhou X, et al. Secondline and subsequent therapy and outcomes for follicular lymphoma in the United States: data from the observational National LymphoCare Study. Br J Haematol. 2019;184(4):660-663. 23. Batlevi CL, Sha F, Alperovich A, et al. Follicular lymphoma in the modern era: survival, treatment outcomes, and identification of high-risk subgroups. Blood Cancer J. 2020;10(7):74. 24. Weide R, Feiten S, Friesenhahn V, et al. Retreatment with bendamustine-containing regimens in patients with relapsed/refractory chronic lymphocytic leukemia and indolent B-cell lymphomas achieves high response rates and some long lasting remissions. Leuk Lymphoma. 2013;54(8):16401646. 25. Morschhauser F, Fowler NH, Feugier P, et al.

Rituximab plus lenalidomide in advanced untreated follicular lymphoma. N Engl J Med. 2018;379(10):934-947. 26. Strati P, Samaniego F, Gallardo M, Noorani M. Treatment and outcome of patients with follicular lymphoma relapsed or progressed after frontline lenalidomide and rituximab. Blood. 2020;136(Suppl 1):31-32. 27. Harrop S, Polliack A, Tam CS. Chronic lymphoproliferative disorders and secondary cancers in the era of purine analogues and beyond. Leuk Lymphoma. 2020;62(4):771778. 28. Ghione P, Cavallo F, Visco C, et al. A retrospective study of R-DHAP/Ox for early progressing follicular lymphoma. Br J Haematol. 2018;183(5):828-831. 29. Martinelli G, Montoro J, Vanazzi A, et al. Chlorambucil–rituximab as first-line therapy in patients affected by follicular nonHodgkin’s lymphoma: a retrospective single-centre study. Hematol Oncol. 2015;33 (4):129-135. 30. Peterson BA, Petroni GR, Frizzera G, et al. Prolonged single-agent versus combination chemotherapy in indolent follicular lymphomas: a study of the Cancer and Leukemia Group B. J Clin Oncol. 2016;21 (1):5-15. 31. Forstpointner R, Unterhalt M, Dreyling M, et al. Maintenance therapy with rituximab leads to a significant prolongation of response duration after salvage therapy with a combination of rituximab, fludarabine, cyclophosphamide, and mitoxantrone (RFCM) in patients with recurring and refractory follicular and mantle cell lymphomas: results of a prospective randomized study of the German Low Grade Lymphoma Study Group (GLSG). Blood. 2006;108(13):40034008. 32. Van Oers MHJ, Van Glabbeke M, Giurgea L, et al. Rituximab maintenance treatment of relapsed/resistant follicular non-Hodgkin’s lymphoma: long-term outcome of the EORTC 20981 phase III randomized intergroup study. J Clin Oncol. 2010;28(17):28532858. 33. Vidal L, Gafter-Gvili A, Salles G, et al. Rituximab maintenance improves overall survival of patients with follicular lymphoma—individual patient data meta-analysis. Eur J Cancer. 2017;76:216-225. 34. Schouten HC, Qian W, Kvaloy S, et al. Highdose therapy improves progression-free survival and survival in relapsed follicular nonHodgkin’s lymphoma: results from the randomized European CUP trial. J Clin Oncol. 2003;21(21):3918-3927. 35. le Gouill S, de Guibert S, Planche L, et al. Impact of the use of autologous stem cell transplantation at first relapse both in naïve and previously rituximab exposed follicular lymphoma patients treated in the GELA/GOELAMS FL2000 study. Haematologica. 2011;96(8):1128-1135. 36. Jurinovic V, Metzner B, Pfreundschuh M, et al. Autologous stem cell transplantation for patients with early progression of follicular lymphoma: a follow-up study of 2 randomized trials from the German Low Grade Lymphoma Study Group. Biol Blood Marrow Transplant. 2018;24(6):1172-1179. 37. Casulo C, Friedberg JW, Ahn KW, et al. Autologous transplantation in follicular lymphoma with early therapy failure: a National LymphoCare Study and Center for International Blood and Marrow Transplant Research analysis. Biol Blood Marrow Transplant. 2018;24(6):1163-1171. 38. Pettengell R, Uddin R, Boumendil A, et al.

Durable benefit of rituximab maintenance post-autograft in patients with relapsed follicular lymphoma: 12-year follow-up of the EBMT Lymphoma Working Party Lym1 trial. Bone Marrow Transplant. 2021;56(6): 1413-1421. 39. Sarkozy C, Trneny M, Xerri L, et al. Risk factors and outcomes for patients with follicular lymphoma who had histologic transformation after response to first-line immunochemotherapy in the PRIMA trial. J Clin Oncol. 2016;34(22):2575-2582. 40. El-Najjar I, Boumendil A, Luan JJ, et al. The impact of total body irradiation on the outcome of patients with follicular lymphoma treated with autologous stem-cell transplantation in the modern era: a retrospective study of the EBMT Lymphoma Working Party. Ann Oncol. 2014;25(11):2224-2229. 41. Hoskin P, Popova B, Schofield O, et al. 4 Gy versus 24 Gy radiotherapy for follicular and marginal zone lymphoma (FoRT): long-term follow-up of a multicentre, randomised, phase 3, non-inferiority trial. Lancet Oncol. 2021;22(3):332-340. 42. Illidge T, Morschhauser F. Radioimmunotherapy in follicular lymphoma. Best Pract Res Clin Haematol. 2011;24(2):279-293. 43. Kolstad A, Illidge T, Bolstad N, et al. Phase 1/2a study of 177Lu-lilotomab satetraxetan in relapsed/refractory indolent non-Hodgkin lymphoma. Blood Adv. 2020;4(17):4091. 44. Phillips TJ, Michot JM, Ribrag V. Can nextgeneration PI3K inhibitors unlock the full potential of the class in patients with B-cell lymphoma? Clin Lymphma Myeloma Leuk. 2021;21(1):8-20. 45. Salles G, Schuster SJ, De Vos S, et al. Efficacy and safety of idelalisib in patients with relapsed, rituximab- and alkylating agentrefractory follicular lymphoma: a subgroup analysis of a phase 2 study. Haematologica. 2017;102(4):e156-e159. 46. Gopal AK, Kahl BS, de Vos S, et al. PI3Kδ inhibition by idelalisib in patients with relapsed indolent lymphoma. N Engl J Med. 2014;370(11):1008-1018. 47. Flinn IW, Miller CB, Ardeshna KM, et al. DYNAMO: a phase II study of duvelisib (IPI-145) in patients with refractory indolent non-Hodgkin lymphoma. J Clin Oncol. 2019;37(11):912-922. 48. Fowler NH, Samaniego F, Jurczak W, et al. Umbralisib, a dual PI3Kδ/CK1ε inhibitor in patients with relapsed or refractory indolent lymphoma. J Clin Oncol. 2021;39(15):16091618. 49. Dreyling M, Santoro A, Mollica L, et al. Long-term safety and efficacy of the PI3K inhibitor copanlisib in patients with relapsed or refractory indolent lymphoma: 2-year follow-up of the CHRONOS-1 study. Am J Hematol. 2020;95(4):362-371. 50. Gopal AK, Kahl BS, Flowers CR, et al. Idelalisib is effective in patients with highrisk follicular lymphoma and early relapse after initial chemoimmunotherapy. Blood. 2017;129(22):3037-3039. 51. Lynch R, Paneesha S, Avigdor A, et al. Phase 2 study evaluating the efficacy and safety of parsaclisib in patients with relapsed or refractory follicular lymphoma (CITADEL203). Blood. 2020;136(Suppl 1):36-37. 52. Zelenetz AD, Jagadeesh D, Reddy NM, et al. Results of the PI3Kδ inhibitor ME-401 alone or with rituximab in relapsed/refractory (R/R) follicular lymphoma (FL). J Clin Oncol. 2019;37(15_suppl):7512. 53. Matasar MJ, Capra M, Özcan M, et al. Copanlisib plus rituximab versus placebo

haematologica | 2022; 107(1)

Recent developments for patients with R/R FL

plus rituximab in patients with relapsed indolent non-Hodgkin lymphoma (CHRONOS-3): a double-blind, randomised, placebo-controlled, phase 3 trial. Lancet Oncol. 2021;22(5):678-689. 54. Gopal AK, Schuster SJ, Fowler NH, et al. Ibrutinib as Treatment for patients with relapsed/refractory follicular lymphoma: results from the open-label, multicenter, phase II DAWN study. J Clin Oncol. 2018;36(23):2405-2412. 55. Hu N, Wang F, Sun T, et al. Follicular lymphoma–associated BTK mutations are Inactivating resulting in augmented AKT activation. Clin Cancer Res. 2021;27(8):2301-2313. 56. Fowler NH, Coleman M, Stevens DA, et al. Acalabrutinib alone or in combination with rituximab (R) in follicular lymphoma (FL). J Clin Oncol. 2018;36(15_suppl):7549-7549. 57. Tam CS, Quach H, Nicol A, et al. Zanubrutinib (BGB-3111) plus obinutuzumab in patients with chronic lymphocytic leukemia and follicular lymphoma. Blood Adv. 2020;4(19):4802-4811. 58. Bennani NN, LaPlant BR, Ansell SM, et al. Efficacy of the oral mTORC1 inhibitor everolimus in relapsed or refractory indolent lymphoma. Am J Hematol. 2017;92(5):448453. 59. Smith SM, Besien K van, Karrison T, et al. Temsirolimus has activity in non–mantle cell non-Hodgkin’s lymphoma subtypes: the University of Chicago phase II consortium. J Clin Oncol. 2010;28(31):4740-4746. 60. Davids MS, Roberts AW, Kenkre VP, et al. Long-term follow-up of patients with relapsed or refractory non–Hodgkin lymphoma treated with venetoclax in a phase I, first-in-human study. Clin Cancer Res. 2021;27(17):4690-4695. 61. Zinzani PL, Flinn IW, Yuen SLS, et al. Venetoclax-rituximab with or without bendamustine vs bendamustine-rituximab in relapsed/refractory follicular lymphoma. Blood. 2020;136(23):2628. 62. Blombery P, Birkinshaw RW, Nguyen T, et al. Characterization of a novel venetoclax resistance mutation (BCL2 Phe104Ile) observed in follicular lymphoma. Br J Haematol. 2019 186(6):e188-191. 63. Morschhauser F, Tilly H, Chaidos A, et al. Tazemetostat for patients with relapsed or refractory follicular lymphoma: an openlabel, single-arm, multicentre, phase 2 trial. Lancet Oncol. 2020;21(11):1433-1442. 64. Ribrag V, Kim WS, Bouabdallah R, et al. Safety and efficacy of abexinostat, a pan-histone deacetylase inhibitor, in non-Hodgkin lymphoma and chronic lymphocytic leukemia: results of a phase II study. Haematologica. 2017;102(5):903-909. 65. Kirschbaum M, Frankel P, Popplewell L, et al. Phase II study of vorinostat for treatment of relapsed or refractory indolent nonHodgkin’s lymphoma and mantle cell lymphoma. J Clin Oncol. 2011;29(9):1198-1203. 66. Ogura M, Ando K, Suzuki T, et al. A multicentre phase II study of vorinostat in patients with relapsed or refractory indolent B-cell non-Hodgkin lymphoma and mantle cell lymphoma. Br J Haematol. 2014;165(6):768-776. 67. Evens AM, Balasubramanian S, Vose JM, et al. A phase I/II multicenter, open-label study of the oral histone deacetylase inhibitor abexinostat in relapsed/refractory lymphoma. Clin Cancer Res. 2016;22(5):10591066. 68. Assouline SE, Nielsen TH, Yu S, et al. Phase 2 study of panobinostat with or without rit-

haematologica | 2022; 107(1)

uximab in relapsed diffuse large B-cell lymphoma. Blood. 2016;128(2):185-194. 69. Heward JA, Koniali L, D’Avola A, et al. KDM5 inhibition offers a novel therapeutic strategy for the treatment of KMT2D mutant lymphomas. Blood. 2021;138(5): 370-381. 70. Kahl BS, Hong F, Williams ME, et al. Rituximab extended schedule or re-treatment trial for low-tumor burden follicular lymphoma: Eastern Cooperative Oncology Group protocol E4402. J Clin Oncol. 2014;32(28):3096-3102. 71. Sehn LH, Goy A, Offner FC, et al. Randomized phase II trial comparing obinutuzumab (GA101) with rituximab in patients with relapsed CD20+ indolent Bcell non-Hodgkin lymphoma: final analysis of the GAUSS study. J Clin Oncol. 2015;33(30):3467-3474. 72. Leonard JP, Trneny M, Izutsu K, et al. AUGMENT: a phase III study of lenalidomide plus rituximab versus placebo plus rituximab in relapsed or refractory indolent lymphoma. J Clin Oncol. 2019;37(14):11881199. 73. Davis TA, Grillo-López AJ, White CA, et al. Rituximab anti-CD20 monoclonal antibody therapy in non-Hodgkin’s lymphoma: safety and efficacy of re-treatment. J Clin Oncol. 2000;18(17):3135-3143. 74. Johnston A, Bouafia-Sauvy F, BroussaisGuillaumot F, et al. Retreatment with rituximab in 178 patients with relapsed and refractory B-cell lymphomas: a single institution case control study. Leuk Lymphoma. 2010;51(3):399-405. 75. Maloney DG, Ogura M, Fukuhara N, et al. A phase 3 randomized study (HOMER) of ofatumumab vs rituximab in iNHL relapsed after rituximab-containing therapy. Blood Adv. 2020;4(16):3886-3893. 76. Ramsay AG, Clear AJ, Kelly G, et al. Follicular lymphoma cells induce T-cell immunologic synapse dysfunction that can be repaired with lenalidomide: Implications for the tumor microenvironment and immunotherapy. Blood. 2009;114(21):47134720. 77. Chiu H, Trisal P, Bjorklund C, et al. Combination lenalidomide-rituximab immunotherapy activates anti-tumour immunity and induces tumour cell death by complementary mechanisms of action in follicular lymphoma. Br J Haematol. 2019;185(2):240-253. 78. Vo DN, Alexia C, Allende-Vega N, et al. NK cell activation and recovery of NK cell subsets in lymphoma patients after obinutuzumab and lenalidomide treatment. Oncoimmunology. 2018;7(4):e1409322. 79. Morschhauser F, Le Gouill S, Feugier P, et al. Obinutuzumab combined with lenalidomide for relapsed or refractory follicular Bcell lymphoma (GALEN): a multicentre, single-arm, phase 2 study. Lancet Haematol. 2019;6(8):e429-437. 80. Ghosh N, Salles GA, Lossos IS, et al. Obinutuzumab (G)-atezolizumab (atezo)lenalidomide (len) for the treatment of relapsed/refractory (R/R) follicular lymphoma (FL): final analysis of a phase Ib/II trial. J Clin Oncol. 2021;39(15_suppl):7511. 81. Jurczak W, Zinzani PL, Gaidano G, et al. Phase IIa study of the CD19 antibody MOR208 in patients with relapsed or refractory B-cell non-Hodgkin’s lymphoma. Ann Oncol. 2018;29(5):1266-1272. 82. Jacobson CA, Chavez JC, Sehgal AR, et al. Interim analysis of ZUMA-5: a phase II study of axicabtagene ciloleucel (axi-cel) in

patients (pts) with relapsed/refractory indolent non-Hodgkin lymphoma (R/R iNHL). J Clin Oncol. 2020;38(15_suppl):8008. 83. Jacobson CA, Chavez JC, Sehgal A, et al. Outcomes in ZUMA-5 with axicabtagene ciloleucel (axi-cel) in patients (pts) with relapsed/refractory (R/R) indolent nonHodgkin lymphoma (iNHL) who had the high-risk feature of progression within 24 months from initiation of first anti-CD20– containing chemoimmunotherapy (POD24). J Clin Oncol. 2021;39(15_suppl):7515. 84. Ghione P, Patel A, Bobillo S, Deighton K, Jacobson C, Nahas M. A comparison of clinical outcomes from ZUMA-5 (axicabtagene ciloleucel) and the international SCHOLAR5 external control cohort in relapsed/refractory follicular lymphoma (R/R FL). European Hematology Association Congress. 2021. 85. Fowler NH, Dickinson M, Dreyling M, et al. Efficacy and safety of tisagenlecleucel in adult patients with relapsed/refractory follicular lymphoma: interim analysis of the phase 2 Elara trial. Blood. 2020;136(Suppl 1):1-3. 86. Qualls D, Salles GA. Optimizing CAR T cell therapy in lymphoma. Hematol Oncol. 2021;39 (Suppl 1):104-112. 87. Sehn LH, Kamdar M, Herrera AF, et al. Randomized phase 2 trial of polatuzumab vedotin (pola) with bendamustine and rituximab (BR) in relapsed/refractory (r/r) FL and DLBCL. J Clin Oncol. 2018;36(15_suppl): 7507. 88. Caimi PF, Ai W, Alderuccio JP, et al. Loncastuximab tesirine in relapsed or refractory diffuse large B-cell lymphoma (LOTIS2): a multicentre, open-label, single-arm, phase 2 trial. Lancet Oncol. 2021;22(6):790800. 89. Morschhauser F, Flinn IW, Advani R, et al. Polatuzumab vedotin or pinatuzumab vedotin plus rituximab in patients with relapsed or refractory non-Hodgkin lymphoma: final results from a phase 2 randomised study (ROMULUS). Lancet Haematol. 2019;6(5):e254-265. 90. Diefenbach C, Kahl BS, Banerjee L, et al. Polatuzumab vedotin plus obinutuzumab and lenalidomide in patients with relapsed/refractory follicular lymphoma: primary analysis of the full efficacy population in a phase Ib/II trial. Blood. 2019;134 (Suppl_1):126. 91. Hamadani M, Radford J, Carlo-Stella C, et al. Final results of a phase 1 study of loncastuximab tesirine in relapsed/refractory B-cell non-Hodgkin lymphoma. Blood. 2021;137 (19):2634-2645. 92. Assouline SE, Kim WS, Sehn LH, et al. Mosunetuzumab shows promising efficacy in patients with multiply relapsed follicular lymphoma: updated clinical experience from a phase I dose-escalation trial. Blood. 2020;136(Suppl 1):42-44. 93. Hutchings M, Morschhauser F, Iacoboni G, et al. Glofitamab, a novel, bivalent CD20targeting T-cell–engaging bispecific antibody, induces durable complete remissions in relapsed or refractory B-cell lymphoma: a phase I trial. J Clin Oncol. 2021;39(18):19591970. 94. Hutchings M, Mous R, Clausen MR, et al. Dose escalation of subcutaneous epcoritamab in patients with relapsed or refractory Bcell non-Hodgkin lymphoma: an open-label, phase 1/2 study. Lancet. 2021;6736(21):1-13. 95. Bannerji R, Allan JN, Arnason JE, et al. Odronextamab (REGN1979), a human CD20 x CD3 bispecific antibody, induces durable, complete responses in patients with

D. Qualls and G. Salles highly refractory B-cell non-Hodgkin lymphoma, including patients refractory to CAR T therapy. Blood. 2020;136(Suppl 1):42-43. 96. Feng M, Jiang W, Kim BYS, Zhang CC, Fu YX, Weissman IL. Phagocytosis checkpoints as new targets for cancer immunotherapy. Nat Rev Cancer. 2019;19(10):568-586. 97. Advani R, Bartlett NL, Smith SM, et al. The first-in-class anti-CD47 antibody HU5F9-G4 + rituximab induces durable responses in relapsed/refractory DLBCL and indolent lymphoma: interim phase 1B/2 results. Hematol Oncol. 2019;37(S2):89-90. 98. Kim TM, Lakhani N, Gainor J, et al. ALX148, a CD47 blocker, in combination with rituximab in patients with nonHodgkin lymphoma. Blood. 2020;136(Suppl 1):13-14. 99. Patel K, Ramchandren R, Maris M, et al. Investigational CD47-blocker TTI-622 shows single-agent activity in patients with advanced relapsed or refractory lymphoma: update from the ongoing first-in-human dose escalation study. Blood. 2020;136 (Suppl 1):46-47. 100. Dheilly E, Moine V, Broyer L, et al. Selective blockade of the ubiquitous checkpoint receptor CD47 is enabled by dual-targeting bispecific antibodies. Mol Ther. 2017;25(2): 523-533. 101. Nakamura K, Casey M, Oey H, et al. Targeting an adenosine-mediated “don’t eat me signal” augments anti-lymphoma immunity by anti-CD20 monoclonal antibody. Leukemia. 2020;34(10):2708-2721. 102. Liu X, Pu Y, Cron K, et al. CD47 blockade triggers T cell-mediated destruction of immunogenic tumors. Nat Med. 201521(10):

1209-1215. 103. Laurent C, Charmpi K, Gravelle P, et al. Several immune escape patterns in nonHodgkin’s lymphomas. Oncoimmunology. 2015;4(8):e1026530. 104. Armand P, Janssens A, Gritti G, et al. Efficacy and safety results from CheckMate 140, a phase 2 study of nivolumab for relapsed/refractory follicular lymphoma. Blood. 2021;137(5):637-645. 105. Ding W, Laplant B, Witzig TE, et al. PD-1 blockade with pembrolizumab in relapsed low grade non-Hodgkin lymphoma. Blood. 2017;130(Suppl 1):4055. 106. Salles G, Ghosh N, Lossos IS, et al. Atezolizumab in combination with obinutuzumab and lenalidomide demonstrates favorable activity and manageable toxicity in patients with relapsed/refractory follicular lymphoma (FL): an interim analysis of a phase Ib/II trial. Blood. 2018;132(Suppl 1):1603. 107. Palomba ML, Till BG, Park SI, et al. A phase IB study evaluating the safety and clinical activity of atezolizumab combined with obinutuzumab in patients with relapsed or refractory non-Hodgkin lymphome (NHL). Hematol Oncol. 2017;35(S2):137-138. 108. Nastoupil LJ, Westin JR, Fowler NH, et al. Response rates with pembrolizumab in combination with rituximab in patients with relapsed follicular lymphoma: interim results of an on open-label, phase II study. J Clin Oncol. 2017;35(15_suppl):7519. 109. Houot R, Goldstein MJ, Kohrt HE, et al. Therapeutic effect of CD137 immunomodulation in lymphoma and its enhancement by Treg depletion. Blood. 2009;114(16):

3431-3438. 110. Gopal AK, Levy R, Houot R, et al. First-inhuman study of utomilumab, a 41BB/CD137 agonist, in combination with rituximab in patients with follicular and other CD20+ non-Hodgkin lymphomas. Clin Cancer Res. 2020;26(11):2524-2534. 111. Segal NH, Logan TF, Hodi FS, et al. Results from an integrated safety analysis of urelumab, an agonist anti-CD137 monoclonal antibody. Clin Cancer Res. 2017;23(8):19291936. 112. Timmerman J, Herbaux C, Ribrag V, et al. Urelumab alone or in combination with rituximab in patients with relapsed or refractory B-cell lymphoma. Am J Hematol. 2020;95(5):510-520. 113. Xu Y, Szalai AJ, Zhou T, et al. FcgRs modulate cytotoxicity of anti-Fas antibodies: implications for agonistic antibody-based therapeutics. J Immunol. 2003;171(2):562568. 114. Herter S, Sam J, Ferrara Koller C, et al. RG6076 (CD19-4-1BBL): CD19-targeted 41BB ligand combination with glofitamab as an off-the-shelf, enhanced T-cell redirection therapy for B-cell malignancies. Blood. 2020;136(Suppl 1):40. 115. Andorsky DJ, Coleman M, Yacoub A, et al. MAGNIFY phase IIIb interim analysis of induction R2 followed by maintenance in relapsed/refractory indolent NHL. J Clin Oncol. 2020;38(15_suppl):8046. 116. Dreyling M, Santoro A, Leppä S, et al. Efficacy and safety in high-risk relapsed or refractory indolent follicular lymphoma patients treated with copanlisib. Hematol Oncol. 2019;37(Suppl):387-389.

haematologica | 2022; 107(1)

REVIEW ARTICLE

Marginal zone lymphoma: present status and future perspectives

Ferrata Storti Foundation

Chan Y. Cheah,1,2 Emanuele Zucca,3 Davide Rossi4 and Thomas M. Habermann5 Department of Haematology, Sir Charles Gairdner Hospital, Nedlands, Western Australia, Australia; 2Medical School, University of Western Australia, Crawley, Western Australia, Australia; 3Oncology Institute of Southern Switzerland, University of Bern and International Extranodal Lymphoma Study Group, Director of Operation Office, Bern, Switzerland; 4 Laboratory of Experimental Hematology, Institute of Oncology Research, Bellinzona, Switzerland and 5Department of Internal Medicine, Division of Hematology, Mayo Clinic, Rochester, MN, USA 1

Haematologica 2022 Volume 107(1):35-43

Introduction Marginal zone lymphomas (MZL) are collectively the second most common indolent lymphoma comprising 7% of all non-Hodgkin lymphomas with 7,460 patients diagnosed in the USA in 2016.1-3 There are three distinct subtypes: extranodal MZL of mucosa-associated lymphoid tissue (MALT lymphoma), which accounts for 50–70% of cases, splenic MZL (20%) and nodal MZL (10%).4,5 The optimal treatment in many cases is not well defined because of the diversity of clinical presentation, incomplete understanding of the underlying disease biology and tendency to group MZL with follicular lymphoma in clinical trials. In this review, we provide an overview of the biology, epidemiology, clinical presentation, current management strategies and emerging data for novel agents in the management of relapsed/refractory disease. The epidemiology and natural history of MZL remain poorly understood.6 A family history of lymphoma is a salient risk factor for MZL. Genetic and environmental risk factors for extranodal MZL include infectious agents and autoimmune disorders such as Sjögren syndrome, systemic lupus erythematosus and Hashimoto thyroiditis.3 Preliminary data suggest a significant risk factor for nodal MZL is being a metal worker (odds ratio 3.6), whereas significant risk factors for splenic MZL include asthma (odds ratio 2.3) and use of hair dye (odds ratio 6.5).7

Extranodal marginal zone lymphoma of mucosa-associated lymphoid tissue (MALT lymphoma)

Correspondence: MALT lymphoma is often caused by chronic antigenic stimulation by infectious pathogens or autoimmunity leading to inflammatory lymphoid populations and can arise in widely varied sites.5 Gastric MALT accounts for more than 30% of cases.8 Other common sites include the ocular adnexa, salivary glands, skin, conjunctiva, lungs, thyroid and breasts, with diverse site-specific etiologies.2 The strongest evidence for a specific etiological pathogen relates to Helicobacter pyloriinduced chronic gastritis implicated in around two-thirds of cases of gastric MALT lymphoma. Autoimmune diseases such as Sjögren syndrome and Hashimoto thyroiditis are associated with increased risk of MALT lymphoma of the salivary gland and thyroid, respectively. The clinical presentation of MALT lymphoma varies widely according to the site(s) of involvement. Typically, MALT lymphoma has an indolent behavior and favorable outcomes.1 In MALT lymphoma, particularly gastric, contact with foreign antigens and mucosal permeability are likely important.9 Although most MALT lymphomas are localized, around 20% are stage IV and extranodal dissemination is typical in this case. Other infections associated with MALT lymphoma include hepatitis C virus, Chlamydophila psittaci (previously Chlamydia) in the conjunctiva and ocular adnexa, Borrelia burgdorferi in the skin, and Achromobacter xylosoxidans in the lungs. Histological transformation of extranodal MZL is associated with an inferior prognosis and may be driven by TP53 mutations, loss of p16 protein, or rearrangements in MYC.10,11 Critical signals required to support growth of marginal zone B cells include BAFF, CD40, TLR, BCR, and NOTCH receptor signaling. Chronic antigenic stimulation through infection or autoimmunity can drive B-cell receptor stimulation, biased immunoglobulin heavy chain gene usage, as well as genetic abnormalities in signaling pathways

haematologica | 2022; 107(1)

CHAN CHEAH chan.cheah@health.wa.gov.au Received: September 10, 2021. Accepted: November 3, 2021. https://doi.org/10.3324/haematol.2021.278755

C.Y. Cheah et al.

that regulate the NFkB pathway including t(1;14), t(11;18), A20 inactivation and MYD88 mutations.12

Diagnosis and staging MALT lymphomas characteristically remain localized for prolonged periods although multi-focal single organ involvement and systemic dissemination can occur in up to 25% of cases (more likely with non-gastric sites).5,13 Patients with advanced stage disease have an inferior prognosis and require different therapeutic strategies from patients with localized disease.8 Thus, careful staging is required and the diagnostic work-up should be tailored according to the site involved and any possible underlying infectious or autoimmune causes.5 Bone marrow involvement is present in fewer than 10% of patients with initially localized MALT lymphoma and patients without cytopenias can possibly be spared this procedure as those with radiologically defined stage IE disease have excellent lymphoma-specific outcomes irrespective of whether bone marrow biopsy is performed or not.14.15 The MALT International Prognostic Index (MALT-IPI) identified three factors (advanced stage disease, age ≥70 years, and elevated lactate dehydrogenase) that may be useful for prognostication.16

Current treatment approaches Localized disease Gastric MALT lymphoma In patients positive for H. pylori infection, standard eradication therapy with a proton pump inhibitor plus dual or triple antibiotics should be instituted. H. pylori eradication alone causes regression of gastric MALT lymphoma in 75% of cases.5 Re-testing at 2 months with a breath test can be considered - following cessation of proton pump inhibitors for at least 1 month - to ensure eradication before re-assessing the lymphoma status endoscopically 3 months after eradication.17,18 Waiting for 3 months for repeat endoscopy is important, as earlier evaluation may not reflect the eventual disease response. Patients with tumors carrying the t(11;18) translocation have a lower response rate to H. pylori eradication and alternative approaches (see below) should be considered for these patients.19 For patients with localized disease who are H. pylorinegative, empiric eradication therapy may still be beneficial in a significant proportion of patients.20 Similarly, clarithromycin therapy has resulted in meaningful response rates in some patients with gastric MALT lymphoma.21 In cases in which eradication therapy has failed, involved site radiotherapy is a reasonable approach with favorable outcomes using moderate doses (24-30 Gy over 3-4 weeks).17,22 One study which included patients with localized gastric or non-gastric MALT lymphoma reported 10-year overall and recurrence-free survival rates of 87% and 76%, respectively, with cause-specific survival of 98%.23 Other treatment options include rituximab monotherapy,24 and chemo-immunotherapy such as rituximab plus chlorambucil25 or rituximab plus cyclophosphamide, vincristine and prednisolone (R-CVP).18 Gastrectomy results in significant morbidity and is no longer recommended. Non-gastric MALT lymphoma Patients with localized disease in other sites associated with a postulated causative pathogen should be consid36

ered for eradication therapy, although the etiological relationship and outcomes following eradication are less well established. Some investigators have found ocular adnexal MALT to be associated with C. psittaci with considerable geographic variability.26 Doxycycline or clarithromycin has resulted in response rates of 45-65%.27 Furthermore, disease regression using antibiotics has been reported in C. psittaci-negative cases.28 Thus testing and an empiric trial of eradication can be considered. Data regarding response rates to antibiotics in the other subtypes are scant, and no firm conclusions can be drawn. In contrast to other lymphomas, radiation therapy has a significant role in extranodal MZL.29 The phase II Trans-Tasman Radiation Oncology Group/Australasian Leukemia and Lymphoma Group 05.02 trial established that involved field radiotherapy is a reasonable treatment for localized non-gastric MALT lymphoma, resulting in 5year progression free survival (PFS) and overall survival (OS) of 79% and 95% respectively.30 Outcomes of dural MZL have been reported to be favorable following radiation therapy.31,32 Extranodal MALT of the thyroid, small bowel, colon, and rectum have been managed with observation, surgical resection, radiation therapy, and rituximab. MALT lymphoma of the salivary glands has an excellent prognosis irrespective of the primary therapy.33 According to National Comprehensive Cancer Network guidelines, surgery may be considered for lymphomas in certain sites and some selected asymptomatic patients can also be observed.34

Advanced stage disease Advanced stage MZL of MALT type is incurable and the usually indolent biology allows for a ‘watch and wait’ approach in many patients. When treatment is required, systemic chemo-immunotherapy has been used successfully. The addition of rituximab to chlorambucil improved outcomes compared to either agent alone.25 RCVP followed by rituximab maintenance has been shown to be well tolerated and effective.35 Bendamustine and rituximab was safe and effective in a phase II trial of 60 patients with a median follow-up of 43 months.36 Eventfree survival was 88% at 4 years. A USA-Italian observational series (n=136) confirmed these observations, with estimated 5-year PFS and OS of 72.3% and 85.6%, respectively.37 Similar results were evident from German prospective registry data.38

Nodal marginal zone lymphoma Nodal MZL is the least common of all the subtypes of MZL, accounting for approximately 10% of MZL and <2% of all non-Hodgkin lymphomas.2,39 The median age at presentation is 60 years and both genders are equally affected.40 The understanding of nodal MZL has been hampered by its rarity, with therapeutic strategies largely based on data from follicular or small lymphocytic lymphoma. In common with these disorders, the disease generally behaves in an indolent fashion and is often disseminated at presentation. Histological transformation is reported in 3-15% of patients with nodal MZL and is often associated with a poor outcome.41 While there is an association with hepatitis C infection,42 a history of autoimmunity is less common than with other forms of MZL.43 haematologica | 2022; 107(1)

Marginal zone lymphoma

Diagnosis and staging Peripheral lymphadenopathy involving the head and neck is common at presentation, with up to one third of patients having bulky tumors (>5 cm) and about half having stage III/IV disease.43 Approximately 10% of patients will present with an IgM paraprotein,39 which can result in the diagnosis being confused with Waldenström macroglobulinemia. The absence of an MYD88 L265P mutation (a feature of Waldenström macroglobulinemia) supports the diagnosis of nodal MZL although this mutation may also be observed in less than 10% of cases of nodal MZL.38 PTPRD mutations are observed in 20% of patients with nodal MZL and the finding appears specific to this entity.44 Nodal MZL demonstrates similar cytological, immunophenotypic and genetic features to those of both splenic and extranodal MZL which may result in diagnostic difficulty, particularly in cases with involvement of the spleen or extranodal sites.39 Validated prognostic scoring systems are lacking in nodal MZL, with conflicting data regarding the applicability of the Follicular Lymphoma International Prognostic Index (FLIPI).45,46 Increased age and advanced stage have been associated with an adverse prognosis.47

Current treatment approaches The standard therapy for nodal MZL is yet to be defined with many centers employing strategies used in follicular lymphoma. Patients with localized disease respond well to radiotherapy, and those with minimally symptomatic, low tumor burden, advanced stage disease are suitable for a strategy of watchful waiting.39 Reports of regression of MZL with eradication of hepatitis C infection support this strategy as an initial approach in hepatitis C virus-infected patients.48,49 Patients with disseminated disease and high tumor burden can be treated with chemo-immunotherapy.40

cyclophosphamide and rituximab53 and bendamustine and rituximab.51,54 Fludarabine regimens are not routinely utilized because of toxicities.

Splenic marginal zone lymphoma Splenic MZL makes up less than 2% of all lymphoid malignancies, and 20% of all MZL. It is usually indolent, with a median survival of 8-10 years, but can transform to diffuse large B-cell lymphoma in approximately 5-10% of cases.2,55 Approximately one third of patients have no symptoms, and a watch and wait approach has no adverse impact on overall survival.56 The subtypes and the biology and function of the splenic marginal zone B cell remain poorly understood. Splenic marginal zone B cells bridge the gap between early innate immune responses and late adaptive immune responses. Marginal zone B cells are sustained in their local microenvironment by cytokine-secreting cells such as group 3 innate lymphoid cells, which produce copious amounts of the cytokine BAFF and induce IgM, IgG, and IgA production in marginal zone B cells. NOTCH pathway genes are mutated in splenic MZL and nodal MZL, in addition to other marginal zone differentiation-associated genes, in as many as 60% of patients.57 A common mutation in splenic MZL occurs in the KLF2 transcription factor, leading to activation of NFkB signaling with further hits to TRAF3, MAP3K14, and BIRC3.11 Distinguishing splenic MZL from other CD5and CD10-negative indolent B-cell lymphoproliferative disorders can be challenging, with a definitive diagnosis best achieved with spleen histology.55 However, in most patients, the diagnosis can be suggested by the characteristic morphology of peripheral blood lymphocytes with bipolar cytoplasmic villous projections and a round nucleus (in contrast to hairy cell leukemia in which cells have circumferential projections and an ovoid nucleus).

Chemo-immunotherapy Despite the lack of prospective studies, chemoimmunotherapy with rituximab is generally considered standard treatment for patients with symptomatic advanced stage disease. Numerous regimens have been explored including R-CVP,50 rituximab, cyclophosphamide, doxorubicin, vincristine and prednisone (RCHOP),51 fludarabine and rituximab,52 fludarabine,

Diagnosis and staging Immunophenotyping of circulating or bone marrow lymphocytes demonstrates IgM +/- IgD, CD19, CD20, CD22 and BCL-2 expression. CD23, CD25, and CD103 and cyclin D1 negativity assist in excluding chronic lymphocytic leukemia, mantle cell and hairy cell leukemia (which also causes prominent splenomegaly). In common

Table 1. Selected phase II studies evaluating novel agents in patients with relapsed/refractory marginal zone lymphoma. Efficacy data from the publication with longest follow-up reported where available.

Class

Agent

First author

BTK inhibitor

ibrutinib zanubrutinib

Noy79, 80 Opat83

63 68

66 70

2 2

PI3K inhibitor

idelalisib copanlisib

Gopal84 Martin85 Dreyling89 Panayiotidis90 Fowler91 Phillips92 Jacobson93

15 23

NA 69

69 99 22

67 71 66

Anti-CD19 CAR T-cell

umbralisib parsaclisib axi-cel

N. of MZL Median cases age (years)

Median Extranodal / prior nodal / lines splenic (%)

ORR (%)

CR (%)

Median DOR (months)

Median PFS (months)

OS (months)

51 / 27 / 22 38 / 38 / 18

58 74

3 24

NA 17 / 65 / 17

43 70

5 13

15.7 15m PFS 68% 6.6 24.1

NE NA

NA 3

27.6 12m DOR 81% NA 17.4

NA 2y OS 83%

2 2 3

33 / 31 / 35 NA

49 54 85

16 6 60

NE 9.3 10.6

2y PFS 50.5% 13.8 11.8

NE NA NE

MZL: marginal zone lymphoma; ORR, objective response rate; CR, complete response; DOR, duration of response; PFS, progression free survival; OS, overall survival; m: months; y: years: NA: not available; NE: not evaluable; BTK, Bruton tyrosine kinase; PI3K, phosphatidylinositol-3-kinase; axi-cel: . axicabtagene ciloleucel; CAR: T-cell: chimeric antigen receptor T-cell therapy.

haematologica | 2022; 107(1)

C.Y. Cheah et al.

with nodal MZL, an IgM paraprotein may occur and MYD88 mutation testing can help distinguish splenic MZL from Waldenström macroglobulinemia.58 Hepatitis C infection should be treated if present as treatment can result in lymphoma regression.48 Splenic hilar lymphadenopathy occurs in 25% of cases of splenic MZL but peripheral lymphadenopathy is rare.59,60 Most patients present with splenomegaly, lymphocytosis and cytopenias.59 Autoimmune hemolytic anemia and other autoimmune phenomena can occur. Computed tomography (CT) scan is adequate for staging; positron emission tomography (PET) with CT can be reserved for patients in whom histological transformation is suspected.55

imab.64 There was a significant improvement in time to treatment failure and PFS in the maintenance therapy arm. Single-agent rituximab (with short or protracted administration) has become the preferred approach in most patients, so that splenectomy or chemoimmunotherapy is reserved for patients not responding to single-agent treatment.65 Recent single institution data showed that CD5 expression, although rare in MZL, was associated with a lower ORR following rituximab monotherapy but not bendamustine and rituximab, suggesting the latter combination might be preferred if systemic therapy is required.66

Current treatment approaches

Histological transformation

Asymptomatic patients can be managed with observation. Symptomatic splenomegaly, cytopenia, systemic symptoms or progressive nodal disease are indications for treatment.55,56 Frontline treatment options include splenectomy, rituximab monotherapy, and chemo-immunotherapy.55 While these three approaches have not been directly compared, rituximab monotherapy or chemoimmunotherapy are typically preferred.

Splenectomy Splenectomy was the mainstay of therapy before rituximab monotherapy was adopted, and its role in modern management is now often in the second-line setting or beyond. Nonetheless, splenectomy removes disease bulk, abdominal discomfort and improves cytopenias due to splenic sequestration which is more common than heavy marrow involvement.55 Splenectomy typically results in durable disease control and facilitates a definitive diagnosis of splenic MZL.55 Short-term perioperative complications may be reduced with a laparoscopic approach and prophylaxis against venous thromboembolism. The late risk of infections with encapsulated bacteria can be minimized with vaccinations at least 2 weeks before elective splenectomy and, potentially, prophylactic antibiotics.55

Chemo-immunotherapy Chemo-immunotherapy is appropriate for fit patients with disseminated disease, constitutional symptoms, and/or high-grade transformation.55 R-CVP and R-CHOP, both commonly used in follicular lymphoma, can be delivered to patients with splenic MZL; however, in a prospective phase II trial R-COMP (rituximab with cyclophosphamide, vincristine, non-pegylated liposomal doxorubicin and prednisone) resulted in an objective response rate (ORR) of 84% and a 6-year PFS of 54%.55 Bendamustine and rituximab resulted in durable responses in the BRIGHT61 and STiL51 studies, and the use of rituximab maintenance for 2 years following initial treatment with bendamustine and rituximab in both nodal and splenic MZL prolonged PFS (but not OS) relative to no maintenance (hazard ratio=0.35, 95% confidence interval: 0.17-0.76, P=0.008) in the STiL NHL7-2008 MAINTAIN trial.62

Single-agent rituximab Rituximab monotherapy resulted in an ORR of 92% and 10-year freedom from progression of 64% in a large retrospective series.63 The RESORT trial treated patients with 375 mg/m2 weekly for 4 weeks and then randomized patients to observation versus maintenance ritux38

Histological transformation to diffuse large B-cell lymphoma is associated with a poor outcome and inferior overall survival. Histological transformation occurs with an annual incidence of approximately 1% per year.11 Failure to achieve complete remission, elevated lactate dehydrogenase concentration, more than four nodal sites involved at diagnosis,11 involvement of multiple mucosal sites,66 CD5 expression65 and in splenic MZL, complex karyotype, are associated with greater risk of histological transformation.67

Mantle zone lymphoma assessment criteria and response evaluation Several classifications are used for treatment response assessment in MZL. CT-scan response assessment is not well defined in MZL, which is primarily an extranodal disease. Splenic MZL and gastric MZL, or MZL from other extranodal sites prove difficult to assess based on CT-scan criteria. Various response assessment criteria for these diseases, such as the Lugano,68 Matutes69 and GELA70 classifications need to be homogenized, to facilitate better comparison of the results between different clinical trials. In localized gastric MZL, serial endoscopy and gastric biopsy are recommended, and responses can take up to 12 months. The presence of residual microscopic lymphoma prior to that time should not prompt initiation of another treatment if the patient has improved clinically and macroscopically.8 There is a wide range of endpoints that are currently used in clinical trials and in routine practice.71-73 However, the endpoints currently validated by the Food and Drug Administration (FDA) for use in phase II clinical trials are the ORR and complete response rate, which may not fully capture patients’ outcomes in MZL. Surrogate endpoints in MZL may include PET-CT, minimal residual disease, and progression of disease within 2 years.74 The value of PET-CT and minimal residual disease criteria in assessing response to treatment in MZL requires further characterization. The value of PET-CT in routine evaluation of MALT lymphoma remains unestablished.75 The sensitivity of PET-CT is highly variable in the disease, ranging from 5080% in various studies.76 Furthermore, fluorodeoxyglucose avidity in MZL is strongly dependent on the histopathological subtype.77 Nevertheless, PET-CT is useful because CT-based staging is of limited utility in the evaluation of extranodal disease.75 PET-CT is useful in haematologica | 2022; 107(1)

Marginal zone lymphoma

staging MZL to confirm localized disease and ensure effective radiotherapy. However, gastric and ocular MALT possess low fluorodeoxyglucose avidity. In histologically transformed MZL, PET-CT is necessary to confirm transformation and PET-CT-based response criteria are also used in the Lugano classification to identify transformed MZL.78 Site-specific imaging is required to monitor response in MZL (for example, magnetic resonance imaging in ocular adnexal MALT).

Novel agents in mantle zone lymphoma Bruton tyrosine kinase inhibitors The covalent Bruton tyrosine kinase (BTK) inhibitor ibrutinib was approved for patients with relapsed/refractory MZL on the basis of the results of a phase II trial.79 Noy et al. treated 63 patients with a median age of 66 years and median of two prior lines of therapy. MALT lymphoma (51%) was the most common subtype of MZL, followed by nodal MZL (27%) and splenic MZL (22%). Among the 60 patients in whom efficacy could be evaluated, the ORR was 48% (complete responses, 3%) and consistent across disease subtypes. The median duration of response was not reached, while the median PFS

was 14.2 months. The safety profile of ibrutinib was consistent with that following its use in other settings, with anemia (14%), pneumonia (8%) and fatigue (6%) being the most common grade ≥3 treatment-emergent adverse events (TEAE). Bleeding events occurred in 59% of patients and were all grade 1-2 apart form one grade 5 cerebral hemorrhage that occurred in a patient therapeutically anticoagulated with dalteparin. Atrial fibrillation occurred in four patients (6%). The long-term follow-up of this study was recently published, and after a median follow-up of 33.1 months the ORR was 58%, the median duration of response was 27.6 months, the median PFS 15.7 months and the median OS had not been reached.80 The ORR for extranodal, nodal and splenic MZL was 63%, 47% and 62%, respectively. Mutations in KMT2D and CARD11 were associated with shorter duration of response. The B-cell lymphoma-2 (BCL2) inhibitor venetoclax showed activity in a phase I study of patients with relapsed/refractory lymphoma, including MZL.81 Although this agent has not been further explored as monotherapy in MZL, the combination of ibrutinib and venetoclax was investigated in a small phase II study.82 An interim analysis demonstrated an ORR of 84% (complete responses, 42%) at week 16.

Figure 1. Mechanism of action of selected novel agents for the treatment of marginal zone lymphoma. Bold type indicates approved agents, normal type indicates agents under investigation. IgH: immunoglobulin heavy chain; CD: cluster of differentiation; Syk: spleen tyrosine kinase; Btk: Bruton tyrosine kinase; PLCγ2: phospholipase Cγ2; PKCb: protein kinase Cb; PI3K: phosphatidylinositol-3-kinase; Akt: protein kinase B; bax: B-cell lymphoma-2 associated X protein; Bcl2: B-cell lymphoma protein-2; CAR: chimeric antigen receptor; PD-1, programmed cell death-1. Adapted from Cheah et al. J Clin Oncol 2016.

haematologica | 2022; 107(1)

C.Y. Cheah et al.

The selective BTK inhibitor zanubrutinib was examined in a single-arm phase II study by Opat et al.83 Eligible patients with MZL who had had one or more prior line of therapy were treated with 160 mg of zanubrutinib twice daily. Among the 68 patients, 38% had MALT lymphoma, 38% had nodal MZL, 18% had splenic MZL and 6% had an indeterminate subtype. Patients had received a median of two prior lines of therapy and the ORR was 68% (complete responses, 26%) with similar rates between MZL subtypes. The estimated 15 month PFS rate was 82.5%. The most common TEAE were diarrhea (22%), bruising (21%), and constipation (15%) with neutropenia (10%) being the most common grade ≥3 TEAE. Atrial fibrillation and hypertension each occurred in 3% of patients. On the basis of these data zanubrutinib was recently approved for patients with relapsed MZL.

Phosphatidylinositol-3-kinase inhibitors Three phosphatidylinositol-3-kinase (PI3K) inhibitors have been approved by the FDA for the treatment of relapsed/refractory MZL. The PI3Kd inhibitor idelalisib was the first-in-class agent, with significant activity in a range of indolent B-cell malignancies. In the phase II registration trial, Gopal et al. treated 125 patients with indolent lymphoma including 15 with MZL with 150 mg of idelalisib orally twice daily.84 Efficacy was encouraging, but the infectious and immune toxicities were noteworthy. Although results for the subset of MZL patients were not reported separately in that study; Martin et al. reported pooled data from 21 patients treated in the phase I and phase II trials: two of six patients (both with partial responses) in phase I and seven of 15 (one with a complete response) in phase II achieved a response for a cumulative ORR of 43%.85 The median PFS in phase II was 6.6 months and the toxicity profile was consistent with that observed in other histological subtypes. Although phase III trials in indolent lymphoma were commenced, the toxicity observed when combining idelalisib with chemo-immunotherapy and other novel agents was considerable and further clinical development was halted.86-88 Other PI3K inhibitors have been developed, including the intravenous pan-class I PI3K inhibitor copanlisib. In the phase II CHRONOS-1 study Dreyling et al. enrolled 142 patients with relapsed/refractory indolent B-cell lymphoma, of whom 23 had MZL.89 Patients received 60 mg of copanlisib intravenously on days 1, 8 and 15 of 28-day cycles until disease progression or unacceptable toxicity. At the primary analysis after four cycles of therapy, the ORR was 70%, resulting in breakthrough therapy designation for adults with MZL who had received two or more prior systemic therapies. Long-term follow-up of the 23 patients in this study with relapsed/refractory MZL was recently reported.90 These patients had a median age of 69 years and had received a median of three prior lines of therapy. The most common MZL subtype was nodal MZL (n=15) with four patients each having splenic and MALT lymphoma. The eventual ORR was 78%, with the rates for nodal, splenic and MALT lymphoma being 87%, 75% and 50%, respectively. Complete responses were observed in three (13%) patients, all with splenic MZL. The median duration of response was 17.4 months; the median PFS was 24.1 months and the median OS was not reached. The estimated 2-year OS rate was 83%. The most frequent TEAE 40

of any grade were fatigue (52%), diarrhea and hyperglycemia (each 48%), while the most common grade ≥3 TEAE were hyperglycemia, hypertension (each 39%), fatigue, diarrhea, neutropenia and pneumonia (each 26%). The increase in infectious and immune toxicity observed with idelalisib was not apparent. Umbralisib is a novel, oral, dual inhibitor of PI3Kd and casein kinase-1-e, with minimal PI3Kg inhibition, which is taken once daily. Fowler et al. performed a phase IIb registration trial of umbralisib 800 mg daily in 208 patients with relapsed/refractory indolent B-cell lymphoma, including 69 patients with MZL.91 These patients had a median age of 67 years, they had received a median of two prior lines of therapy and 21% were refractory to their previous line of therapy. The ORR was 49% (complete responses, 16%) and consistent across the MZL subtypes. After a median follow-up of 27 months, the median duration of response was not reached, with the estimated 2-year PFS rate being 50.5%. Among all patients, the most frequent TEAE of any grade were diarrhea (59%), nausea (39%) and fatigue (30%), with the grade ≥3 ones being neutropenia (11%) and diarrhea (10%). TEAE of interest included opportunistic infections (grade ≥3 3.4%), elevation of liver enzymes (all grades 20.2%; grade ≥3 6.7%) pneumonitis (all grades 1.4%; grade ≥3 1.0%) and non-infectious colitis (all grades 1.9%; grade ≥3 0.5%). Overall the agent was active and resulted in durable remissions with an acceptable safety profile and has received accelerated FDA approval for patients with relapsed MZL. Other PI3K inhibitors such as parsaclisib and zandelisib (NCT037685050) have also been explored in phase II studies.92 Although the focus of chimeric antigen receptor T-cell therapy (CAR-T) studies has mainly been on patients with aggressive histological subtypes such as relapsed/refractory diffuse large B-cell lymphoma, a few studies have looked at patients with indolent lymphomas. ZUMA-5 is an ongoing phase II study of axicabtagene ciloleucel in patients with relapsed/refractory indolent B-cell lymphoma.93 Jacobson et al. treated 146 patients, including 22 patients with MZL who had received two or more prior lines of therapy including a CD20 monoclonal antibody combined with an alkylating agent. The median age of the MZL patients was 66 years, the median number of prior lines of therapy was three and 52% had experienced a prior progression of disease within 24 months despite CD20 and alkylator-based therapy. The ORR for MZL patients was 85% (complete responses, 60%) and after a median follow-up of 12.1 months, the median PFS was 11.8 months and the estimated 12-month OS rate was 92.9%. Cytokine release syndrome occurred in all 22 patients (100%), with two (9%) experiencing grade ≥3 events and 15 (68%) requiring tocilizumab. Typically for axicabtagene ciloleucel, the rate of neurological toxicity was substantial: 17 (77%) patients experienced a neurological event and in nine (41%) cases these were grade ≥3 events. Steroids were required in 14 (64%) of patients. Notably the rate of grade ≥3 neurological toxicity appeared higher in MZL than in the follicular lymphoma cohort of the same study (15%) and the median increases in analytes associated with axicabtagene ciloleucel toxicity were higher in patients with MZL than in those with follicular lymphoma. Although these results were disappointing, the number of patients with MZL was small and results are preliminary. In the haematologica | 2022; 107(1)

Marginal zone lymphoma

ongoing TRANSCEND FL study (NCT04245839) patients with relapsed/refractory follicular lymphoma and MZL are being treated with lisocabtagene maraleucel. This and other studies will help to define the role of anti-CD19 CAR T cells in this disease. Lenalidomide and rituximab is an active combination in patients with MZL. Leonard et al. included 63 patients with relapsed/refractory MZL (18% of the study population) in the AUGMENT study, in which patients with indolent B-cell lymphoma were randomized to either rituximab or rituximab and lenalidomide.94 Overall, the study indicated an improvement in PFS (hazard ratio=0.46) from the addition of lenalidomide although this did not reach statistical significance in the small subset of patients with MZL. A phase II trial in treatmentnaïve patients resulted in an ORR of 93% (complete responses, 70%).95 In a report of the long term follow-up of the MZL subset of this study, the median PFS was 59.8 months and the 5-year OS was 96%.96

Future directions The COUP-1 single-arm, phase II study (NCT03474744) is evaluating copanlisib and rituximab combination therapy in treatment-naïve and relapsed MZL patients ineligible for local therapy.22 The German Lymphoma Alliance has also planned the POLE-1 trial (NCT03474744), a single-arm, phase II German and Italian collaborative study designed to evaluate the clinical performance of pembrolizumab in treatment-naïve and relapsed-confirmed MZL patients with nodal, extranodal, or splenic disease who are ineligible for local therapy. The German Lymphoma Alliance’s OLYMP-1 trial (NCT03322865) is a single-arm phase II study that is designed to evaluate the clinical performance of obinutuzumab as a single agent in treatment-naïve MZL patients with nodal, extranodal, or splenic disease who are ineligible for local therapy. The IELSG38 study (NCT018085990), which investigated chlorambucil in combination with subcutaneous rituximab in patients with MALT lymphoma, has completed its enrollment phase, and follow-up and analysis is now ongoing. The IELSG48 randomized phase III study is planned to compare the clinical performance of rituximab single-agent therapy with that of rituximab combined with acalabrutinib in patients with splenic MZL in the first-line setting. The IELSG49 study (NCT04646395) is a run-in pilot study of tafasitamab (an anti-CD19 antibody) in combination with acalabrutinib in patients with relapsed or refractory MZL in whom previous systemic therapy has failed. The ongoing MALIBU-IELSG47 study

References 1. Reid R, Friedberg JW. Management of marginal zone lymphoma. Oncology (Williston Park). 2013;27(9):840, 842, 844. 2. Sriskandarajah P, Dearden CE. Epidemiology and environmental aspects of marginal zone lymphomas. Best Pract Res Clin Haematol. 2017;30(1-2):84-91. 3. Cerhan JR, Habermann TM. Epidemiology of marginal zone lymphoma. Ann Lymphoma. 2021;5:1.

haematologica | 2022; 107(1)

(NCT03697512) is evaluating ibrutinib plus rituximab combination therapy in untreated MZL, including extranodal MZL, splenic MZL, and nodal MZL. The primary endpoints of the MALIBU study are complete response at 12 months and progression-free survival at 5 years.

Conclusions MZL are a group of indolent B-cell lymphomas with considerable heterogeneity in terms of clinical presentation, biology, etiology and therapeutic approaches. Most patients with limited stage MALT lymphoma have an excellent prognosis with either antibiotics (H. pylori-associated gastric MALT) or radiotherapy (gastric and nongastric sites). In the absence of symptoms, splenic MZL can be observed, while symptomatic patients can be managed with rituximab monotherapy or eventually splenectomy or chemo-immunotherapy if needed. Patients with nodal MZL can be managed in a similar fashion to those with follicular lymphoma. Histological transformation is rare but associated with inferior outcomes and should be managed with anthracycline-based chemo-immunotherapy. Newer targeted agents including BTK inhibitors, PI3K inhibitors, and immunomodulatory drugs are active in patients with relapsed/refractory disease. At present, the role of CAR T-cell therapy in MZL is under investigation in several trials. Future studies will define more active novel combinations. Disclosures CYC has received fees for consulting and advisory services and honoraria from Roche, Janssen, MSD, Gilead, Ascentage Pharma, AstraZenecca, Lilly, TG therapeutics, Beigene, Novartis, and BMS; research funding from BMS, Roche, Abbvie; and travel expenses from Roche. TMH has participated in data monitoring committees for Seagen and Tess Therapeutics, has sat on scientific advisory boards for Eli Lilly & Co., Morphosys, Incyte, Biegene, and Loxo Oncology; he receives no personal compensation for these activities. His institution receives the compensation. DR receives no personal compensation. His institution has received honoraria from AbbVie, AstraZeneca, and Janssen; and research grants from AbbVie, AstraZeneca, and Janssen. EZ’s institution has received research support from AstraZeneca, Celgene, Incyte, Janssen, Merck, and Roche; honoraria for advisory boards services from Beigene, Celgene, Incyte Janssen, Merck, Roche, Celltion Healthcare, and Kyte (a Gilead Company); and travel grants from Abbvie and Roche. Contributions All authors revised and approved the manuscript.

4. Swerdlow SH, Campo E, Harris NL, et al. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. Revised Fourth Edition. IARC, 2017. 5. Zucca E, Bertoni F. The spectrum of MALT lymphoma at different sites: biological and therapeutic relevance. Blood. 2016;127(17):2082-2092. 6. Teras LR, DeSantis CE, Cerhan JR, Morton LM, Jemal A, Flowers CR. 2016 US lymphoid malignancy statistics by World Health Organization subtypes. CA

Cancer J Clin. 2016;66(6):443-459. 7. Bracci PM, Benavente Y, Turner JJ, et al. Medical history, lifestyle, family history, and occupational risk factors for marginal zone lymphoma: the InterLymph NonHodgkin Lymphoma Subtypes Project. J Natl Cancer Inst Monogr. 2014;2014 (48):52-65. 8. Zucca E, Arcaini L, Buske C, et al. Marginal zone lymphomas: ESMO clinical practice guidelines for diagnosis, treatment and follow-up. Ann Oncol.

C.Y. Cheah et al. 2020;31(1):17-29. 9. Isaacson P, Wright DH. Malignant lymphoma of mucosa-associated lymphoid tissue. A distinctive type of B-cell lymphoma. Cancer. 1983;52(8):1410-1416. 10. Maeshima AM, Taniguchi H, Toyoda K, et al. Clinicopathological features of histological transformation from extranodal marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue to diffuse large B-cell lymphoma: an analysis of 467 patients. Br J Haematol. 2016;174(6):923-931. 11. Alderuccio JP, Zhao W, Desai A, et al. Risk factors for transformation to highergrade lymphoma and its impact on survival in a large cohort of patients with marginal zone lymphoma from a single institution. J Clin Oncol. 2018;36(34): 3370-3380. 12. Du MQ. MALT lymphoma: a paradigm of NF-kappaB dysregulation. Semin Cancer Biol. 2016;39:49-60. 13. Raderer M, Wöhrer S, Streubel B, et al. Assessment of disease dissemination in gastric compared with extragastric mucosa-associated lymphoid tissue lymphoma using extensive staging: a singlecenter experience. J Clin Oncol. 2006;24(19):3136-3141. 14. Alderuccio JP, Isrow D, Reis IM, et al. Diagnostic bone marrow biopsy in patients with stage I EMZL treated with radiation therapy: needed or not? Blood. 2020;135(15):1299-1302. 15. Chae H, Cho H, Sa H-S, et al. The limited role of comprehensive staging work-up in ocular adnexal extranodal marginal zone lymphoma of mucosa-associated lymphoid tissue type (MALToma) with excellent prognosis. Br J Haematol. 2021;193 (4):848-851. 16. Thieblemont C, Cascione L, Conconi A, et al. A MALT lymphoma prognostic index. Blood. 2017;130(12):1409-1417. 17. National Comprehensive Cancer Network. National Comprehensive Cancer Network (NCCN) guidelines - Bcell lymphomas 2017 November 15, 2017; version 6. 18. Zucca E, Copie-Bergman C, Ricardi U, Thieblemont C, Raderer M, Ladetto M. Gastric marginal zone lymphoma of MALT type: ESMO clinical practice guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2013;24(Suppl 6):vi144-148. 19. Liu H, Ruskon-Fourmestraux A, Lavergne-Slove A, et al. Resistance of t(11;18) positive gastric mucosa-associated lymphoid tissue lymphoma to Helicobacter pylori eradication therapy. Lancet. 2001;357(9249):39-40. 20. Ryu KD, Kim GH, Park SO, et al. Treatment outcome for gastric mucosaassociated lymphoid tissue lymphoma according to Helicobacter pylori infection status: a single-center experience. Gut Liver. 2014;8(4):408-414. 21. Ferreri AJ, Cecchetti C, Kiesewetter B, et al. Clarithromycin as a “repurposing drug” against lymphomas: safety and efficacy profiles in 55 patients with extranodal marginal zone lymphoma (EMZL). Hematol Oncol. 2017;35(S2):86-87. 22. Saifi O, Lester SC, Rule W, et al. Comparable efficacy of reduced dose radiation therapy for the treatment of early stage gastric extranodal marginal zone lymphoma of mucosa-associated lymphoid tissue. Adv Radiat Oncol. 2021;6(4):100714.

23. Goda JS, Gospodarowicz M, Pintilie M, et al. Long-term outcome in localized extranodal mucosa-associated lymphoid tissue lymphomas treated with radiotherapy. Cancer. 2010;116(16):3815-3824. 24. Martinelli G, Laszlo D, Ferreri AJ, et al. Clinical activity of rituximab in gastric marginal zone non-Hodgkin's lymphoma resistant to or not eligible for antiHelicobacter pylori therapy. J Clin Oncol. 2005;23(9):1979-1983. 25. Zucca E, Conconi A, Laszlo D, et al. Addition of rituximab to chlorambucil produces superior event-free survival in the treatment of patients with extranodal marginal-zone B-cell lymphoma: 5-year analysis of the IELSG-19 randomized study. J Clin Oncol. 2013;31(5):565-572. 26. Desai A, Joag MG, Lekakis L, et al. Longterm course of patients with primary ocular adnexal MALT lymphoma: a large single-institution cohort study. Blood. 2017;129(3):324-332. 27. Kiesewetter B, Raderer M. Antibiotic therapy in nongastrointestinal MALT lymphoma: a review of the literature. Blood. 2013;122(8):1350-1357. 28. Ferreri AJ, Ponzoni M, Guidoboni M, et al. Regression of ocular adnexal lymphoma after Chlamydia psittaci-eradicating antibiotic therapy. J Clin Oncol. 2005;23(22):5067-5073. 29. Teckie S, Qi S, Lovie S, et al. Long-term outcomes and patterns of relapse of early-stage extranodal marginal zone lymphoma treated with radiation therapy with curative intent. Int J Radiat Oncol Biol Phys. 2015;92(1):130-137. 30. MacManus MP, Roos D, O'Brien P, et al. Prospective phase II trial of radiation therapy in localised non-gastric marginal zone lymphoma with prospective evaluation of autoimmunity and Helicobacter pylori status: TROG 05.02/ALLG NHL15. Eur J Cancer. 2021;152:129-138. 31. de la Fuente MI, Haggiagi A, Moul A, et al. Marginal zone dural lymphoma: the Memorial Sloan Kettering Cancer Center and University of Miami experiences. Leuk Lymphoma. 2017;58(4):882-888. 32. Sunderland AJ, Steiner RE, Al Zahrani M, et al. An international multicenter retrospective analysis of patients with extranodal marginal zone lymphoma and histologically confirmed central nervous system and dural involvement. Cancer Med. 2020;9(2):663-670. 33. Jackson AE, Mian M, Kalpadakis C, et al. Extranodal marginal zone lymphoma of mucosa-associated lymphoid tissue of the salivary glands: a multicenter, international experience of 248 patients (IELSG 41). Oncologist. 2015;20(10): 1149-1153. 34. Zelenetz AD, Gordon LI, Abramson JS, et al. NCCN guidelines insights: B-cell lymphomas, version 3.2019. J Natl Compr Canc Netw. 2019;17(6):650-661. 35. Oh SY, Kim WS, Kim JS, et al. Phase II study of R-CVP followed by rituximab maintenance therapy for patients with advanced marginal zone lymphoma: Consortium for Improving Survival of Lymphoma (CISL) study. Cancer Commun (Lond). 2019;39(1):58. 36. Salar A, Domingo-Domenech E, Panizo C, et al. First-line response-adapted treatment with the combination of bendamustine and rituximab in patients with mucosa-associated lymphoid tissue lymphoma (MALT2008-01): a multicentre, single-arm, phase 2 trial. Lancet

Haematol. 2014;1(3):e104-111. 37. Alderuccio JP, Beaven AW, Shouse G, et al. Frontline bendamustine and rituximab in extranodal marginal zone lymphoma: an international analysis. Blood. 2020;136(Suppl 1):2-3. 38. Knauf W, Abenhardt W, Koenigsmann M, et al. Rare lymphomas in routine practice - treatment and outcome in marginal zone lymphoma in the prospective German Tumour Registry Lymphatic Neoplasms. Hematol Oncol. 2021;39(3): 313-325. 39. Thieblemont C, Molina T, Davi F. Optimizing therapy for nodal marginal zone lymphoma. Blood. 2016;127(17): 2064-2071. 40. Pileri S, Ponzoni M. Pathology of nodal marginal zone lymphomas. Best Pract Res Clin Haematol. 2017;30(1):50-55. 41. Conconi A, Franceschetti S, Aprile von Hohenstaufen K, et al. Histologic transformation in marginal zone lymphomas. Ann Oncol. 2015;26(11):2329-2335. 42. Arcaini L, Paulli M, Boveri E, et al. Splenic and nodal marginal zone lymphomas are indolent disorders at high hepatitis C virus seroprevalence with distinct presenting features but similar morphologic and phenotypic profiles. Cancer. 2004; 100(1):107-115. 43. van den Brand M, van Krieken JH. Recognizing nodal marginal zone lymphoma: recent advances and pitfalls. A systematic review. Haematologica. 2013; 98(7):1003-1013. 44. Spina V, Khiabanian H, Messina M, et al. The genetics of nodal marginal zone lymphoma. Blood. 2016;128(10):1362-1373. 45. Starr AG, Caimi PF, Fu P, et al. Dual institution experience of nodal marginal zone lymphoma reveals excellent long-term outcomes in the rituximab era. Br J Haematol. 2016;175(2):275-280. 46. Heilgeist A, McClanahan F, Ho AD, Witzens-Harig M. Prognostic value of the Follicular Lymphoma International Prognostic Index score in marginal zone lymphoma: an analysis of clinical presentation and outcome in 144 patients. Cancer. 2013;119(1):99-106. 47. Olszewski AJ, Castillo JJ. Survival of patients with marginal zone lymphoma: analysis of the Surveillance, Epidemiology, and End Results database. Cancer. 2013;119(3):629-638. 48. Arcaini L, Besson C, Frigeni M, et al. Interferon-free antiviral treatment in Bcell lymphoproliferative disorders associated with hepatitis C virus infection. Blood. 2016;128(21):2527-2532. 49. Kelaidi C, Rollot F, Park S, et al. Response to antiviral treatment in hepatitis C virusassociated marginal zone lymphomas. Leukemia. 2004;18(10):1711-1716. 50. Kang HJ, Kim WS, Kim SJ, et al. Phase II trial of rituximab plus CVP combination chemotherapy for advanced stage marginal zone lymphoma as a first-line therapy: Consortium for Improving Survival of Lymphoma (CISL) study. Ann Hematol. 2012;91(4):543-551. 51. Rummel MJ, Niederle N, Maschmeyer G, et al. Bendamustine plus rituximab versus CHOP plus rituximab as first-line treatment for patients with indolent and mantle-cell lymphomas: an open-label, multicentre, randomised, phase 3 non-inferiority trial. Lancet. 2013;381(9873):12031210. 52. Brown JR, Friedberg JW, Feng Y, et al. A phase 2 study of concurrent fludarabine

haematologica | 2022; 107(1)

Marginal zone lymphoma

and rituximab for the treatment of marginal zone lymphomas. Br J Haematol. 2009;145(6):741-748. 53. Ferrario A, Pulsoni A, Olivero B, et al. Fludarabine, cyclophosphamide, and rituximab in patients with advanced, untreated, indolent B-cell nonfollicular lymphomas: phase 2 study of the Italian Lymphoma Foundation. Cancer. 2012; 118(16):3954-3961. 54. Laribi K, Tempescul A, Ghnaya H, et al. The bendamustine plus rituximab regimen is active against primary nodal marginal zone B-cell lymphoma. Hematol Oncol. 2017;35(4):536-541. 55. Arcaini L, Rossi D, Paulli M. Splenic marginal zone lymphoma: from genetics to management. Blood. 2016;127(17):20722081. 56. Kalpadakis C, Pangalis GA, Angelopoulou MK, Vassilakopoulos TP. Treatment of splenic marginal zone lymphoma. Best Pract Res Clin Haematol. 2017;30(1-2):139-148. 57. Rossi D, Trifonov V, Fangazio M, et al. The coding genome of splenic marginal zone lymphoma: activation of NOTCH2 and other pathways regulating marginal zone development. J Exp Med. 2012;209 (9):1537-1551. 58. Swerdlow SH, Kuzu I, Dogan A, et al. The many faces of small B cell lymphomas with plasmacytic differentiation and the contribution of MYD88 testing. Virchows Arch. 2016;468(3):259-275. 59. Santos TSd, Tavares RS, Farias DLCd. Splenic marginal zone lymphoma: a literature review of diagnostic and therapeutic challenges. Rev Bras Hematol Hemoter. 2017;39(2):146-154. 60. Piris MA, Onaindia A, Mollejo M. Splenic marginal zone lymphoma. Best Pract Res Clin Haematol. 2017;30(1-2):56-64. 61. Flinn I, van der Jagt R, Chang JE, et al. First-line treatment of iNHL or MCL patients with BR or R-CHOP/R-CVP: results of the BRIGHT 5-year follow-up study. J Clin Oncol. 2017;35(15 suppl): 7500. 62. Rummel MJ, Koenigsmann M, Chow KU, et al. Two years rituximab maintenance vs. observation after first line treatment with bendamustine plus rituximab (B-R) in patients with marginal zone lymphoma (MZL): results of a prospective, randomized, multicenter phase 2 study (the StiL NHL7-2008 MAINTAIN trial). J Clin Oncol. 2018;36(15 suppl):7515. 63. Kalpadakis C, Pangalis GA, Sachanas S, et al. Rituximab monotherapy in splenic marginal zone lymphoma: prolonged responses and potential benefit from maintenance. Blood. 2018;132(6):666670. 64. Williams ME, Hong F, Gascoyne RD, et al. Rituximab extended schedule or retreatment trial for low tumour burden non-follicular indolent B-cell nonHodgkin lymphomas: Eastern Cooperative Oncology Group protocol E4402. Br J Haematol. 2016;173(6):867875. 65. Alderuccio JP, Zhao W, Desai A, et al. Short survival and frequent transformation in extranodal marginal zone lymphoma with multiple mucosal sites presentation. Am J Hematol. 2019;94(5):585596. 66. Hsu A, Kurt H, Zayac AS, Olszewski AJ. CD5 expression in marginal zone lymphoma predicts differential response to rituximab or bendamustine/rituximab.

haematologica | 2022; 107(1)

Leuk Lymphoma. 2021 Sep 1. [Epub ahead of print] 67. Bastidas-Mora G, Bea S, Navarro A, et al. Clinico-biological features and outcome of patients with splenic marginal zone lymphoma with histological transformation. Br J Haematol. 2021 Sep 14. [Epub ahead of print] 68. Cheson BD, Fisher RI, Barrington SF, et al. Recommendations for initial evaluation, staging, and response assessment of Hodgkin and non-Hodgkin lymphoma: the Lugano classification. J Clin Oncol. 2014;32(27):3059-3068. 69. Matutes E, Oscier D, Montalban C, et al. Splenic marginal zone lymphoma proposals for a revision of diagnostic, staging and therapeutic criteria. Leukemia. 2008;22(3):487-495. 70. Copie-Bergman C, Gaulard P, LavergneSlove A, et al. Proposal for a new histological grading system for post-treatment evaluation of gastric MALT lymphoma. Gut. 2003;52(11):1656. 71. Thieblemont C, Cascione L, Conconi A, et al. A MALT lymphoma prognostic index generated from the dataset of the IELSG-19 prospective clinical trial. Blood. 2017;130(12):1409-1417. 72. Montalban C, Abraira V, Arcaini L, et al. Simplification of risk stratification for splenic marginal zone lymphoma: a point-based score for practical use. Leuk Lymphoma. 2014;55(4):929-931. 73. Arcaini L, Lazzarino M, Colombo N, et al. Splenic marginal zone lymphoma: a prognostic model for clinical use. Blood. 2006;107(12):4643-4649. 74. Conconi A, Thieblemont C, Cascione L, et al. Early progression of disease predicts shorter survival in MALT lymphoma patients receiving systemic treatment. Haematologica. 2020;105(11):2592-2597. 75. Beal KP, Yeung HW, Yahalom J. FDG-PET scanning for detection and staging of extranodal marginal zone lymphomas of the MALT type: a report of 42 cases. Ann Oncol. 2005;16(3):473-480. 76. Albano D, Borghesi A, Bosio G, et al. Pulmonary mucosa-associated lymphoid tissue lymphoma: (18)F-FDG PET/CT and CT findings in 28 patients. Br J Radiol. 2017;90(1079):20170311. 77. Schöder H, Noy A, Gönen M, et al. Intensity of 18fluorodeoxyglucose uptake in positron emission tomography distinguishes between indolent and aggressive non-Hodgkin’s lymphoma. J Clin Oncol. 2005;23(21):4643-4651. 78. Noy A, Schoder H, Gonen M, et al. The majority of transformed lymphomas have high standardized uptake values (SUVs) on positron emission tomography (PET) scanning similar to diffuse large B-cell lymphoma (DLBCL). Ann Oncol. 2009;20 (3):508-512. 79. Noy A, de Vos S, Thieblemont C, et al. Targeting Bruton tyrosine kinase with ibrutinib in relapsed/refractory marginal zone lymphoma. Blood. 2017;129(16): 2224-2232. 80. Noy A, de Vos S, Coleman M, et al. Durable ibrutinib responses in relapsed/refractory marginal zone lymphoma: long-term follow-up and biomarker analysis. Blood Adv. 2020;4(22): 5773-5784. 81. Davids MS, Roberts AW, Seymour JF, et al. Phase I first-in-human study of venetoclax in patients with relapsed or refractory non-Hodgkin lymphoma. J Clin Oncol. 2017;35(8):826-833.

82. Handunnetti SM, Khot A, Anderson MA, et al. Safety and efficacy of ibrutinib in combination with venetoclax in patients with marginal zone lymphoma: preliminary results from an open label, phase II study. Blood. 2019;134(Suppl_1):3999. 83. Opat S, Tedeschi A, Linton K, et al. The MAGNOLIA trial: zanubrutinib, a nextgeneration Bruton tyrosine kinase inhibitor, demonstrates safety and efficacy in relapsed/refractory marginal zone lymphoma. Clin Cancer Res. 2021 Sep 15. [Epub ahead of print] 84. Gopal AK, Kahl BS, de Vos S, et al. PI3Kd inhibition by idelalisib in patients with relapsed indolent lymphoma. N Engl J Med. 2014;370(11):1008-1018. 85. Martin P, Armas A, Gopal AK, et al. Idelalisib monotherapy and durable responses in patients with relapsed or refractory marginal zone lymphoma (MZL). Blood. 2015;126(23):1543. 86. Cheah CY, Nastoupil LJ, Neelapu SS, Forbes SG, Oki Y, Fowler NH. Lenalidomide, idelalisib, and rituximab are unacceptably toxic in patients with relapsed/refractory indolent lymphoma. Blood. 2015;125(21):3357-3359. 87. Gilead Sciences Inc. Idelalisib Urgent Safety Letter. 2016. 88. Lampson BL, Kasar SN, Matos TR, et al. Idelalisib given front-line for treatment of chronic lymphocytic leukemia causes frequent immune-mediated hepatotoxicity. Blood. 2016;128(2):195-203. 89. Dreyling M, Santoro A, Mollica L, et al. Phosphatidylinositol 3-kinase inhibition by copanlisib in relapsed or refractory indolent lymphoma. J Clin Oncol. 2017;35(35):3898-3905. 90. Panayiotidis P, Follows GA, Mollica L, et al. Efficacy and safety of copanlisib in patients with relapsed or refractory marginal zone lymphoma. Blood Adv. 2021;5(3):823-828. 91. Fowler NH, Samaniego F, Jurczak W, et al. Umbralisib, a dual PI3Kd/CK1e inhibitor in patients with relapsed or refractory indolent lymphoma. J Clin Oncol. 2021;39(15):1609-1618. 92. Phillips TJ, Corradini P, Gurion R, et al. Phase 2 study evaluating the efficacy and safety of parsaclisib in patients with relapsed or refractory marginal zone lymphoma (CITADEL-204). Blood. 2020;136 (Suppl 1):27-28. 93. Jacobson C, Chavez JC, Sehgal AR, et al. Primary analysis of Zuma-5: a phase 2 study of axicabtagene ciloleucel (axi-cel) in patients with relapsed/refractory (R/R) indolent non-Hodgkin lymphoma (iNHL). Blood. 2020;136(Suppl 1):40-41. 94. Leonard JP, Jung S-H, Johnson J, et al. Randomized trial of lenalidomide alone versus lenalidomide plus rituximab in patients with recurrent follicular lymphoma: CALGB 50401 (Alliance). J Clin Oncol. 2015;33(31):3635-3640. 95. Fowler NH, Davis RE, Rawal S, et al. Safety and activity of lenalidomide and rituximab in untreated indolent lymphoma: an open-label, phase 2 trial. Lancet Oncol. 2014;15(12):1311-1318. 96. Becnel MR, Nastoupil LJ, Samaniego F, et al. Lenalidomide plus rituximab (R(2) ) in previously untreated marginal zone lymphoma: subgroup analysis and long-term follow-up of an open-label phase 2 trial. Br J Haematol. 2019;185(5):874-882.

EDITORIALS Fusion genes in acute myeloid leukemia: do acute myeloid leukemia diagnostics need to fuse with RNA-sequencing? Felicitas Thol Department of Hematology, Hemostasis, Oncology, and Stem Cell Transplantation, Hannover Medical School, Hannover, Germany E-mail: FELICITAS THOL - thol.felicitas@mh-hannover.de doi:10.3324/haematol.2021.278983

n this issue of Haematologica, Kerbs and colleagues present a detailed analysis of fusion gene detection by RNAsequencing in acute myeloid leukemia (AML).1 Fusion genes, i.e., genes made by joining parts of two different genes, occur frequently in carcinogenesis. These fusion transcripts can occur within one chromosome as well as between two chromosomes (inter-chromosomal) and include translocations, deletions, insertions as well as inversions. Thousands of chromosomal aberrations and fusion genes have been identified as being recurrently present in cancer.2 Interestingly, in pediatric acute lymphocytic leukemia it was shown that many fusion genes are already present in utero and that a further transformation occurs in early childhood through secondary mutations.3 The discovery of BCR-ABL1 in patients with chronic myeloid leukemia as well as gene fusions involving neurotrophic tyrosine receptor kinase (NTRK)4 and neuregulin-1 (NRG1)5 in solid cancers are examples of how the identification of fusion genes have not only revolutionized the diagnosis but also the therapy of a disease. It is estimated that fusion genes occur in 30% of patients with AML6 and it has been recognized that certain recurrent chromosomal and genetic aberrations delineate distinct types of AML. Along this line, the current World Health Organization (WHO) classification of AML describes seven types of recurrent genetic abnormalities in AML as well as the provisional entity “AML with BCR-ABL1”.7 Furthermore, the WHO classification lists recurrent genetic aberrations that classify the disease as “AML with myelodysplasia-related changes”. This

underscores the relevance of identifying chromosomal aberrations and fusion genes in AML patients. The European LeukemiaNet recommends cytogenetics and screening for certain gene mutations for all AML patients at the time of diagnosis.8 Screening for gene rearrangements (e.g., with fluorescence in situ hybridization or reverse transcription polymerase chain reaction [RT-PCR]) should also be performed in some special circumstances (e.g., poor quality of chromosome morphology, need for rapid information). However, although cytogenetic analysis is still the gold standard, many small aberrations can be easily missed with this technique. RNA-sequencing is a next-generation sequencing technique that detects the presence and quantity of RNA and thus describes the transcriptome. Additionally, it gives information about alternative splicing, mutations, changes in gene expression, fusion genes, etc. We can expect additional information that we would not be able to obtain with routine cytogenetic analysis. Thus, RNA-sequencing can give us a deeper and dynamic picture of what is happening in cells. Kerbs and colleagues applied RNA-sequencing to a cohort of 806 AML samples that also underwent conventional diagnostics with karyotyping and molecular diagnostics (RT-PCR and fluorescence in situ hybridization) (Figure 1). This allowed a direct comparison between results obtained with these different techniques. Special attention was paid to the detection of fusion genes. Here, 90% of fusion genes reported by conventional diagnostics were also detected by RNA-sequencing. Nevertheless, the fact that some fusion genes were not recog-

Figure 1. Design and outlook of an integrated analysis that includes routine diagnostics (karyotyping, molecular diagnostics) as well as RNA-sequencing in acute myeloid leukemia. In this case, 806 samples from four studies were investigated. RNA-seq: RNA-sequencing; MDx: molecular diagnostics; AML: acute myeloid leukemia.

haematologica | 2022; 107(1)

Editorials

nized by RNA-sequencing indicates a weakness of this novel technique. Some fusion genes can be missed by RNA-sequencing if only low or uneven coverage is achieved. Thus, a high read depth of samples is essential for RNA-sequencing in order to detect all fusion genes. Furthermore, some affected genes might be more easily missed if they are not highly expressed (i.e., RNAsequencing is highly dependent on gene expression). However, RNA-sequencing allowed the authors to detect 26 cases with known recurrent fusion events that failed to be detected by routine diagnostics. Furthermore, Kerbs and colleagues also detected two novel recurrent fusion genes by RNA-sequencing. The fusion gene NRIP1MIR99AHG was found in 1.1% of patients, and the fusion gene LTN1-MX1 in 0.25% of patients. These are just examples that demonstrate that many more recurrent fusion genes will be detected if RNA-sequencing becomes employed more widely in AML. It is therefore very likely that we currently still underestimate the occurrence of fusion genes in AML with conventional diagnostics. The encouraging results with RNA-sequencing also raise the questions of whether the technique should be applied to all AML patients at the time of diagnosis and whether it could replace conventional diagnostics. Currently, there are still a few obstacles to routine use of RNA-sequencing in all AML patients. Besides costs and labor-consuming aspects of this technique, analysis of results and data interpretation are very challenging. In particular, the high rate of false positive calls for fusion genes raises concerns and requires awareness. So far, interpretation of RNA-sequencing data has not been standardized and over 20 algorithms for fusion gene detection by RNA-sequencing have been described.9 However, standardization is a critical requirement for using this technique more widely and in routine clinical practice. In this sense, it is encouraging as well as important that Kerbs and colleagues have developed and report a workflow with integrated filtering strategies for the identification of robust fusion gene candidates by RNA-sequencing. Similar steps need to be undertaken in the future in order to implement RNA-sequencing in routine diagnostics. The wider use of RNA-sequencing in AML is likely to have positive impacts on diagnostic classification as well as prognostic stratification. RNA-sequencing can also open the door to novel therapeutic strategies in individual AML patients. Gene fusions involving the NTRK gene constitute an encouraging example. NTRK fusions rarely occur in AML and are easily missed with routine diagnostics.10 Larotrectinib is a targeted therapy that has shown efficacy in a broad spectrum of NTRK fusion-positive cancers. Many fusion genes occurring through chromosomal translocations encode tyrosine kinases that are involved in signal transduction. The development of tyrosine kinase inhibitors represents an attrac-

haematologica | 2022; 107(1)

tive therapeutic strategy for patients with such fusions and RNA-sequencing will be essential for identifying these patients. It is therefore very likely that the increase in detection of recurrent gene fusions will also expand opportunities for targeted therapies in AML. RNA-sequencing may also have further implications for monitoring measurable residual disease in AML. Fusion genes are ideally suited for molecular measurable residual disease assessment as they can be detected by RT-PCR as well as next-generation sequencing very sensitively.11 Of note, RNA-sequencing is evolving as an upcoming standard in solid cancers as well. In summary, the study by Kerbs and colleagues is an important step towards implementing RNA-sequencing in routine AML diagnostics. Further work is undoubtedly required (especially regarding standardizing data analysis), but the manuscript already gives an encouraging outlook of how RNA-sequencing can translate into clinical applications. Disclosures No conflicts of interest to disclose.

References 1. Kerbs P, Vosberg S, Krebs S, et al. Fusion gene detection by RNAsequencing complements diagnostics of acute myeloid leukemia and identifies recurring NRIP1-MIR99AHG rearrangements. Haematologica. 2021;107(1):100-111. 2. Mitelman DatabaseChromosome Aberrations and Gene Fusions in Cancer. https://mitelmandatabase.isb-cgcorg/. 2021. Last accessed June 30, 2021. 3. Wiemels JL, Cazzaniga G, Daniotti M, et al. Prenatal origin of acute lymphoblastic leukaemia in children. Lancet. 1999;354(9189):14991503. 4. Drilon A, Laetsch TW, Kummar S, et al. Efficacy of larotrectinib in TRK fusion-positive cancers in adults and children. N Engl J Med. 2018;378(8):731-739. 5. Laskin J, Liu SV, Tolba K, et al. NRG1 fusion-driven tumors: biology, detection, and the therapeutic role of afatinib and other ErbB-targeting agents. Ann Oncol. 2020;31(12):1693-1703. 6. Grimwade D, Hills RK, Moorman AV, et al. Refinement of cytogenetic classification in acute myeloid leukemia: determination of prognostic significance of rare recurring chromosomal abnormalities among 5876 younger adult patients treated in the United Kingdom Medical Research Council trials. Blood. 2010;116(3):354-365. 7. Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood. 2016;127(20):2391-2405. 8. Dohner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017;129(4):424-447. 9. Haas BJ, Dobin A, Li B, Stransky N, Pochet N, Regev A. Accuracy assessment of fusion transcript detection via read-mapping and de novo fusion transcript assembly-based methods. Genome Biol. 2019;20(1):213. 10. Joshi SK, Qian K, Bisson WH, et al. Discovery and characterization of targetable NTRK point mutations in hematologic neoplasms. Blood. 2020;135(24):2159-2170. 11. Thol F, Gabdoulline R, Liebich A, et al. Measurable residual disease monitoring by NGS before allogeneic hematopoietic cell transplantation in AML. Blood. 2018;132(16):1703-1713.

Editorials

Adding eltrombopag to immunosuppression: the importance of predicting outcome André Tichelli,1,2 Régis Peffault de Latour,2,3 Carlo Dufour2,4 and Alicia Rovó2,5 1

Hematology, University Hospital, Basel, Switzerland; 2Severe Aplastic Anemia Working Party of the European Group for Blood and Marrow Transplantation, Leiden, the Netherlands; 3French Reference Center for Aplastic Anemia and PNH, BMT Unit, Saint-Louis Hospital. – AP-HP, Université de Paris, Paris, France; 4Hemato-Oncology and Stem Cell Transplantation Pole, G. Gaslini, Research Children’s Hospital (IRCCS), Genova, Italy and 5Department of Hematology and Central Hematology Laboratory, Inselspital, Bern University Hospital, University of Bern, Bern, Switzerland E-mail: ANDRÉ TICHELLI - andre.tichelli@gmail.com doi:10.3324/haematol.2021.278761

evere aplastic anemia (SAA) can be successfully treated today, with either allogeneic hematopoietic cell transplantation (HCT) or standard immunosuppressive therapy (IST) with horse antithymocyte globulin (hATG) and cyclosporine. The choice of one of these treatments is usually based on the availability of a matched sibling donor and the age of the patient. Despite good survival rates for both forms of treatment, there are relevant differences with respect to the long-term outcome. Thus, patients undergoing HCT are at risk of early complications mainly related to the toxicity of the preparative regimen, infections and acute graft-versus-host disease (GvHD), while in the long-term, they are prone to develop chronic GvHD. In contrast, patients treated with IST often do not have a complete recovery of blood counts, remaining with a certain degree of cytopenia, and are exposed to a series of risks including relapse of the underlying disease, development of a clinical paroxysmal nocturnal hemoglobinuria as well as late malignant clonal complications, such as myelodysplastic syndrome or acute myeloid leukemia. A major challenge of treating SAA patients is the identification of pretreatment factors predicting response, other than just age and donor availability, which could help to make the best therapeutic decision. A group from the National Institutes of Health contributed in this regard, showing in a previous study that pretreatment peripheral blood values were able to predict response of SAA patients treated with hATG and cyclosporine.1 More recently, eltrombopag, a thrombopoietin-receptor agonist added to standard IST, has been shown to improve the rate of response even for SAA patients, refractory to standard IST.2,3 Although a randomized comparison to standard immunosuppression is warranted and should be available soon (RACE; ClinicalTrials.gov number, NCT02099747; https://clinicaltrials.gov/ct2/show/NCT02099747, the use of a combination of hATG, cyclosporine and eltrombopag is tending to supplant standard IST for SAA and factors predicting response need to be confirmed with this triple combination. In this issue of Haematologica, Zaimoku et al. show that baseline blood counts, such as reticulocyte, neutrophil, and lymphocyte counts, and thrombopoietin level, which are surrogates of residual hematopoiesis, remain the best predictors of response at 6 months after treatment with hATG, cyclosporine and eltrombopag.4 Interestingly, when compared to IST with hATG and cyclosporine alone, adding eltrombopag was predictive of response at 6 months, even with reticulocyte baseline counts below 30x109/L (between 10 and 30x109/L), suggesting that the addition of eltrombopag to IST could increase the

response rate also for SAA patients with less residual hematopoiesis. Furthermore, the authors suggested an age-dependent association between response and absolute lymphocyte counts, demonstrating that lower absolute lymphocyte counts in children under 10 years of age were predictive of complete response, while for older children and adults the reverse was the case. Despite overall good survival after IST, event-free survival remains largely unsatisfactory. In a randomized study on IST with or without granulocyte colony-stimulating factor, the overall survival at 15 years was 60%, but the event-free survival was as low as 24%. This divergence was even more pronounced in patients 30 years of age or younger, who had an overall survival of 89% but an event-free survival of only 27% (Figure 1A)5, meaning that children and young adults treated with IST, despite excellent overall survival, still have a substantial number of complications later in life. After HCT, long-term follow-up shows greater coherence between overall survival and event-free survival, even when unrelated donors have been used (Figure 1B, C).6,7 A retrospective study of a selected cohort of pediatric patients with SAA demonstrated that upfront unrelated HCT produced similar results to those of matched sibling donor HCT and better than those of matched unrelated donor HCT after failure of IST.8 These data raise the question of whether upfrontunrelated donor HCT could be a first-line option in selected pediatric or young adult patients who lack a matched sibling donor and are at risk of refractoriness to IST. The broader the therapeutic options, the more important factors predicting response and outcome of IST become. Event-free survival in patients treated with IST is mainly influenced by early deaths due to infections, refractoriness to treatment, relapse, the development of clinical paroxysmal nocturnal hemoglobinuria or transformation into a clonal myeloid neoplasm. Predicting the outcome of IST in SAA does, therefore, include three relevant issues: prediction of immediate response to treatment (affected by early infections and refractoriness to treatment), prediction of a robust and sustainable response in the long-term (affected by relapse of SAA), and prediction of the outcome in respect to late hematologic complications (affected by clonal malignant myeloid disease and clinical paroxysmal nocturnal hemoglobinuria). In a convincing manner, Zaimoku et al. determined factors predicting response to treatment at 6 months. Predicting events occurring after 6 months of IST would also be important in SAA patients, but this remains an unmet need in the field. In fact, identification of risk factors for relapse, development of paroxysmal nocturnal hemoglobinuria or transformation into

haematologica | 2022; 107(1)

Editorials

malignant clonal hematopoietic diseases, which are relevant events affecting event-free survival after 6 months, would contribute considerably to predicting long-term outcomes. The age-dependent association between response and absolute lymphocyte counts in the study by Zaimoku et al. is an interesting but not fully understood finding. The

question of why lower absolute lymphocyte counts in children below 10 years predict complete response, while the reverse is true in older children and adults remains unanswered. A confirmatory study would be needed to validate the conflicting results. These paradoxical results remind us of another study in which pediatric patients with very SAA treated with IST showed unexpectedly

Figure 1. Survival outcomes of patients with severe aplastic anemia treated with immunosuppressive therapy or hematopoietic stem cell transplantation and a treatment algorithm. (A) Comparison of overall survival (OS) and event-free survival (EFS) of patients with severe aplastic anemia (SAA) 30 years of age and younger treated for SAA with immunosuppressive therapy (IST) with or without granulocyte colony-stimulating factor (from unpublished data of the dataset of a randomized study5). Events were defined as refractoriness to first IST, relapse of SAA, development of clinical paroxysmal nocturnal hemoglobinuria or transformation into a myelodysplastic syndrome or acute leukemia. OS at 15 years is 89% and EFS 27% There is a great divergence between OS and EFS, meaning that children and young adults treated with IST, despite an excellent overall survival, still have a substantial number of complications later in life. These patients could, therefore, be considered as candidates for an alternative donor hematopoietic cell transplantation (HCT) if no matched sibling donor is available. (B, C) Survival outcomes in terms of OS (B) and EFS (C) following upfront-unrelated HCT in children with SAA. The divergence of OS and EFS is much less in patients treated with HCT than in those treated with IST (Figure from Dufour C et al.8). (D) The algorithm for first-line treatment adapted from the recommendations of the European Blood and Marrow Transplant Severe Aplastic Anemia Working Party (SAA-WP).10 Treatment with eltrombopag has been added in red, since it is not yet officially in the guideline. According to the recommendations of the SAA-WP, patients <20 years) could be considered for first-line HCT with a matched unrelated donor (MUD) if the donor is available within 2 to 3 months. The pretreatment predictive values discussed in the study by Zaimoku et al.4 could be helpful for decision-making. Indeed, patients with factors predicting an unfavorable response to IST could be the patients considered for first-line transplantation from a MUD. HSCT: hematopoietic stem cell transplantation; SE: standard error; MSD: matched sibling donor; ATG: antithymocyte globulin; CSA: cyclosporine A; MTX: methotrexate; hATG: horse antithymocyte globulin; EPAG: eltrombopag; BMT: bone marrow transplantation.

haematologica | 2022; 107(1)

Editorials

better response to treatment and overall survival than children with less severe SAA.9 Unfortunately, baseline lymphocyte counts in that study were not available. We could hypothesize that similar mechanisms could explain discrepant results between young children and adult patients. The data in the present study could also become a supportive tool to approach children and young adults with SAA who lack a matched sibling donor. The management of newly diagnosed pediatric and young adult SAA patients could be based on the following algorithm (Figure 1D): when baseline blood values are predictive of a unsatisfactory response to IST combined with eltrombopag, and the chance of rapidly finding a suitable alternative donor exists, upfront first-line HCT with this alternative donor should be taken in serious consideration. Future research in this field should be directed at identifying factors predicting long-term outcome, which would bring more precise information to adapt up-front treatment strategies. Disclosures No conflicts of interest to disclose. Contributions AT, RPL, CD and AR wrote the mansucript.

References 1. Scheinberg P, Wu CO, Nunez O, Young NS. Predicting response to immunosuppressive therapy and survival in severe aplastic anaemia.

Br J Haematol. 2009;144(2):206-216. 2. Assi R, Garcia-Manero G, Ravandi F, et al. Addition of eltrombopag to immunosuppressive therapy in patients with newly diagnosed aplastic anemia. Cancer. 2018;124(21):4192-4201. 3. Townsley DM, Scheinberg P, Winkler T, et al. Eltrombopag added to standard immunosuppression for aplastic anemia. N Engl J Med. 2017;376(16):1540-1550. 4. Zaimoku Y, Patel BA, Shalhoub R, et al. Predicting response of severe aplastic anemia to immunosuppression combined with eltrombopag. Haematologica. 2021;107(1):126-133. 5. Tichelli A, de Latour RP, Passweg J, et al. Long-term outcome of a randomized controlled study in patients with newly diagnosed severe aplastic anemia treated with antithymocyte globulin and cyclosporine, with or without granulocyte colony-stimulating factor: a Severe Aplastic Anemia Working Party trial from the European Group of Blood and Marrow Transplantation. Haematologica. 2020;105(5):1223-1231. 6. Yang S, Yuan X, Ma R, et al. Comparison of outcomes of frontline immunosuppressive therapy and frontline haploidentical hematopoietic stem cell transplantation for children with severe aplastic anemia who lack an HLA-matched sibling donor. Biol Blood Marrow Transplant. 2019;25(5):975-980. 7. Dufour C, Pillon M, Passweg J, et al. Outcome of aplastic anemia in adolescence: a survey of the Severe Aplastic Anemia Working Party of the European Group for Blood and Marrow Transplantation. Haematologica. 2014;99(10):1574-1581. 8. Dufour C, Veys P, Carraro E, et al. Similar outcome of upfront-unrelated and matched sibling stem cell transplantation in idiopathic paediatric aplastic anaemia. A study on behalf of the UK Paediatric BMT Working Party, Paediatric Diseases Working Party and Severe Aplastic Anaemia Working Party of EBMT. Br J Haematol. 2015;171(4):585-594. 9. Fuhrer M, Rampf U, Baumann I, et al. Immunosuppressive therapy for aplastic anemia in children: a more severe disease predicts better survival. Blood. 2005;106(6):2102-2104. 10. de Latour RP, Risitano A, Dufour C. Severe Aplastic Anemia and PNH. In: Carreras E, Dufour C, Mohty M, Kroger N, editors. The EBMT Handbook: Hematopoietic Stem Cell Transplantation and Cellular Therapies. 7th edition. Cham (CH): 2019. p. 579-585.

haematologica | 2022; 107(1)

ARTICLE

Acute Lymphoblastic Leukemia

Pre-existing antibodies against polyethylene glycol reduce asparaginase activities on first administration of pegylated E. coli asparaginase in children with acute lymphocytic leukemia Alaeddin Khalil,1 Gudrun Würthwein,1 Jana Golitsch,1 Georg Hempel,2 Manfred Fobker,3 Joachim Gerss,4 Anja Möricke,5 Martin Zimmermann,6 Petr Smisek,7 Massimo Zucchetti,8 Christa Nath,9 Andishe Attarbaschi,10 Arend von Stackelberg,11 Nicola Gökbuget,12 Carmelo Rizzari,13 Valentino Conter,13 Martin Schrappe,5 Joachim Boos1 and Claudia Lanvers-Kaminsky1

Ferrata Storti Foundation

Haematologica 2022 Volume 107(1):49-57

Department of Pediatric Hematology and Oncology, University Children's Hospital Münster, Münster, Germany; 2Department of Pharmaceutical and Medical Chemistry, Clinical Pharmacy, University of Münster, Münster, Germany; 3Center of Laboratory Medicine, University Hospital Münster, Albert-Schweitzer-Campus 1, Münster, Germany; 4Institute of Biostatistics and Clinical Research, University of Münster, Münster, Germany; 5Department of Pediatrics, University Medical Center SchleswigHolstein, Campus Kiel, Kiel, Germany: 6Department of Pediatric Hematology and Oncology, Medical School Hannover, Hannover, Germany; 7Department of Pediatric Hematology and Oncology, University Hospital Motol, Praha, Czech Republic; 8 Laboratory of Cancer Pharmacology, Department of Oncology, Istituto di Ricerche Farmacologiche Mario Negri IRCCS, Milan, Italy; 9Departments of Biochemistry and Oncology, The Children’s Hospital at Westmead, Sydney Pharmacy School, University of Sydney, Sydney, Australia; 10Department of Pediatric Hematology and Oncology, St. Anna Children’s Hospital, Department of Pediatrics and Adolescent Medicine, Medical University of Vienna, Vienna, Austria; 11Departments of Pediatric Oncology/Hematology and of General Pediatrics, Charité -University Medicine Berlin, Berlin, Germany; 12 Department of Medicine, University Hospital, Frankfurt am Main, Germany and 13 Pediatric Hematology-Oncology Unit, Department of Pediatrics, University of MilanoBicocca, MBBM Foundation, ASST-Monza, Monza, Italy 1

ABSTRACT

ntibodies against polyethylene glycol (PEG) in healthy subjects raise concerns about the efficacy of pegylated drugs. We evaluated the prevalence of antibodies against PEG among patients with acute lymphoblastic leukemia (ALL) prior to and/or immediately after their first dose of pegylated E.coli asparaginase (PEG-ASNase). Serum samples from 701 children (673 with primary ALL, 28 with relapsed ALL) and 188 adults with primary ALL were analyzed for anti-PEG IgG and IgM. Measurements in 58 healthy infants served as a reference to define cut-points for antibody-positive and -negative samples. The prevalence of anti-PEG antibodies in ALL patients prior to the first administration of PEG-ASNase was 13.9% for anti-PEG IgG and 29.1% for anti-PEG IgM. After administration of PEG-ASNase the prevalence of anti-PEG antibodies decreased to 4.2% for anti-PEG IgG and to 4.5% for anti-PEG IgM. Pre-existing anti-PEG antibodies did not inhibit PEGASNase activity but significantly reduced PEG-ASNase activity levels in a concentration-dependent manner. Although pre-existing anti-PEG antibodies were not boosted, pre-existing anti-PEG IgG were significantly associated with first-exposure hypersensitivity reactions (Common Terminology Criteria for Adverse Events grade 2) (P<0.01; Fisher exact test). Two of four patients with pre-existing anti-PEG IgG and first-exposure hypersensitivity reactions were not switched to Erwinia ASNase and continued on PEG-ASNase with sufficient activity (≥100 U/L). In conclusion, pre-existing anti-PEG antibodies were detected in a considerable proportion of patients with ALL and although they did not inhibit PEG-ASNase activity, they were associated with lower haematologica | 2022; 107(1)

Correspondence: CLAUDIA LANVERS-KAMINSKY lanvers@uni-muenster.de Received: May 19, 2020. Accepted: November 25, 2020. Pre-published: December 10, 2020. https://doi.org/10.3324/haematol.2020.258525

A. Khalil et al.

serum PEG-ASNase activity levels. Patients with pre-existing antibodies may show mild to moderate signs of hypersensitivity reaction after their first administration PEG-ASNase, which may be successfully addressed by re-challenge.

Introduction Due to its favorable toxicity profile polyethylene glycol (PEG) is widely used in foods, cosmetics, and pharmaceuticals.1 Pegylation can improve the therapeutic benefit of protein drugs. It prolongs their elimination by increasing the molecular mass and protecting them from enzymatic cleavage and it decreases their immunogenicity by shielding potential antigenic epitopes.2-4 Numerous pegylated drugs are currently marketed in the USA and Europe including pegylated uricase (KrystexxaTM), pegylated interferon (PegasysTM, PegIntronTM) and pegylated E. coli asparaginase (PEG-ASNase) (OncasparTM, calaspargase, AsparlasTM).5-8 The asparagine-hydrolyzing enzyme asparaginase (ASNase) is crucial for the successful treatment of acute lymphoblastic leukemia (ALL)9,10 and because of its favorable drug characteristics PEG-ASNase is increasingly replacing its unmodified native form in frontline treatment of ALL.11-13 While pegylated proteins, despite their higher molecular mass, tend to be less immunogenic than their non-pegylated forms of protein drugs, antibodies against PEG have been detected in patients treated with pegylated proteins as well as in healthy volunteers.14 The reported prevalence varies widely between studies (0.2-72%) which is partly due to the use of different detection methods and cutpoint definitions (Online Supplementary Tables S1 and S2). In animal studies anti-PEG antibodies, especially anti-PEG IgM, were considered responsible for the accelerated blood clearance of pegylated proteins, liposomes, and nanoparticles.15,16 In human studies, the reported effects of anti-PEG antibodies on the therapeutic efficacy of pegylated drugs have been ambiguous; no effects of antibodies against PEG have been observed for pegylated interferons to date,17 whereas in patients with gout anti-PEG IgM and anti-PEG IgG were associated with a faster elimination of PEG-uricase.18,19 Drug authorities now require evaluation of the relevance of anti-PEG antibodies during drug development and registration processes.20,21 Published data suggest that anti-PEG antibodies may have important effects on the efficacy of PEG-ASNase. Armstrong et al. detected anti-PEG antibodies in 12 of 15 patients with undetectable ASNase activities after PEGASNase administration and also in four of 12 patients before their first PEG-ASNase administration.22 Liu et al. recently showed that anti-PEG ASNase antibodies consisted mainly of antibodies against PEG rather than E. coli Table 1. Demographics of the patients in the three cohorts of acute lymphocytic leukemia cases.

Protocol Number Sex (M/F) Age, years Median Range

ALL-cohort 1 ALL-cohort 2 ALL-cohort 3 AIEOP-BFM ALL 2009 ALL-REZ BFM 2002 & GMALL 07/2003 ALL-REZ BFM Observ.a 673 401/272

28 19/9

188 120/68

5.6 1 - 18

8.5 5 - 17

36 18 - 74

Observational Study and Biobank. M: male; F: female.

ASNase and were significantly associated with hypersensitivity reactions to PEG ASNase.23 Given the increasing use of PEG-ASNase in frontline treatment for ALL, the aims of this study were to: (i) evaluate the prevalence of anti-PEG antibodies in three cohorts of patients (children and adults with primary ALL and children with relapsed ALL) before and/or immediately after their first dose of PEG-ASNase during induction treatment, and (ii) investigate the effects of pre-existing anti-PEG antibodies on PEG-ASNase activities and hypersensitivity reactions.

Methods Patients Samples for anti-PEG antibody determination were obtained from children with primary ALL (ALL-cohort 1), children with relapsed ALL (ALL-cohort 2), adults with primary ALL (ALLcohort 3) and healthy infants, who served as the reference cohort. Patients in ALL-cohort 1 were treated according to the AIEOPBFM ALL 2009 trial (ClinicalTrials.gov identifier: NCT01117441) and a total of 673 plasma samples were collected from 673 pediatric patients (401 males, 272 females) prior to their first administration of PEG-ASNase. In addition, 646 patients provided one or two more serum samples (1,183 in total) taken within 15 days after the first PEG-ASNase dose on day 12 of induction. Patients in ALL-cohort 2 were diagnosed with relapsed ALL and treated according to the protocol of the ALL-REZ BFM 2002 (ClinicalTrials.gov identifier: 00114348) or the ALL-REZ BFM Observational Study and Biobank study. Twenty-eight samples were collected from 28 patients (19 males, 9 females) 0 to 2 days after the first dose of PEG-ASNase Patients in ALL-cohort 3 were treated according to the multicenter GMALL 07/2003 trial (ClinicalTrials.gov identifier: 00198991). A total of 188 samples from 120 males and 68 females were taken on the same day after the first administration of PEGASNase (n=16) or the following day (n=172). Further details on the ALL cohorts are provided in Table 1 and in the Online Supplementary Data. The respective ALL studies were approved by national and local review boards in accordance with the Helsinki Declaration and national laws. The approvals included monitoring antibodies against PEG-ASNase and determination of ASNase activity. Patients and/or their guardians gave their signed informed consent to participate in the monitoring of ASNase activities and antibodies against PEG-ASNase. Serum samples from 58 infants <1 year, who were considered naïve to PEG were used as reference. The Central Laboratory of the University Hospital Muenster provided anonymized remainders of routine serum samples from infants. Only age in months was disclosed. Thus, these samples were considered as completely anonymized leftover material.

Determination of antibodies against PEG For the detection of anti-PEG IgG and anti-PEG IgM the flow cytometry method described by Armstrong et al.22 was transferred to a 96-well format with fluorescent read-out. TentaGel M OCH particles (10 mm), to which methoxy-polyethylene glycol chains with a mean molecular weight of 5,000 Da were covalently 3

haematologica | 2022; 107(1)

Effect of pre-existing anti-PEG antibodies on PEG-ASNase

The MFI for anti-PEG IgG was between 0.65 and 67.4, whereas that for anti-PEG IgM was between 0.13 and 30.8. Overall, anti-PEG IgG levels correlated with anti-PEG IgM levels (r=0.68, P<0.005, Pearson correlation). However, high anti-PEG IgG levels did not necessarily coincide with

high anti-PEG IgM levels and vice versa (Online Supplementary Tables S3 and S4). On average the lowest levels of anti-PEG IgG and IgM were determined in the reference cohort and the highest levels in children with primary ALL prior to their first dose of PEG-ASNase (ALLcohort 1) (Figure 1). In ALL-cohort 1 anti-PEG IgG and IgM levels were significantly lower after the administration of PEG-ASNase. This difference was statistically significant in an unpaired analysis, when all samples available after the first administration were included (n=1,183, median days after administration: 13; range: 1-15; P<0.001, Mann-Whitney rank sum test) and in a paired analysis, when only the first sample taken after administration was chosen for the pairwise comparison (n=646, median days after administration: 7; range: 1-15; P<0.001, Wilcoxon signed-rank test). In addition, anti-PEG IgG and IgM levels were also significantly lower in ALL-cohorts 2 and 3, which were analyzed after PEG-ASNase administration (P<0.001, Kruskal-Wallis oneway analysis of variance on ranks, all pairwise multiple comparison procedures [Dunn method]) (Figure 1). The prevalence of anti-PEG antibodies was correspondingly lower in samples/patients analyzed after administration of PEG-ASNase (Figure 2C-E). In ALL-cohort 1 13.9% of samples were positive for anti-PEG IgG and 29.1% positive for anti-PEG IgM prior to the administration of PEG-ASNase. After administration of PEG-ASNase the prevalence dropped to 4.2% for anti-PEG IgG and 4.5% for anti-PEG IgM. This represented a significant reduction in prevalence by PEG-ASNase administration (P<0.0001, McNemar χ2 test with continuity correction) (Figure 2B, C). Among

bound, were used as the antigen (RAPP Polymere, Tuebingen, Germany). Mean fluorescent intensities (MFI) of duplicate determinations were calculated for anti-PEG IgG and IgM levels. Based on the MFI, determined in the reference cohort, cut-points of 8 (anti-PEG IgG) and 2 (anti-PEG IgM) were defined to classify samples as positive or negative. A detailed description of the method for the determination of anti-PEG antibodies and its performance characteristics is included in the Online Supplementary Data along with a description of the measurement of PEG-ASNase activity and total IgG and IgM.

Statistics Statistical analyses were performed using SAS® Version 9.4 (SAS Institute Inc., Cary, NC, USA) and RStudio Version 1.2.5033 (RStudio: Integrated Development for R. RStudio, Inc., Boston, MA, USA. URL: http://www.rstudio.com/). Kruskal-Wallis one way analysis of variance on ranks, all pairwise multiple comparison procedures (Dunn method, Holm-Sidak method), the MannWhitney rank sum test, Wilcoxon signed-rank test, χ2 test, McNemar χ2 test with continuity correction, Fisher exact test, Pearson correlation and logistic regression were used as indicated.

Results Anti-PEG IgG and anti-PEG IgM antibody levels

Figure 1. Box plots of anti-PEG antibodies in the reference and acute lymphocytic leukemia cohorts. (A, B) Box plots of anti-PEG IgG (A) and anti-PEG IgM (B) mean fluorescent intensities (MFI) determined in the reference and the acute lymphocytic leukemia (ALL) cohorts. The boxes represent the first and third quartiles, the lines in the box the represent the medians, the whiskers the first quartile – (1.5 x the interquartile range between the first and third quartiles [IQR]) and the third quartile + (1.5 x IQR), and the dots the outliers. The dashed reference lines represent the cut-points for anti-PEG IgG (MFI = 8) (A) and anti-PEG IgM (MFI = 2) (B).

haematologica | 2022; 107(1)

A. Khalil et al. B

Figure 2. Scatter plots of anti-PEG IgM versus anti-PEG IgG levels determined in the reference and the acute lymphocytic leukemia cohorts. (A) Samples of the reference cohort were taken from healthy infants. (B, C) Samples of acute lymphocytic leukemia (ALL)-cohort 1 were taken from patients treated according to the AIEOPBFM ALL 2009 trial and were either collected prior to their first dose of PEG-asparatinase (ASNase) (B) or within 15 days after administration of the first PEG-ASNase dose (C). (D, E) Samples from ALL-cohort 2 (children with relapsed ALL treated according to the ALL-REZ BFM 2002 and the ALL-REZ BFM ALL observational study and biobank) (D) and samples from ALL-cohort 3 (adults treated according to the GMALL 07/2003 protocol) (E) were taken after administration of PEG-ASNase. AntiPEG antibodies were determined by the level of mean fluorescent intensity (MFI). In (A) the dashed light blue vertical line represents the 95th percentile of anti-PEG IgM MFI and the dashed green horizontal line represents the 95th percentile of anti-PEG IgG MFI determined in the reference cohort. The solid reference lines represent the defined cut-points after visual adjustment for anti-PEG IgM (light blue vertical line, MFI = 2) and anti-PEG IgG (green horizontal line, MFI = 8).

patients with pre-existing anti-PEG antibodies, the antibody levels decreased a mean of about 2.7-fold for antiPEG IgG and about 4.1-fold for anti-PEG IgM.

PEG-asparaginase activities PEG-ASNase activities were determined in 1,183 samples from 646 patients of ALL-cohort 1. Samples were collected within 15 days of the first administration of 2,500 U/m2 PEG-ASNase (maximum 3,750 U per dose) on day 12 of induction. Of these samples, 95.5% were collected at the scheduled times (day 7±1 and day 14±1 after administration). The mean (± standard deviation) PEG-ASNase activities determined were 911±311 U/L on day 7±1 and 527±200 U/L on day 14±1. PEG-ASNase activities were significantly lower among patients with elevated anti-PEG IgG (MFI ≥8) or anti-PEG IgM (MFI ≥2) prior to their first dose of PEG-ASNase (P<0.05, Kruskal-Wallis one way analysis of variance on ranks, all pairwise multiple comparison procedures [HolmSidak method]). In addition, the PEG-ASNase activities decreased with increasing anti-PEG antibody levels prior to administration (Figure 3). To evaluate the effect of antiPEG antibodies on PEG-ASNase activities in individual patients, mean PEG-ASNase activities were calculated for the respective day after administration and individual PEGASNase activities were categorized as above or below the respective means. Pre-existing anti-PEG IgG (MFI ≥8) as well as pre-existing anti-PEG IgM (MFI ≥2) increased the risk of PEG-ASNase activities below average (anti-PEG 52

IgG: odds ratio [OR]: 2.06, 95% confidence interval [95% CI]: 1.44-2.96; anti-PEG IgM: OR: 1.65; 95% CI: 1.27-2.15; P<0.001, χ2 test). No such associations were observed for anti-PEG IgG and IgM levels determined after PEG-ASNase administration. Pre-existing anti-PEG antibodies reduced but did not eliminate PEG-ASNase activities (Figure 3). Comparing the distribution of PEG-ASNase activities above and below 400 U/L, 100 U/L and the lower limit of quantitation (LLOQ = 5 U/L), significantly more samples with PEG-ASNase activities <400 U/L were found in patients with already existing anti-PEG antibodies (Table 2). No differences were observed for the distribution of PEG-ASNase activities above and below 100 U/L and the LLOQ (Table 2). Thus, silent inactivation of PEG-ASNase, which is defined by PEG-ASNase activities <100 U/L within 7±1 days and/or undetectable PEG-ASNase activities within 14±1 days after administration without signs of hypersensitivity reaction, was not affected by pre-existing anti-PEG antibodies.24,25 Anti-PEG antibodies did not inhibit the catalytic activity of PEG-ASNase and anti-PEG IgG and/or IgM had no effect on asparagine hydrolysis by PEGASNase (Online Supplementary Figure S4).

Anti-PEG antibodies prior to treatment with PEG-asparaginasese and hypersensitivity reactions to PEG-asparaginase After initial exposure to PEG-ASNase, seven patients in ALL-cohort 1 (1.0%) showed hypersensitivity reactions (all Common Terminology Criteria for Adverse Events haematologica | 2022; 107(1)

Effect of pre-existing anti-PEG antibodies on PEG-ASNase

Figure 3. Box plots of PEG-asparaginase activities after the first dose of the drug. PEG-asparaginase (ASNase) activities were determined in patients of acute lymphoblastic leukemia (ALL) cohort 1 on day 7±1 and day 14±1 after administration of the first dose of PEG-ASNase and grouped according to various cut-points for pre-existing anti-PEG IgG and IgM mean fluorescent intensity (MFI). The light gray boxes represent PEG-ASNase activities determined in patients with anti-PEG antibody levels below the respective cut-point and the dark gray boxes represent PEG-ASNase activities determined in patients with anti-PEG antibody levels above the respective cut-point. The boxes represent the first and third quartiles, the lines in the box the medians, the whiskers the first quartile – (1.5 x the interquartile range between the first and third quartiles [IQR]) and the third quartile + (1.5 x IQR), and the dots the outliers. The dashed reference line represents the target PEG-ASNase activity of 100 U/L.

[CTCAE] grade 2) which were significantly associated with pre-existing anti-PEG IgG levels. Four of seven patients had pre-existing anti-PEG IgG (MFI ≥8) before their first PEG-ASNase dose (Table 3). No pre-existing antiE. coli ASNase antibodies were detected in these patients. Four patients (2 with and 2 without pre-existing anti-PEG antibodies) were switched to Erwinia ASNase. Among the four patients with a first-exposure hypersensitivity reaction and pre-existing anti-PEG IgG no further boosts of anti-PEG IgG levels were observed. The two patients with pre-existing anti-PEG IgG, who continued on PEGASNase, completed the scheduled PEG-ASNase treatment without further signs of hypersensitivity. The relative risk of a hypersensitivity reaction upon first exposure to PEGASNase was 8 times higher for patients with anti-PEG IgG MFI ≥8 and 50 times higher for patients with anti-PEG IgG MFI ≥25 prior to their first PEG-ASNase (Table 3). This association was only observed for pre-existing anti-PEG IgG and not for pre-existing anti-PEG IgM.

Discussion We detected a high prevalence of anti-PEG IgG (13.9%) haematologica | 2022; 107(1)

and IgM (29.1%) among children with primary ALL prior to their first PEG-ASNase. Antibodies against PEG had already been detected in healthy volunteers of different ages and ethnicity and in patients who had never been treated with pegylated drugs before. The prevalence of anti-PEG antibodies in ALLcohort 1 was within the range of reported prevalences (0.2 to 72%).14,18,19,22,26–30 However, it must be acknowledged that it is difficult to compare the prevalence between different studies when different methods and cut-points were used (Online Supplementary Tables S1 and S2). The reported effects of pre-existing anti-PEG antibodies on the efficacy and tolerability of pegylated drugs vary.14,17– 19,31,32 Unexpectedly, the first administration of PEG-ASNase did not trigger the formation of further anti-PEG antibodies. Instead, anti-PEG antibody levels and their prevalence decreased, which was different from a typical hypersensitivity reaction to PEG-ASNase that developed after repeated administration of PEG-ASNase.23,33–35 We also observed no increase in anti-PEG antibodies in the seven patients with hypersensitivity reaction at first exposure to PEGASNase. Four of these patients had pre-existing anti-PEG IgG (MFI ≥8) (Table 3). This association was significant and the risk 53

A. Khalil et al.

Table 2. Distribution of PEG-asparaginase activities below and above various thresholds among patients with and without preexisting anti-PEG antibodies.

PEG-ASNase levels < LLOQ ≥ LLOQ < 100 U/L ≥ 100 U/L < 400 U/L ≥ 400 U/L

Pre-existing anti-PEG IgG yes no [MFI ≥ 8] [MFI < 8] 2 88 6 84 35 55

4 552 7 549 116 440

Pre-existing anti-PEG IgM yes no [MFI ≥ 2] [MFI < 2] 2 189 6 185 55 136

4 451 7 448 96 359

odds ratio (95% CI)

P-value

adjusted P-value

3.14

(0.57-17.4)

0.198

0.535

5.60

(5.6-17.1)

0.005

0.085

2.41

(1.51-3.86)

0.0005

0.002

P-value

adjusted P-value

odds ratio (95% CI) 1.19

(0.22-6.57)

0.535

2.08

(0.69-6.62)

0.22

0.476

1.51

(1.03-2.22)

0.045

0.021

For each patient of ALL-cohort 1 only the lowest PEG-ASNase activities determined within 15 days after administration of 2500 U/m2 PEG-ASNase were evaluated. PEG-ASNase: polyethylene glycol asparaginase; MFI: mean fluorescence intensity; 95% CI: 95% confidence interval; LLOQ: lower limit of quantification; >: above threshold; <: below threshold.

for hypersensitivity reaction increased with increasing antiPEG IgG levels prior to PEG-ASNase administration. A significant association between pre-existing anti-PEG antibodies and first-exposure hypersensitivity reaction was also documented in the RADAR phase IIb clinical trial, which evaluated pegnivacogin, a 2’-fluoropyrimidine-modified RNA aptamer, in patients with acute coronary syndrome.31 Among the six patients with the highest anti-PEG antibody levels prior to pegnivacogin administration, three suffered from a first-exposure hypersensitivity reaction. The firstexposure hypersensitivity reactions in the RADAR phase IIb clinical trial affected only 0.5% of patients but were considered serious and led to early termination of the trial.31 In ALL-cohort 1 the first-exposure hypersensitivity reactions to PEG-ASNase were, however, only moderate (CTCAE grade 2). Symptoms of moderate hypersensitivity reactions (CTCAE grade ≤2) and infusion-related adverse events are often difficult to distinguish.36 Typically, hypersensitivity reactions occur after re-challenge to the antigen and are associated with an increase in antibodies, which can persist in the blood for up to several weeks.23,37 Since this was not the case in patients with first-exposure hypersensitivity reactions to PEG-ASNase and only moderate hypersensitivity reactions (CTCAE grade 2) occurred, one might conclude that pre-existing anti-PEG IgG simply predispose to mild hypersensitivity reactions for which re-challenge with PEG-ASNase may be possible. The two ALL patients with pre-existing anti-PEG IgG who developed first-exposure hypersensitivity reactions to PEG-ASNase and did not switch to Erwinia ASNase tolerated their subsequent PEG-ASNase administrations well. Pre-existing anti-PEG antibodies are most likely triggered by repeated contact with PEG-containing products of daily life, such as cosmetics, pharmaceuticals and food. Depending on the nature of the PEG-containing compound, different immunological mechanisms are supposed to facilitate the anti-PEG antibody response.28,38,39 Experiments in nude mice showed that pegylated proteins induced the production of anti-PEG IgM in a T-cell54

dependent manner, whereas the induction of anti-PEG IgM by pegylated liposomes was T-cell independent.39,40 Furthermore, studies in animals indicated that these different immunological processes may also lead to antibodies with different properties.40 Similar processes might also be feasible in humans and might explain why patients with pre-existing antibodies showed different antibody responses after their first PEG-ASNase dose than patients who developed a hypersensitivity reaction to the PEG-ASNase after repeated administrations and in whom the PEG covalently bound to the bacterial ASNase acted as a hapten.23,37 According to the “Consensus expert recommendations for identification and management of ASNase hypersensitivity and silent inactivation” discontinuation of treatment is recommended for grade ≥2 allergic reactions.24 Recently, the National Comprehensive Cancer Network clinical practice guidelines and other expert reviews on ASNase hypersensitivity recommended switching of ASNase preparations only in the event of grade ≥3, severe or life-threatening allergic or anaphylactic reactions.36,41–43 In addition, because of the repeated shortage of Erwinia ASNase, various strategies were evaluated in order to avoid or delay a switch to Erwinia ASNase as long as possible.36,44,45 Thus, PEG-ASNase was either generally administered under premedication or in the case of hypersensitivity reactions grade ≤2 under premedication at initially reduced infusion rates. In each case, PEG-ASNase activity was monitored to detect silent inactivation or premedication-masked hypersensitivity reactions. Pre-existing anti-PEG antibodies reduced the PEGASNase activity levels as a function of concentration, but did not reduce the PEG-ASNase activity levels to such an extent that the criteria of silent inactivation were fulfilled. Silent inactivation (or subclinical hypersensitivity reaction) is characterized by the development of antibodies without overt symptoms of a hypersensitivity reaction.24,25 According to the “Consensus expert recommendations for identification and management of asparaginase hypersensitivity and silent inactivation” silent inactihaematologica | 2022; 107(1)

Effect of pre-existing anti-PEG antibodies on PEG-ASNase

Table 3. Distribution of pre-existing anti-PEG IgG levels (at various thresholds) among patients of ALL-cohort 1 with and without first-exposure hypersensitivity reactions to PEG-asparaginase.

Anti-PEG IgG level prior to first PEG-ASNase MFI > 25 MFI ≤ 25 MFI > 15 MFI ≤ 15 MFI > 8 MFI ≤ 8

Hypersensitivity reaction to first PEG-ASNase No Yes 7 659 26 640 90 576

3 4 4 3 4 3

Odds ratio (95% CI)

Relative risk (95% CI)

Statistics PPV (95% CI)

NPV (95% CI)

Sensitivity (95% CI)

Specificity (95% CI)

0.00008

70.6 (10.1-497) 32.8 (5.80-197) 8.53 (1.59-48.9)

49.7 (9.33-228) 28.6 (5.57-157) 8.21 (1.58-45.7)

0.30 (0.07-0.65) 0.13 (0.04-0.31) 0.04 (0.01-0.11)

0.99 (0.98-1.00) 0.99 (0.99-1.00) 0.99 (0.98-1.00)

0.43 (0.10-0.82) 0.57 (0.18-0.90) 0.57 (0.18-0.90)

0.99 (0.98-1.00) 0.96 (0.94-0.97) 0.86 (0.84-0.89)

0.0001 0.009

a Fisher exact test; PEG-ASNase: polyethylene glycol asparaginase; 95% CI: 95% confidence interval; PPV: positive predictive value; NPV: negative predictive value; MFI: mean fluorescence intensity

vation of PEG-ASNase is defined by PEG-ASNase activities ≤100 U/L within 7 days and undetectable PEGASNase activities within 14 days after administration.24 Neutralizing antibodies but also an accelerated elimination of antigen-antibody complexes are being discussed as the underlying cause for the rapid decrease in ASNase activity.24,25 We could not detect any inhibition of asparagine hydrolysis by anti-PEG antibodies (Online Supplementary Figure S4). Animal studies have shown an increased clearance of pegylated proteins, liposomes and nanoparticles in the presence of anti-PEG IgM and antiPEG IgG.15,16 In nude mice, anti-PEG IgM induced a rapid clearance of pegylated protein from serum with simultaneous accumulation in the liver. Similar processes could also be conceivable in humans. The increased clearance of PEG uricase in gout patients was associated with an increase in anti-PEG IgG and IgM levels.18,19 The lower PEG-ASNase activity levels in patients with pre-existing antibodies might have been caused by an increased clearance of PEG-ASNase. When comparing the distribution of PEG-ASNase activities above and below 400 U/L, 100 U/L and the LLOQ (5 U/L), we found a significant difference between patients with and without pre-existing anti-PEG antibodies only at 400 U/L. The 400 U/L value was chosen in addition to the generally accepted target activity of 100 U/L and the LLOQ because 400 U/L have been shown to result in deeper asparagine depletion.46,47 This higher activity level and its associated glutaminase activity have been suggested to increase the effectiveness of ASNase against leukemic blasts with residual asparagine synthetase activity.48 Several studies have recently shown that PEGASNase clearance in a patient can vary significantly between different parts of the protocol.49,50 Thus, the effect of increased PEG-ASNase clearance due to preexisting antibodies on the intensity of ASNase therapy would depend on the dose and concomitant ALL treatment. Therefore, the effects of pre-existing anti-PEG antibodies on the pharmacokinetics of PEG-ASNase must be determined separately for each protocol. In summary, we observed a considerable number of patients with pre-existing antibodies against PEG. AntiPEG antibody kinetics after PEG-ASNase administration were not the same in patients with pre-existing antibodies as in patients with hypersensitivity reactions after repeated PEG-ASNase administration.23,37 Pre-existing anti-PEG antibodies may cause mild to moderate symptoms of hypersensitivity reaction with the first administration of PEG-ASNase, which might be addressed by rechallenge. They do not inhibit PEG-ASNase activity but haematologica | 2022; 107(1)

lower PEG-ASNase activity levels, which, depending on the dose and protocol, may interfere with the targeted PEG-ASNase treatment intensity. Disclosures MS and the ALL-BFM Study Center have received funding from medac, Shire, SigmaTau, Servier, and Jazz Pharmaceuticals for research, and MS has participated in advisory boards. CR received institutional grants for PEG-asparaginase pharmacological studies aimed at the therapeutic monitoring of asparaginase activity within the AIEOP-BFM ALL 2009 study, as well as fees for participation in advisory boards and invited lectures for the companies involved in marketing different asparaginase products, medac GmbH, Sigma-Tau, Baxalta, Shire, Servier. CLK has held invited talks for Sigma tau, Erytech and Jazz Pharmaceuticals, received honoraria for consultancy from Erytech and travel expenses from Erytech and Servier. AA has consulting and advisory roles for Jazz Pharmaceuticals, and has also received travel and accommodation fees and expenses from Jazz Pharmaceuticals. AM has received honoraria from Baxter. JB has served personally as a consultant and participated in advisory as well as in safety boards for medac GmbH; he has received support for travel from Eusa Pharma, Jazz Pharmaceuticals, Baxalta Shire and Servier; has held invited lectures for medac GmbH, Eusa Pharma, Jazz Pharmaceuticals, Baxalta, Servier, Shire and Sigma-Tau; and, in addition, he has received institutional grants in the context of ASNase drug monitoring from more or less all ASNase providers contributing to the therapeutic drug monitoring program, including medac GmbH, Eusa Pharma, Jazz Pharmaceuticals, Baxalta, Servier, Shire, and Sigma-Tau (all representing the varying marketing authorization holders of E. coli ASNase, PEG-ASNase and Erwinase). NG has received research support from Jazz Pharmaceuticals and honoraria as an advisory board member from Jazz Pharmaceuticals and Baxalta. The other authors declare that they have no potential conflicts of interest. Contributions AK, CLK, GW and JB designed the research. CR, MZu, VC, CN, AA, AM, MS, JB, AvS and NG collected data. AA, CN, MZu, MF, CLK, GH and JB analyzed and interpreted the data. AK, GW, JG, MZi, AM and CLK performed the statistical analysis. AK and CLK wrote the manuscript. Acknowledgments The authors would like to thank Ms Schulze Westhoff, Ms Hoogestraat and Ms Günter for their excellent technical assistance. We acknowledge support from the Open Access Publication Fund of the University of Muenster. 55

A. Khalil et al.

Funding This work was supported by the Deutsche José Carreras Leukämie-Stiftung e.V. (DJCLSR 13/01). The monitoring of PEG-ASNase activities and of antibodies against PEG-ASNase and native E. coli ASNase was supported by Servier (Suresnes, France), Shire (Lexington, USA), Baxalta (Westlake Village, USA), Sigma Tau Pharmaceuticals (Zofingen, Switzerland) and medac GmbH (Wedel, Germany). In Australia, ASNase monitoring and the collection of samples for antibody determination for the

References 1. Fruijtier-Pölloth C. Safety assessment on polyethylene glycols (PEGs) and their derivatives as used in cosmetic products. Toxicology. 2005;214(1-2):1-38. 2. Caliceti P, Veronese FM. Pharmacokinetic and biodistribution properties of poly(ethylene glycol)-protein conjugates. Adv Drug Deliv Rev. 2003;55(10):1261-1277. 3. Harris JM, Chess RB. Effect of pegylation on pharmaceuticals. Nat Rev Drug Discov. 2003;2(3):214-221. 4. Werle M, Bernkop-Schnürch A. Strategies to improve plasma half life time of peptide and protein drugs. Amino Acids. 2006;30(4):351367. 5. Sherman MR, Saifer MGP, Perez-Ruiz F. PEG-uricase in the management of treatment-resistant gout and hyperuricemia. Adv Drug Deliv Rev. 2008;60(1):59-68. 6. Dinndorf PA, Gootenberg J, Cohen MH, Keegan P, Pazdur R. FDA drug approval summary: pegaspargase (oncaspar) for the first-line treatment of children with acute lymphoblastic leukemia (ALL). Oncologist. 2007;12(8):991-998. 7. Barnard DL. Pegasys (Hoffmann-La Roche). Curr Opin Investig Drugs. 2001;2(11):15301538. 8. Graham ML. Pegaspargase: a review of clinical studies. Adv Drug Deliv Rev. 2003;55(10):1293-1302. 9. Pieters R, Hunger SP, Boos J, et al. L-asparaginase treatment in acute lymphoblastic leukemia: a focus on Erwinia asparaginase. Cancer. 2011;117(2):238-249. 10. Hunger SP, Mullighan CG. Acute lymphoblastic leukemia in children. N Engl J Med. 2015;373(16):1541-1552. 11. Place AE, Stevenson KE, Vrooman LM, et al. Intravenous pegylated asparaginase versus intramuscular native Escherichia coli Lasparaginase in newly diagnosed childhood acute lymphoblastic leukaemia (DFCI 05001): a randomised, open-label phase 3 trial. Lancet Oncol. 2015;16(16):1677-1690. 12. Zeidan A, Wang ES, Wetzler M. Pegasparaginase: where do we stand? Expert Opin Biol Ther. 2009;9(1):111-119. 13. Rytting M. Peg-asparaginase for acute lymphoblastic leukemia. Expert Opin Biol Ther. 2010;10(5):833-839. 14. Yang Q, Lai SK. Anti-PEG immunity: emergence, characteristics, and unaddressed questions. Wiley Interdiscip Rev Nanomed Nanobiotechnol. 2015;7(5):655-677. 15. Poppenborg SM, Wittmann J, Walther W, et al. Impact of anti-PEG IgM antibodies on the pharmacokinetics of pegylated asparaginase preparations in mice. Eur J Pharm Sci. 2016;91:122-130. 16. Ishida T, Kashima S, Kiwada H. The contribution of phagocytic activity of liver macrophages to the accelerated blood clearance (ABC) phenomenon of PEGylated lipo-

AIEOP-BFM ALL 2009 trial were funded by the Kids Cancer Alliance (KCA), a Translational Cancer Research Center of the Cancer Institute of New South Wales (15/TRC/1-04). CN is supported by the Leukemia Research and Support Fund at The Children’s Hospital at Westmead (Australia). In Austria monitoring of ASNase activity and collection of samples for antibody determination for the AIEOP-BFM ALL 2009 trial were funded by the St. Anna Kinderkrebsforschung (BFM-Austria). The work is part of the thesis of AK.

somes in rats. J Control Release. 2008;126 (2):162-165. 17. Myler H, Hruska MW, Srinivasan S, et al. Anti-PEG antibody bioanalysis: a clinical case study with PEG-IFN-g-1a and PEG-IFNa2a in naive patients. Bioanalysis. 2015;7(9):1093-1106. 18. Ganson NJ, Kelly SJ, Scarlett E, Sundy JS, Hershfield MS. Control of hyperuricemia in subjects with refractory gout, and induction of antibody against poly(ethylene glycol) (PEG), in a phase I trial of subcutaneous PEGylated urate oxidase. Arthritis Res Ther. 2006;8(1):R12. 19. Hershfield MS, Ganson NJ, Kelly SJ, Scarlett EL, Jaggers DA, Sundy JS. Induced and preexisting anti-polyethylene glycol antibody in a trial of every 3-week dosing of pegloticase for refractory gout, including in organ transplant recipients. Arthritis Res Ther. 2014;16 (2):R63. 20. U.S. Department of Health and Human Services, Food and Drug Administration. Guidance for Industry: Immunogenicity Assessment for Therapeutic Protein Products. https://www.fda.gov/regulatoryinformation/search-fda-guidance-documents/immunogenicity-assessment-therapeutic-protein-products. Accessed 11.11.2020 21. Parenky A, Myler H, Amaravadi L, et al. New FDA draft guidance on immunogenicity. AAPS J. 2014;16(3):499-503. 22. Armstrong JK, Hempel G, Koling S, et al. Antibody against poly(ethylene glycol) adversely affects PEG-asparaginase therapy in acute lymphoblastic leukemia patients. Cancer. 2007;110(1):103-111. 23. Liu Y, Smith CA, Panetta JC, et al. Antibodies predict pegaspargase allergic reactions and failure of rechallenge. J Clin Oncol. 2019;37(23):2051-2061. 24. van der Sluis IM, Vrooman LM, Pieters R, et al. Consensus expert recommendations for identification and management of asparaginase hypersensitivity and silent inactivation. Haematologica. 2016;101(3):279-285. 25. Asselin B, Rizzari C. Asparaginase pharmacokinetics and implications of therapeutic drug monitoring. Leuk Lymphoma. 2015;56 (8):2273-2280. 26. Richter AW, Akerblom E. Polyethylene glycol reactive antibodies in man: titer distribution in allergic patients treated with monomethoxy polyethylene glycol modified allergens or placebo, and in healthy blood donors. Int Arch Allergy Appl Immunol. 1984;74(1):36-39. 27. Liu Y, Reidler H, Pan J, et al. A double antigen bridging immunogenicity ELISA for the detection of antibodies to polyethylene glycol polymers. J Pharmacol Toxicol Methods. 2011;64(3):238-245. 28. Lubich C, Allacher P, de la Rosa M, et al. The mystery of antibodies against polyethylene glycol (PEG) - what do we know?. Pharm

Res. 2016;33(9):2239-2249. 29. Chen B-M, Su Y-C, Chang C-J, et al. Measurement of pre-existing IgG and IgM antibodies against polyethylene glycol in healthy individuals. Anal Chem. 2016;88 (21):10661-10666. 30. Yang Q, Jacobs TM, McCallen JD, et al. Analysis of pre-existing IgG and IgM antibodies against polyethylene glycol (PEG) in the general population. Anal Chem. 2016;88(23):11804-11812. 31. Ganson NJ, Povsic TJ, Sullenger BA, et al. Pre-existing anti-polyethylene glycol antibody linked to first-exposure allergic reactions to pegnivacogin, a PEGylated RNA aptamer. J Allergy Clin Immunol. 2016;137(5):1610-1613. 32. Tillmann H, Ganson NJ, Patel K, et al. High prevalence of pre-existing antibodies against polyethylene glycol (PEG) in hepatitis C (HCV) patients which is not associated with impaired response to PEG-interferon. J Hepatol. 2010;52:S129. 33. Henriksen LT, Nersting J, Raja RA, et al. Cerebrospinal fluid asparagine depletion during pegylated asparaginase therapy in children with acute lymphoblastic leukaemia. Br J Haematol. 2014;166(2):213220. 34. Henriksen LT, Harila-Saari A, Ruud E, et al. PEG-asparaginase allergy in children with acute lymphoblastic leukemia in the NOPHO ALL2008 protocol. Pediatr Blood Cancer. 2015;62(3):427-433. 35. Tong WH, Pieters R, Kaspers GJL, et al. A prospective study on drug monitoring of PEGasparaginase and Erwinia asparaginase and asparaginase antibodies in pediatric acute lymphoblastic leukemia. Blood. 2014;123(13):2026-2033. 36. Burke MJ, Rheingold SR. Differentiating hypersensitivity versus infusion-related reactions in pediatric patients receiving intravenous asparaginase therapy for acute lymphoblastic leukemia. Leuk Lymphoma. 2017;58(3):540-551. 37. Willer A, Gerss J, König T, et al. AntiEscherichia coli asparaginase antibody levels determine the activity of second-line treatment with pegylated E coli asparaginase: a retrospective analysis within the ALL-BFM trials. Blood. 2011;118(22):5774-5782. 38. Mond JJ, Vos Q, Lees A, Snapper CM. T cell independent antigens. Curr Opin Immunol. 1995;7(3):349-354. 39. Ishida T, Wang X, Shimizu T, Nawata K, Kiwada H. PEGylated liposomes elicit an anti-PEG IgM response in a T cell-independent manner. J Control Release. 2007;122(3): 349-355. 40. Mima Y, Hashimoto Y, Shimizu T, Kiwada H, Ishida T. Anti-PEG IgM is a major contributor to the accelerated blood clearance of polyethylene glycol-conjugated protein. Mol Pharm. 2015;12(7):2429-2435. 41. Marini BL, Perissinotti AJ, Bixby DL, Brown

haematologica | 2022; 107(1)

Effect of pre-existing anti-PEG antibodies on PEG-ASNase

J, Burke PW. Catalyzing improvements in ALL therapy with asparaginase. Blood Rev. 2017;31(5):328-338. 42. Benitez L, Perissinotti AJ, Santarossa M, Marini BL. Pharmacokinetic and clinical considerations for monitoring asparaginase activity levels during pegaspargase therapy. Pediatr Blood Cancer. 2015;62(6):1115. 43. Verma A, Chen K, Bender C, Gorney N, Leonard W, Barnette P. PEGylated E. coli asparaginase desensitization: an effective and feasible option for pediatric patients with acute lymphoblastic leukemia who have developed hypersensitivity to pegaspargase in the absence of asparaginase Erwinia chrysanthemi availability. Pediatr Hematol Oncol. 2019;36(5):277-286. 44. Marini BL, Brown J, Benitez L, et al. A singlecenter multidisciplinary approach to managing the global Erwinia asparaginase short-

haematologica | 2022; 107(1)

age. Leuk Lymphoma. 2019;60(12):28542868. 45. Cooper SL, Young DJ, Bowen CJ, Arwood NM, Poggi SG, Brown PA. Universal premedication and therapeutic drug monitoring for asparaginase-based therapy prevents infusion-associated acute adverse events and drug substitutions. Pediatr Blood Cancer. 2019;66(8):e27797. 46. Angiolillo AL, Schore RJ, Devidas M, et al. Pharmacokinetic and pharmacodynamic properties of calaspargase pegol Escherichia coli L-asparaginase in the treatment of patients with acute lymphoblastic leukemia: results from Children's Oncology Group Study AALL07P4. J Clin Oncol. 2014;32(34): 3874-3882. 47. Avramis VI, Panosyan EH. Pharmacokinetic/ pharmacodynamic relationships of asparaginase formulations: the past, the present and

recommendations for the future. Clin Pharmacokinet. 2005;44 (4):367-393. 48. Chan WK, Lorenzi PL, Anishkin A, et al. The glutaminase activity of L-asparaginase is not required for anticancer activity against ASNS-negative cells. Blood. 2014;123(23): 3596-3606. 49. Kloos RQH, Mathôt R, Pieters R, van der Sluis IM. Individualized dosing guidelines for PEGasparaginase and factors influencing the clearance: a population pharmacokinetic model. Haematologica. 2021;106(5):12541261. 50. Lanvers-Kaminsky C, Niemann A, Eveslage M, et al. Asparaginase activities during intensified treatment with pegylated E. coli asparaginase in adults with newly-diagnosed acute lymphoblastic leukemia. Leuk Lymphoma. 2020;61(1):138-145.

ARTICLE Ferrata Storti Foundation

Acute Myeloid Leukemia

Targeting MCL-1 dysregulates cell metabolism and leukemia-stroma interactions and re-sensitizes acute myeloid leukemia to BCL-2 inhibition Bing Z. Carter,1 Po Yee Mak,1* Wenjing Tao,1* Marc Warmoes,2 Philip L. Lorenzi,2 Duncan Mak,1 Vivian Ruvolo,1 Lin Tan,2 Justin Cidado,3 Lisa Drew,3 and Michael Andreeff1

Haematologica 2022 Volume 107(1):58-76

Section of Molecular Hematology & Therapy, Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX; 2Department of Bioinformatics & Computational Biology, The University of Texas MD Anderson Cancer Center, Houston, TX and 3Bioscience Oncology R&D, AstraZeneca, Boston, MA, USA

*PYM and WT contributed equally as co-second authors.

ABSTRACT

Correspondence: BING Z. CARTER bicarter@mdanderson.org MICHAEL ANDREEFF mandreef@mdanderson.org Received: May 21, 2020 Accepted: December 14, 2020. Pre-published: December 23, 2020. https://doi.org/10.3324/haematol.2020.260331

CL-1 and BCL-2 are both frequently overexpressed in acute myeloid leukemia (AML) and critical for the survival of AML cells and AML stem cells. MCL-1 is a key factor in venetoclax resistance. Using genetic and pharmacological approaches, we discovered that MCL-1 regulates leukemia cell bioenergetics and carbohydrate metabolisms, including the TCA cycle, glycolysis and pentose phosphate pathway and modulates cell adhesion proteins and leukemia-stromal interactions. Inhibition of MCL-1 sensitizes to BCL-2 inhibition in AML cells and AML stem/progenitor cells, including those with intrinsic and acquired resistance to venetoclax through cooperative release of pro-apoptotic BIM, BAX, and BAK from binding to anti-apoptotic BCL2 proteins and inhibition of cell metabolism and key stromal microenvironmental mechanisms. The combined inhibition of MCL-1 by MCL-1 inhibitor AZD5991 or CDK9 inhibitor AZD4573 and BCL-2 by venetoclax greatly extended survival of mice bearing patient-derived xenografts established from an AML patient who acquired resistance to venetoclax/decitabine. These results demonstrate that co-targeting MCL-1 and BCL-2 improves the efficacy of and overcomes pre-existing and acquired resistance to BCL-2 inhibition. Activation of metabolomic pathways and leukemia-stroma interactions are newly discovered functions of MCL-1 in AML, which are independent from canonical regulation of apoptosis by MCL-1. Our data provide new mechanisms of synergy and a rationale for co-targeting MCL-1 and BCL-2 clinically in patients with AML and potentially other cancers.

Introduction Acute myeloid leukemia (AML) is a highly heterogeneous and aggressive hematological malignancy with dismal treatment outcomes. The survival of AML cells, including stem/progenitor cells, depends on deregulated apoptosis, which partially results from the altered expression of BCL-2 family proteins. Thus, BCL-2 family proteins are promising therapeutic targets in AML.1,2 While combined inhibition of BCL-2 and BCL-XL caused profound thrombocytopenia,3 a highly selective BCL-2 inhibitor, venetoclax, demonstrated potent preclinical activity, but only marginal efficacy in clinical trials in AML.2,4 Conversely, combinations of venetoclax with hypomethylating agents have yielded complete response (CR) or CR with incomplete cell count recovery rates of >65% in elderly AML patients, but overall survival is only 10 to 17 months due to the development of resistance.5-7 Unlike other BCL-2 proteins, MCL-1 is short-lived and highly regulated at the transcriptional, post-transcriptional, and post-translational levels8 in response to various stimuli. MCL-1 is required for the sustained growth of diverse oncogene-

haematologica | 2022; 107(1)

Targeting AML MCL-1 re-sensitizes BCL-2 inhibition

driven hematological malignancies9,10 and essential for AML development and AML stem cell self-renewal and survival.11-13 MCL-1 is upregulated in about half of resistant/relapsed AML patients and associated with poor prognosis.14 Importantly, high levels of MCL-1 expression are associated with resistance to venetoclax.2,15,16 Furthermore, MCL-1 was reported to increase the generation of reactive oxygen species (ROS) in lung cancer cells,17 and to positively regulate mitochondrial oxidative phosphorylation (OxPhos) in breast cancer, inducing the breast cancer stem cells, and promoting tumor chemoresistance.18 These studies suggest that, in addition to its anti-apoptotic function, MCL-1 is coupled to tumor cell metabolism to promote cell survival. Indeed, AML cells and stem cells exhibit increased mitochondrial OxPhos,19-21 suggesting that MCL-1 has a role in regulating mitochondrial OxPhos and altering AML cell metabolism. Several specific MCL-1 inhibitors have been developed recently. S63845 (Servier/Novartis) was very effective and well tolerated in in vivo models.22 Others such as AZD5991 (AstraZeneca)23 and AMG176/AMG397 (Amgen)24 have entered clinical trials. Additionally, inhibitors of CDK9 that transcriptionally regulates shortlived proteins like MCL-1 are under development as a strategy to target MCL-1, and the CDK9 inhibitor AZD457325 has entered clinical trials. Rationales supporting BCL-2 and MCL-1 co-targeting have focused on apoptotic mechanisms. Targeting MCL1 with a BIM transgene enhanced venetoclax activity against AML.26 We reported a combinatorial regimen of BCL-2 inhibition and MDM2 inhibition-mediated p53activation that targeted MCL-1 activity and stability. This combination was synthetically lethal to AML cells27 and in early clinical trials elicits 40-50% CR rates among relapsed/refractory AML patients.28 Recent studies in murine models of AML showed that the combined inhibition of BCL-2 and MCL-1 has potent antitumor activity, enhances venetoclax activity, and kills venetoclaxresistant disease.23,24,29,30 However, the mechanisms underlying this synergism, including its effects on AML stem and venetoclax-resistant cells, have not been fully investigated. Here, we determine the effects of MCL-1 inhibition with the novel MCL-1 inhibitor AZD5991 or the CDK9 inhibitor AZD4573 alone and in combination with venetoclax on multiple phenotypic AML stem/progenitor cell populations in vitro and in an in vivo patient-derived xenograft (PDX) model. Additionally, our in vitro study was conducted with bone marrow (BM)-derived mesenchymal stromal cell (MSC) co-culture, which mimics the BM stromal microenvironment.31 In the in vivo study, for the first time, we utilized a PDX model derived from a patient that acquired resistance to venetoclax/ decitabine treatment thus revealing a novel mechanism of synergy involving potential non-apoptotic, metabolic and microenvironmental functions of MCL-1.

resistant AML cells were generated as previously described.27 BAX, BAK, or BAX/BAK double KD OCI-AML3 cells using small interfering RNA (siRNA) (control or ON-TARGETplus SMARTpool siRNA for each target, Dharmacon, Chicago, IL) were generated as previously described.32 Human BM-derived MSC were isolated as previously described.33 Cells, cultured under the conditions previously described34 were treated with venetoclax, AZD5991, AZD4573, IACS-10759 (Institute for Applied Cancer Science, MD Anderson, Houston, TX), BL-8040 (BioLineRx Ltd., Israel), and various combinations without or with MSC (AML-to-MSC=4:1). For AZD4573 treatments, cells were exposed to AZD4573 for 6 hours (h), and then the drug was removed.

Cell viability assay Viable cells and apoptosis were assessed as previously described.34 For leukemia cells co-cultured with MSC, CD45+ cells were counted. Apoptotic cells were defined as annexin V+ and/or 7-aminoactinomycin D+ (AnnV+/7AAD+) CD45+ cells. For patient samples, annexin V positivity was determined in bulk (CD45+), CD34+CD38+/CD38-, and CD34+CD38+/CD38-CD123+ cells.

Western blot analysis and co-immunoprecipitation Western blot analysis was conducted as described previously.34 Antibodies used are shown in the Online Supplementary Appendix.

Protein determination by flow cytometry MCL-1 and cell surface CD44 and CXCR4 were also determined by flow cytometry as shown in the Online Supplementary Appendix.

Adhesion and migration assay Leukemia cells’ migration and adhesion to MSC were assessed as previously described.35

Mitochondrial respiration assay Mitochondrial respiration was measured using a Seahorse XF extracellular flux analyzer (Agilent Technologies, Inc., Santa Clara, CA) following the manufacturer’s instructions. The oxygen consumption rate (OCR) was expressed as pmol/min/1,000 cells or relative to a control.

Reactive oxygen species and glutathione assay Cellular and mitochondrial ROS were determined by flow cytometry after cells were stained with CellROS deep red or MitoSOX red (ThermoFisher Scientific), for 30 minutes (min) at 37°C and expressed as mean fluorescence intensity shift between stained and unstained live cells (AnnV-/DAPI-). Total glutathione and glutathione disulfide (GSSG) were determined using a glutathione (GSH) assay kit (Cayman Chemical; Ann Arbor, MI) following the manufacturer’s instructions. Reduced GSH was calculated by subtracting the amount of GSSG from total GSH.

C -1,2-glucose and 13C -glutamine tracing and ion chromatography-mass spectrometry metabolite analysis 13

Methods

Tracing and subsequence metabolite analysis are detailed in the Online Supplementary Appendix.

Cells and treatments

In vivo experiments

AML cell lines (Molm13, MV4-11, and OCI-AML3) and primary AML cells (Online Supplementary Table S1) were obtained as described in the Online Supplementary Appendix. MCL-1-overexpressing (OE), MCL-1-knockdown (KD), and acquired venetoclax

Mouse care and experiments were performed in accordance with Institution Animal Care and Use Committee approved protocols. See the Online Supplementary Appendix for detailed experimental procedures.

haematologica | 2022; 107(1)

B.Z. Carter et al.

CyTOF mass cytometry Cells were stained with metal-tagged antibodies (Online Supplementary Table S2) against cell surface markers, barcoded, pooled, then stained with metal-tagged antibodies against intracellular proteins and subjected to cytometry by time of flight (CyTOF) analysis as described previously.34,36,37 Briefly, viable single cells (cisplatin-low) were gated with FlowJo (software v10.7, FlowJo LLC) and exported as flow cytometry standard (FCS) data for subsequent analysis in Cytofkit.38 Cell populations identified and embedded by RPhenoGraph in the “Cytofkit_analyzedFCS” files were gated in FlowJo to quantify marker expression. ArcSinh-transformed counts for each protein expression in desired cell populations were visualized with heat maps.

Statistical analysis

Results MCL-1 regulates metabolic functions in acute myeloid leukemia (AML) cell lines and its inhibition decreases AML metabolism MCL-1-KD or -OE did not alter viability of AML cell line Molm13, MV4-11, or OCI-AML3 in vitro (Online Supplementary Figure S1A). However, NSGS mice harboring MCL-1-OE Molm13 cells exhibited higher leukemia burden and succumbed sooner to AML, while mice harboring MCL-1-KD OCI-AML3 cells had decreased leukemia burden and survived longer compared with their respective controls (Online Supplementary Figure S1B), suggesting that MCL-1 levels affect leukemia expansion. In order to determine whether this effect requires the modulation of bioenergetic/metabolic activity, we assayed mitochondrial respiration and found that MCL-1-OE increased, whereas MCL-1-KD or inhibition with AZD5991 (at doses not affecting cell viability) decreased, OCR and ATP production in OCI-AML3 cells (Figure 1A). Compared with parental MV4-11 (MV4-11P), the venetoclax-resistant MV4-11 (MV4-11R) cells that expressed increased MCL-127 had higher OCR and ATP production. In both cell types, AZD5991 inhibited OCR and ATP generation (Online Supplementary Figure S1C). In addition, compared with controls, MCL-1-OE OCI-AML3 cells exhibited higher levels of cellular and mitochondrial ROS, lower GSH levels, and lower GSH/GSSG ratios, whereas MCL-1-KD OCI-AML3 cells displayed lower ROS, higher GSH levels, and higher GSH/GSSG ratios (Figure 1B and C). In order to further explore metabolic functions of MCL1, we traced 13C incorporation from 13C -1,2-glucose and 13 C -glutamine into downstream metabolites40 using an MCL-1-KD clone (shC16) with an approximately 65% MCL-1 reduction (Online Supplementary Figure S1D), selected by limiting dilution of MCL-1-KD OCI-AML3 cells and OCI-AML3 cells treated with 50 nM AZD5991 2

Cell line experiments were performed in triplicates. Results were expressed as the mean ± standard error of the mean (SEM). The Student t-text was used to assess differences between groups; P-values ≤0.05 were considered statistically significant. The combination index (CI),39 determined using Calcusyn software, was expressed as mean CI values at effective dose (ED) , ED , and ED . A CI<1 was considered synergistic; CI=1, additive; and CI>1, antagonistic. EC and half maximal inhibitory concentration (IC ) values were calculated using Calcusyn. Mouse survival was estimated using the Kaplan-Meier method. Survival data were analyzed using the log-rank test. 75

(Online Supplementary Figure S1E). Metabolomic analysis revealed markedly lower overall incorporations of 13C from both glucose and glutamine into key tricarboxylic acid (TCA) cycle intermediates including citrate, fumarate, and malate, in MCL-1-KD and AZD5991-treated OCIAML3 cells compared to their respective controls (Figure 1D). An examination of the patterns of fractional 13C enrichment from 13C -glutamine into citrate, malate, and fumarate demonstrated increased M+4, but decreased M+3 fractional enrichment, in both AZD5991-treated and MCL-1-KD OCI-AML3 cells (Figure 1E), suggesting decreased glutamine entry into the TCA cycle accompanied by a smaller fraction of glutamine-derived carbon being used for anaplerotic reactions that fuel biosynthesis. In both 13C -1,2-glucose and 13C -glutamine tracing experiments, the total amount of lactate secreted by AZD5991treated or MCL-1-KD cells was statistically significantly lower than that secreted by their respective controls (Figure 1F). Notably, statistically significant 13C enrichment (M>0) in the secreted lactate of both the AZD5991-treated and MCL-1-KD OCI-AML3 cells, was observed from 13C 1,2-glucose, but not 13C -glutamine (Figure 1G). Glycolysis of 13C -1,2-glucose generates lactate with two 13C labels (M+2 lactate). Thus, a reduction in M+2 lactate secretion suggests MCL-1 inhibition decreases glycolytic flux. Next, we assessed whether the decrease in glycolytic flux was accompanied by a decrease in flux through the oxidative branch of the pentose phosphate pathway (oxPPP), a major pathway involved in the generation of the reducing equivalent NADPH that is used for ROS neutralization. When 13C -1,2-glucose is metabolized through the oxPPP and re-enters glycolysis via the nonoxPPP, the elimination of one 13C carbon through decarboxylation yields lactate containing only one 13C. The relative oxPPP flux, which we calculated by multiplying the ratio of [M+1]/([M+1]+[M+2]) from extracellular lactate with lactate secretion rate during 13C -1,2-glucose tracing, was lower in both AZD5991-treated and MCL1-KD OCI-AML3 cells than in their respective controls (Figure 1H). 6-Phosphogluconic acid levels were also greatly decreased under MCL-1 inhibition (Figure 1H), in accordance with oxPPP modulation.41 Consistent with the aforementioned extracellular flux assay results, AZD5991treated, and MCL-1-KD OCI-AML3 cells, exhibited decreased ATP levels (Figure 1I). These findings indicated that MCL-1 inhibition downregulates key pathways that support cellular energetics in AML cells as summarized in Figure 1J. Furthermore, MCL-1 modulated the level of cMYC in OCI-AML3 cells (Online Supplementary Figure S1F), which is a well-known metabolic regulator in malignant cells. 5

MCL-1 regulates adhesion molecule expression in acute myeloid leukemia (AML) cell lines and AML-stroma interactions Since MSC in the BM microenvironment upregulate MCL-1, we investigated whether MCL-1 had a role in leukemia-stroma interactions. MCL-1-OE OCI-AML3 cells exhibited increased surface expression of CXCR4 and CD44, both of which participate in leukemia-MSC interactions and drug resistance, and increased OCI-AML3 migration towards, and adhesion to, MSC (Figure 2A). Conversely, MCL-1-KD OCI-AML3 exhibited decreased surface expression of CXCR4 and CD44 and decreased haematologica | 2022; 107(1)

Targeting AML MCL-1 re-sensitizes BCL-2 inhibition

Aii

E F

Figure 1. Continued on following page.

haematologica | 2022; 107(1)

B.Z. Carter et al.

Figure 1. MCL-1 regulates cellular metabolic functions, and genetic or pharmacological inhibition of MCL-1 decreases mitochondrial respiration and metabolism in acute myeloid leukemia cells. (Ai) Seahorse analysis of mitochondrial respiration in MCL-1-overexpressing (MCL-1-OE), MCL-1 knockdown (MCL-1-KD), and (Aii) AZD5991-treated (24 h) OCI-AML3 cells. For MCL-1-OE and –KD cells, bulk populations were used unless indicated otherwise. (B) Cellular reactive oxygen species (CellROS) and mitochondrial ROS (MitoSOX) in MCL-1-OE and MCL-1-KD OCI-AML3 cells. (C) Glutathione (GSH) and GSH/glutathione and glutathione disulfide (GSSG) ratios in MCL-1-OE and MCL-1-KD OCI-AML3 cells. (D to I) OCI-AML3 MCL-1-KD clone (shC16) with an approximately 65% MCL-1 reduction (Online Supplementary Figure S1D) and OCI-AML3 cells treated with 50 nM AZD5991 were used for metabolomics experiments. WT: wild-type untreated control. (D) Relative enrichment of 13 C from 13C -1,2-glucose or 13C -glutamine (M>0) into key tricarboxylic acid (TCA) cycle intermediates. (E) Fractional enrichment of 13C from 13C -glutamine for citrate, fumarate, and malate. (F) Relative lactate levels in medium after 6 hours (h) of incubation with new medium. Lactate levels were determined by summing MS1 ion intensities for all isotopes in the 13C -1,2-glucose (left) and the 13C -glutamine tracer experiment (right). (G) Fractional enrichment in extracellular lactate after 6 h of incubation in the 13C -1,2-glucose (left) and 13C -glutamine (right) tracer experiments. (H and I) Relative pentose phosphate pathway (oxPPP) flux (left) and 6-phosphogluconic acid (6PG) levels (right, H) and relative levels of unlabeled ATP and 13C enrichment into ATP (I). Average WT and shVec M+0 levels were set to a relative level of 1. WT: parental OCI-AML3 cells; AZD5991: AZD5991-treated OCI-AML3 cells; shVec: vector control OCI-AML3 cells; shC16: MCL-1-KD OCI-AML3 cells. *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001. Experiments were performed in triplicates (for seahorse analysis: duplicates for each experiment). (J) Schematic illustration of carbohydrate metabolic pathways decreased by MCL-1 inhibition. Cit: citrate; Fum: fumarate; Mal: malate; aKG: a-ketoglutarate; Oaa: oxaloacetic acid; MFI: mean fluorescence intensity. 2

OCI-AML3 migration toward, and adhesion to, MSC (Figure 2B). Consistently, inhibition of MCL-1 with AZD5991 decreased OCI-AML3 surface CXCR4 and CD44 expression and interaction with MSC (Figure 2C). Like other MCL-1 inhibitors, AZD5991 increased levels of MCL-1 (Figure 2C). BCL-2 inhibition with venetoclax in 62

OCI-AML3 cells did not appreciably decrease CD44 surface expression, AML cell migration, or adhesion to MSC. However, it decreased CXCR4 surface expression (Figure 2D). These results suggest that MCL-1 enhances surface adhesion molecule expression and promotes leukemiastroma interactions. haematologica | 2022; 107(1)

Targeting AML MCL-1 re-sensitizes BCL-2 inhibition

Figure 2. Continued on following page.

haematologica | 2022; 107(1)

B.Z. Carter et al.

Figure 2. MCL-1 regulates cell surface expression levels of CXCR4 and CD44 in acute myeloid leukemia cells and leukemia-stroma interactions. (A and B) Cell surface CXCR4 and CD44 levels and migration and adhesion of vector control and MCL-1-overexpressing (MCL1-OE), OCI-AML3 cells (A) or vector control and MCL-1-knockdown (MCL-1-KD) OCI-AML3 cells (B) to mesenchymal stromal cell (MSC). Blue bars represent the migration of OCI-AML3 cells to CXCL12 (positive control) or randomly (negative control). (C and D) Cell surface CXCR4 and CD44 levels and migration and adhesion of OCI-AML3 cells treated with AZD5991 (C) or venetoclax (D) for 24 hours (h) to MSC. Cell surface CXCR4 and CD44 and cellular MCL-1 levels were determined by flow cytometry. Migration was performed for 6 h and adhesion for 24 h. Experiments were performed in triplicates. Results are expressed as the mean ± standard error of the mean. IgG: immunoglobulin G; MFI: mean fluorescent intensity.

haematologica | 2022; 107(1)

Targeting AML MCL-1 re-sensitizes BCL-2 inhibition

MCL-1 mediates venetoclax resistance, and its inhibition sensitizes acute myeloid leukemia cell lines to BCL-2 inhibition Molm13 and MV4-11 cells are relatively sensitive, whereas OCI-AML3 cells are resistant, to venetoclax (Online Supplementary Figure S2A). Acquired venetoclaxresistant AML cell lines have increased MCL-1, but they likely also have altered levels of many other proteins. In order to determine whether MCL-1 played a key role in venetoclax resistance and whether co-targeting BCL-2 and MCL-1 overcomes this resistance, we treated MCL-

1-KD OCI-AML3 and MCL-1-OE Molm13 and MV4-11 cells with venetoclax or AZD5991. Whereas MCL-1-KD greatly sensitized OCI-AML3 cells to venetoclax (Figure 3A), MCL-1-OE Molm13 (Figure 3B) and MV4-11 cells (Online Supplementary Figure S2B) were largely insensitive to the drug. Responses to AZD5991 were similar, though to a lesser degree (Figure 3A and B; Online Supplementary Figure S2B). We then treated OCI-AML3 and MCL-1-OE Molm13 and MV4-11 cells with venetoclax plus AZD5991. The combination synergistically induced apoptosis and decreased viable cells in the intrinsically

Figure 3. Continued on following page.

haematologica | 2022; 107(1)

B.Z. Carter et al.

Figure 3. MCL-1 expression contributes to venetoclax resistance, and the combined inhibition of BCL2 and MCL-1 synergistically induces apoptosis in acute myeloid leukemia cells with intrinsic or acquired resistance to venetoclax. (A and B). Control (con) and MCL-1 knockdown (MCL-1-KD) OCI-AML3 cells (A) or control and MCL-1 overexpressing (MCL-1-OE) Molm13 cells (B) were treated with venetoclax (VEN) or AZD5991 for 48 hours (h). (C to E) OCI-AML3 (C), MCL-1-OE Molm13 (D), and control parental MV4-11 cells (MV4-11P) or venetoclax-resistant MV4-11 cells (MV411R) (E) were treated with venetoclax and/or AZD5991 for the indicated times. Apoptosis and viable cell numbers were assessed by flow cytometry. Experiments were performed in triplicates. Results are expressed as the mean ± standard error of the mean. IC : half maximal inhibitory concentration; EC : half maximal effective concentration; CI: combination index; OE: overexpressing; KD: knockdown. 50

haematologica | 2022; 107(1)

Targeting AML MCL-1 re-sensitizes BCL-2 inhibition

venetoclax-resistant OCI-AML3 (Figure 3C), MCL-1-OE Molm13 (Figure 3D), and MV4-11 cells (Online Supplementary Figure S2C). This was also the case in MV4-11R cells (Figure 3E). We then tested CDK9 inhibitor AZD4573. AZD4573 decreased MCL-1 expression and synergized with venetoclax targeting OCI-AML3 cells (Online Supplementary Figure S2D). This combination was also highly synergistic against MV4-11 and MV4-11R cells (Online Supplementary Figure S2E). In order to demonstrate that MCL-1 is indeed the critical target regulated by CDK9, we treated MCL-1OE Molm13 and MV4-11 cells with AZD4573 and found that, similar to the MCL-1 inhibitor AZD5991, MCL-1 OE cells were more resistant than control cells to AZD4573. Furthermore, the combination of venetoclax and AZD4573 was highly synergistic in MCL-1-OE Molm13 and MV4-11 cells (Online Supplementary Figure S2F). We next treated OCI-AML3 cells with venetoclax, AZD5991, or the combination and determined interactions between anti-apoptotic BCL-2, MCL-1, and BCL-XL with the pro-apoptotic activator BIM and effectors BAX and BAK by co-immunoprecipitation to potentially understand the mechanisms of synergy. We observed that BCL2 was bound to BIM and BAX (Figure 4A), MCL-1 was bound to BIM and BAK (Figure 4B), and BCL-XL was bound to all three (Figure 4C). Venetoclax treatment decreased BCL-2/BIM and BCL2/BAX interactions, but increased MCL-1/BIM, BCL-XL/BAX, and BCL-XL/BAK interactions. AZD5991 treatment decreased MCL-1/BIM interaction, but increased BCL-2/BIM and MCL-1/BAK interactions. When the two drugs were combined, BCL2/BIM and BCL2/BAX interactions were largely diminished, MCL-1/BIM interaction decreased, and the single agent treatment-mediated increases of BCL-XL/BAX, BCL-XL/BAK, and MCL-1/BAK interactions were abrogated (Figure 4A to C), suggesting that the combinatorial cooperative release of activator and effector BCL-2 family proteins likely contributed to the synergy in apoptosis induction (Figure 4D). In order to determine the contribution of metabolism and leukemia-stroma interactions on the observed synergism, we treated OCI-AML3 cells with venetoclax in the presence of the OxPhos inhibitor IACS-1075942 or the CXCR4 inhibitor BL-8040 (4F-benzoyl-TN14003).43 As expected, MSC co-culture protected AML cells from venetoclax-, AZD5991-, or IACS-10759-induced apoptosis (Figure 4E). While suppressing OCI-AML3 migration and adhesion to MSC, BL-8040 did not affect cell viability (Figure 4E and F). Under MSC co-culture, venetoclax activity was minimally enhanced by OxPhos or CXCR4 inhibition, but markedly augmented by combinatorial OxPhos and CXCR4 inhibition. Maximal apoptosis induction of OCI-AML3 was observed by combined MCL-1 and BCL2 inhibition (Figure 4E). These results support that cooperative release of pro- from anti-apoptotic BCL-2 family proteins and inhibition of cell metabolism and key stromal microenvironmental mechanisms all contributed to the synergism of co-targeting MCL-1 and BCL-2 in AML cells.

Combined BCL-2 and MCL-1 inhibition synergistically induces apoptosis in primary acute myeloid leukemia (AML) cells and AML stem/progenitor cells Primary AML cells were cultured with BM-derived MSC and treated with venetoclax, AZD5991, AZD4573, venetoclax plus AZD5991, or venetoclax plus AZD4573. haematologica | 2022; 107(1)

Apoptosis and viable cell counts of AML blast cells and CD34+ stem/progenitor cells were assessed and expressed as EC (Figure 5A) or IC (Online Supplementary Figure S3) for each agent used alone or in combination, as indicated. The patient characteristics, including mutations, venetoclax treatment status, and cytogenetics/risk categories, are shown in Figure 5B and Online Supplementary Table S1. Responses to venetoclax, AZD5991, or AZD4573 varied across the patient samples. Compared with the single-agent treatments, the venetoclax plus AZD5991 and venetoclax plus AZD4573 combinations were markedly more effective in inducing apoptosis and decreasing viable cell numbers (markedly lower EC and IC values, respectively) in not only blasts, but also CD34+, CD34+CD38+/CD34+CD38- and CD34+CD38+CD123+/CD34+CD38-CD123+ stem/progenitor cells in all primary AML samples, including those from venetoclax-resistant/relapsed patients (Figure 5A and B; Online Supplementary Figure S3; Online Supplementary Table S1) regardless of mutational status and cytogenetics. Among samples 7, 8, 11, and 12 from venetoclax or venetoclax/decitabine resistant/relapsed patients, three (7, 8, and 12) had relatively high venetoclax EC values. Samples 3 and 10 that exhibited relatively high venetoclax EC values were from resistant/relapsed patients who received venetoclax after sampling and were resistant or relapsed (Figure 5A and B; Online Supplementary Table S1), suggesting that our in vitro system mirrored clinical responses. Patient samples with high venetoclax EC values tended to have high AZD5991 EC values, suggesting that AZD5991 monotherapy would be less effective in venetoclax-resistant patients. However, more samples are needed to confirm our conclusion. We observed that samples with WT1 (3 and 12) or BCORL1 (4 and 7) mutations were the most resistant to AZD5991 and AZD4573, respectively (Figure 5A and B). Experiments with primary AML cells treated without MSC co-culture yielded similar results (Online Supplementary Figure S4), but these cells were generally more sensitive to the treatments supporting the protective role of MSC. The CI values for the combinations for each AML cell population are shown in the Online Supplementary Table S3. 50

Co-targeting BCL-2 and MCL-1 exerts pronounced anti-leukemia activity in a patient-derived xenograft model of clinical venetoclax/decitabine-relapsed acute myeloid leukemia In order to determine whether co-inhibition of BCL-2 and MCL-1 could overcome venetoclax resistance in vivo, we developed an AML PDX model using cells from a patient who initially responded to venetoclax/decitabine therapy, but relapsed after three cycles (Figure 6A; Online Supplementary Table S1). PDX-bearing mice were treated with venetoclax, AZD5991, AZD4573, venetoclax plus AZD5991, or venetoclax plus AZD4573 (Figure 6A). All treatments statistically significantly decreased circulating blasts compared to controls (P≤0.0001, treatment day [d] 18). The venetoclax plus AZD5991 or venetoclax plus AZD4573 group had statistically significantly fewer circulating blasts than the venetoclax (P<0.01) and AZD5991 (P<0.0001) groups or the venetoclax (P=0.0001) and AZD4573 (P<0.001) groups, respectively (Figure 6B). Analyses of BM samples (treatment d 25) yielded similar results (Figure 6C). Spleens from AZD4573- or venetoclax plus AZD4573-treated mice had a statistically significantly lower leukemia burden compared to controls, which were 67

B.Z. Carter et al.

F E

Figure 4. Legend on following page.

haematologica | 2022; 107(1)

Targeting AML MCL-1 re-sensitizes BCL-2 inhibition

Figure 4. Mechanisms of synergy. Interactions of anti-apoptotic and pro-apoptotic BCL-2 proteins in control and BH3 mimetic-treated cells (A to D). OCI-AML3 cells were treated with venetoclax, AZD5991, or both (dose 1: venetoclax, 100 nM and AZD5991, 10 nM; dose 2: venetoclax, 250 nM and AZD5991, 25 nM) for 24 hours (h). (A to C) Interactions of BCL-2 (A), MCL-1 (B), or BCL-XL (C) with BIM, BAX, or BAK were determined by co-immunoprecipitation (co-IP) and western blot analysis. (D) Summary of the interactions. Roles of metabolic function and leukemia-stroma interactions (E and F). (E) OCI-AML3 cells were treated with venetoclax alone or in combination with IACS-10759, BL-8040, or AZD5991 with or without mesenchymal stromal cell (MSC) co-cultures for 48 h. Cell death in CD45+ cells were determined by flow cytometry. (F) Migration (6 h) and adhesion (24 h) of BL-8040 treated-OCI-AML3 cells to MSC were measured. Migration to CXCL12 was used as a positive control and random migration as a negative control. Experiments were performed in triplicates. Results are expressed as the mean ± standard error of the mean. IgG: immunoglobulin G.

similar for spleens from the mice treated with venetoclax plus AZD5991 compared to the untreated or AZD5991treated groups (treatment d 25, Figure 6D). Although all the treatments statistically significantly extended survival in the PDX model, venetoclax (54 d, P=0.0035), as expected, and AZD5991 (53 d, P=0.0081) showed minimal effects similar to that presented for the in vitro data. AZD4573, and to greater degrees the combinations, were more effective (AZD4573: 60 d, P=0.0002; venetoclax plus AZD4573: 71 d, P=0.0005; venetoclax plus AZD5991: 76.5 d, P=0.0005 compared to controls 51 d, Figure 6E). The median survival of the venetoclax plus AZD4573 treatment group was statistically significantly longer than that of the venetoclax or AZD4573 group (P=0.0004 for both), and the venetoclax plus AZD5991 group survival was statistically significantly longer than that of venetoclax (P=0.0004) or AZD5991 group (P=0.0012) (Figure 6E). No obvious weight loss was observed during the various treatments. The mice only began to lose weight several days before they succumbed to the disease (Online Supplementary Figure S5). Of the 60 mice, one in AZD5991 group, one in venetoclax plus AZD5991 group, and two in venetoclax plus AZD4573 group died of treatment procedures and were not included in the survival analyses. In order to elucidate the treatment effects on phenotypically-defined cell populations and on protein expression in these populations, we performed CyTOF analysis on BM samples (treatment d 25). Cells were clustered based on expressions of cell surface markers. While venetoclax had no, and AZD5991 and AZD4573 had minimal, effects in several AML cell populations, venetoclax plus AZD4573 and, to a greater degree, venetoclax plus AZD5991 had profound anti-leukemia activity in all cell populations including AML stem/progenitor cells (Figure 6F). Compared to the vehicle treated controls, the singleagent treatments increased BCL-2, MCL-1, and c-MYC levels in most of the AML cell populations, whereas the combinations, particularly venetoclax plus AZD5991, decreased c-MYC in all populations, decreased BCL-2 in all but the CD34+CD38- population, decreased MCL-1 in the CD45+, CD34+CD38+CD123+, and CD34+CD38CD123+ populations (Figure 7A). Interestingly, AZD5991, AZD4573, and, to a greater degree, venetoclax plus AZD5991 and venetoclax plus AZD4573 greatly decreased cell surface CXCR4 levels in all cell populations. AZD5991 and AZD4573 also decreased CD44 levels in all populations (Figure 7B). These results paralleled our in vitro findings suggesting that the inhibition of MCL-1 decreases CXCR4 and CD44 expression to suppress leukemia-stroma interactions. We also measured several signaling proteins and found that the combination treatments greatly decreased b-CATENIN and p-AKT in several AML cell populations (Online Supplementary Figure S6A), particularly in CD45+ and CD34+CD38-CD123+ cells (Figure 7C). Because mouse BM samples were collected 1, 3, or 2 d after venetoclax, AZD5991, or AZD4573 administration, haematologica | 2022; 107(1)

our analyses may underestimate the treatment effects on proteins/phosphor-proteins in leukemia cells. We also performed CyTOF analysis on BM cells collected from moribund mice. Compared with those in controls, BCL-2, MCL-1, and c-MYC levels in all leukemia cell populations, with the exception of c-MYC in CD34+CD38+ cells, were higher in venetoclax, AZD5991, and the combination treatment groups (Figure 7D). Cell signaling protein analysis (Online Supplementary Figure S6B) showed that in all but the AZD4573 treatment CXCR4 was suppressed (Figure 7E), whereas p-AKT was greatly induced especially in the combination treatment groups (Figure 7F). Although p-AKT was not induced in the AZD4573 group, the CXCR4 level recovered compared with that observed in the controls (Figure 7E and F).

MCL-1 regulates metabolic activity in acute myeloid leukemia cells and leukemia-stromal interactions In order to determine if MCL-1 regulates AML cell metabolic activity and leukemia-stromal interactions independent of its anti-apoptotic functions, we knocked down BAX, BAK, or both with siRNA in OCI-AML3 cells. Suppression of BAX and BAK levels were confirmed by western blot 48 h after siRNA transfection (Figure 8A). siRNA transfected cells were then treated with AZD5991 (50 nM), a dose that did not affect viable cell count (Online Supplementary Figure S1E). Apoptosis, mitochondrial respiration, and cell migration/adhesion to MSC were determined. As shown in Figure 8B, at the 50 nM dose of AZD5991, control siRNA-transfected cells showed no increase in apoptosis at 1 and 4 h, and only 5% more apoptotic cells over baseline at 24 h. The Bax, Bak, and Bax plus Bak siRNA-transfected cells were more resistant to AZD5991-induced apoptosis, as expected. However, inhibition of MCL-1 demonstrated the identical pattern of inhibition of mitochondrial respiration, detectable at 1 h after AZD5991 treatment, in all siRNA-transfected cells (Figure 8B). Furthermore, although AZD5991 treatment (24 h) induced low levels of apoptosis in control siRNAtransfected, but not in Bax and/or Bak siRNA-transfected OCI-AML3 cells, no apparent differences were observed in OCI-AML3 cell migration and adhesion inhibition to MSC in control, Bax, and/or Bak siRNA-transfected and AZD5991 treated cells (Figure 8C). These results suggest that MCL-1 regulates AML metabolic activity and leukemia-stromal interactions independent of its functions as an apoptosis regulator. Finally, we treated AML patient samples (n=4) with low dose venetoclax (2.5 nM) or AZD5991 (12.5 nM) that induced low level apoptosis after 24 h (Online Supplementary Figure S7A) and found that AZD5991, but not venetoclax, effectively inhibited cellular and mitochondrial ROS production, reduced CXCR4 and CD44 levels, and diminished cell migration and adhesion to MSC (Online Supplementary Figure S7B to D), further supporting that MCL-1 regulates mitochondrial respiration and leukemia-stromal-interactions in AML. 69

B.Z. Carter et al.

Figure 5. Combined inhibition of BCL-2 and MCL-1 synergistically induces apoptosis in primary acute myeloid leukemia (AML) cells and AML stem/progenitor cells under mesenchymal stromal cell co-culture conditions. Primary acute myeloid leukemia (AML) cells co-cultured with mesenchymal stromal cellc (MSC) were treated with venetoclax (VEN), AZD5991, AZD4573, VEN plus AZD5991, or VEN plus AZD4573 for 48 hours (h). The apoptosis of blast cells and stem/progenitor cells were assessed by flow cytometry. (A) Half maximal effective concentration (EC ) values of venetoclax alone or combined with AZD5991 or AZD4573, of AZD5991 alone or combined with venetoclax and of AZD4573 alone or combined with venetoclax in blast cells and stem/progenitor cells from individual patient samples are shown. Samples lacking the indicated cell populations or that had excessively high spontaneous apoptosis are not represented. *>3,000 nM. (B) Mutations, venetoclax treatment status, and cytogenetics/risk category of each patient from whom primary AML samples were obtained. HR: complex cytogenetics and high risk; IR: intermediated risk. 50

haematologica | 2022; 107(1)

Targeting AML MCL-1 re-sensitizes BCL-2 inhibition

Discussion Here, we demonstrate that co-targeting BCL-2 and MCL-1 is highly effective in AML cell lines and primary patient samples, which is consistent with recent reports.23,24,26,29 Importantly, combined MCL-1 and BCL-2 inhibition was highly synergistic against AML stem/progenitor cells and exerted pronounced anti-leukemia activity in venetoclax-resistant AML patient samples in vitro, even under conditions mimicking the BM stromal microenvironment. Furthermore, similar findings were obtained in a unique PDX model of clinical venetoclax/decitabinerelapsed AML. Hence, the combination is not just more effective than each agent given alone, but has the potential to overcome intrinsic and acquired venetoclax resistance. In exploring the mechanisms underlying the combination’s efficacy, we demonstrated that targeting MCL-1 sensitizes to BCL-2 inhibition in AML through not only

cooperative release of pro-apoptotic from binding to antiapoptotic BCL-2 proteins but also inhibition of cell metabolism and key stromal microenvironmental mechanisms. The metabolic and stromal functions of MCL-1 can apparently act independently of the protein’s anti-apoptotic activity. These findings provide rationale for the clinical development of combined BCL-2 and MCL-1 targeting in resistant/relapsed AML patients, particularly in patients who were resistant/relapsed from venetoclax-based therapies. We observed that WT1- or BCORL1-mutated primary AML samples were particularly resistant to AZD5991 and AZD4573, respectively. WT1 mutation in AML is associated with poor outcomes and chemoresistance,44 and BCORL1 encodes a transcription co-repressor, opposite to CDK9. We speculate that BCORL1-mutated cells are less transcriptionally suppressed, and therefore more resistant to CDK9 inhibition. However, the sample size for either

Figure 6. Continued on following page.

haematologica | 2022; 107(1)

B.Z. Carter et al.

Figure 6. Combined inhibition of BCL-2 and MCL-1 has strong anti-leukemia activity and prolongs survival in a patient-derived xenograft model of acquired venetoclax/decitabine-resistant acute myeloid leukemia. (A) Patient-derived xenograft (PDX) model and experimental scheme. (B) Circulating human CD45+ (hCD45+) cells in mouse peripheral blood (PB) samples obtained from each treatment group on treatment day (d) 18. (C) hCD45+ cells in mouse bone marrow (BM) samples obtained from each treatment group on treatment d 25. (D) hCD45+ cells in mouse spleens and spleen sizes of each treatment group on treatment d 25. hCD45+ cells were measured by flow cytometry. (E) Survival curves. (F) BM leukemia cell clusters based on the expression of cell surface markers and viable leukemic cells in various cell populations on treatment d 25 in each treatment group as determined by cytometry by time of flight (CyTOF). *P≤0.05; **P≤0.01; ***P≤0.001; ****P≤0.0001. Results are expressed as the mean ± standard error of the mean. Survival data were analyzed using the log-rank test. Con: control; VEN: venetoclax; 5991: AZD5991; 4573: AZD4573.

WT1- or BCORL1-mutated cells was too small to draw definitive conclusions. Additional studies are needed to confirm our finding, which may help guide patient selection. In contrast to a recent study showing that venetoclaxresistant cells are more sensitive to the MCL-1 inhibitor VU0661013,30 we found that intrinsic or acquired venetoclax resistant AML cells were less responsive to the MCL1 inhibitor AZD5991, but the combination of venetoclax and AZD5991 or AZD4573 synergistically induced cell death regardless of the single-agent response, both in vitro and in vivo. These data suggest that patients whose disease is resistant to, or relapses from, venetoclax therapy may 72

potentially benefit from co-targeting MCL-1 and BCL-2. In order to understand the mechanisms of synergy, especially in cells insensitive to BCL-2 or MCL-1 inhibition alone, we examined the interactions of anti- and proapoptotic BCL-2 proteins in OCI-AML3 cells treated with venetoclax, AZD5991, or both. Targeting either BCL-2 or MCL-1 decreased the interactions of one of them with pro-apoptotic BCL-2 proteins, but increased the interactions of the other with pro-apoptotic proteins. Only when both are inhibited, maximal amounts of unbound activator protein BIM and effector proteins BAX/BAK become available, and more effectively induce apoptosis. Interestingly, Pollyea et al. recently attributed the effechaematologica | 2022; 107(1)

Targeting AML MCL-1 re-sensitizes BCL-2 inhibition

Figure 7. Legend on following page.

haematologica | 2022; 107(1)

B.Z. Carter et al.

Figure 7. Protein levels in various leukemia cell populations in mouse bone marrow as determined by CyTOF. Bone marrow (BM) cells were collected from the mice on day (d) 25 of treatment (A to C) or from moribund mice (D to F) from each treatment group (n=2 or 3 mice/group). (A and D). BCL-2, MCL-1, and c-MYC levels in various leukemia cell populations in each treatment group. (B and E) Cell surface expression levels of CXCR4 and CD44 (B) or CXCR4 (E) in various cell populations in each treatment group. (C) β-CATENIN and p-AKT levels in CD45+ and CD34+CD38-CD123+ cells. (F) p-AKT levels in various cell populations. The protein levels of individual samples are presented as heat maps. After stained with antibodies against cell surface markers, all BM samples collected at d 25 treatment were barcoded, pooled into the same tube, stained, and run concomitantly and all BM samples collected from moribund mice were barcoded, pooled into the same tube, stained, and run concomitantly.

tiveness of the venetoclax plus azacitidine combination in its ability to suppress OxPhos, disrupt the TCA cycle, and perturb cell energy metabolism, thereby efficiently targeting leukemia stem cells, an effect that was not achieved with the venetoclax treatment alone.7 We found that MCL-1 regulates redox and metabolic functions in AML cells and that genetic or pharmacological inhibition of MCL-1 suppressed several cellular energetic and metabolic

pathways, including TCA cycle, glycolysis, and PPP. We further demonstrated that inhibiting this function of MCL-1 contributed to the enhanced activity of venetoclax against AML cells. ROS stabilize HIF1 thereby activating hypoxia signaling.45,46 Although MCL-1 alteration in AML cells did not elicit detectable changes in HIF1a levels, it elicited changes in CXCR4, a HIF1a target, and CD44. Both, the

Figure 8. Effects of BAX and/or BAK knockdown on apoptosis, mitochondrial respiration, and migration/adhesion to mesenchymal stromal cell of acute myeloid leukemia cells in responses to MCL-1 inhibition. OCI-AML3 cells were transfected by electroporation with control, Bax, Bak, or Bax and Bak small interfering RNA (siRNA) (4 mM) for 48 hours (h) and then treated with AZD5991 (50 nM) for 24 h. (A) BAX and BAK levels in siRNA-transfected OCI-AML3 cells at 48 h, determined by western blot. The left panel is the result of a representative experiment and the right is the quantitative results of three independent experiments. (B) Apoptosis by flow cytometry and mitochondrial respiration by Seahorse of siRNA-transfected OCI-AML3 cells treated with AZD5991 at various time points. (C) Apoptosis and migration (6 h) and adhesion (24 h) of siRNA-transfected OCI-AML3 cells treated with AZD5991 for 24 h compared with untreated controls. The experiments were performed in triplicates (with duplicates for each experiment in the Seahorse experiment). The results are expressed as mean ± standard error of the mean. AnnV: annexin V; OCR: oxygen consumption rate. MSC: mesenchymal stromal cells.

haematologica | 2022; 107(1)

Targeting AML MCL-1 re-sensitizes BCL-2 inhibition

CXCR4/CXCL12 axis and CD44 are critical for leukemiaBM stromal microenvironment interactions. Indeed, we found that genetic or pharmacological manipulation of MCL-1, but not BCL-2, in AML cells altered cell migration and adhesion to MSC. In confirmation, in vivo MCL-1, but not BCL-2, inhibition decreased CXCR4 and CD44 levels in BM leukemia cells from PDX mice suggesting a novel function of MCL-1, specifically the regulation of leukemia-stroma interactions. The cell intrinsic anti-apoptotic function of MCL-1 may therefore be complemented and enhanced by the cell-extrinsic enhanced adhesion to the BM stroma. In addition to MCL-1, other mechanisms of intrinsic and acquired venetoclax resistance were reported in recent years.47-50 Although highly effective and statistically significantly extending survival, our combination-treated mice eventually died of leukemia. We observed increased pAKT levels in leukemia cells collected from these mice, which warrants further investigation. Collectively, we demonstrated that MCL-1 regulates cell metabolism, leukemia-stroma interactions, and protects leukemia cells from BCL-2 inhibition. MCL-1 inhibition targets multiple cancer cell characteristics and therefore has multifaceted effects in AML. The MCL-1 inhibitionmediated suppression of metabolic activity and inhibition of CXCR4 and CD44 may contribute to its efficacy against AML stem cells in the BM stromal microenvironment. Treatment strategies involving combined MCL-1 and BCL-2 inhibition could improve the outcomes in AML patients for whom BCL-2-targeted therapy has failed, which warrants further clinical evaluation.

References 1. Konopleva M, Contractor R, Tsao T, et al. Mechanisms of apoptosis sensitivity and resistance to the BH3 mimetic ABT-737 in acute myeloid leukemia. Cancer Cell. 2006;10(5):375-388. 2. Pan R, Hogdal LJ, Benito JM, et al. Selective BCL-2 inhibition by ABT-199 causes on-target cell death in acute myeloid leukemia. Cancer Discov. 2014;4(3):362-375. 3. Schoenwaelder SM, Jarman KE, Gardiner EE, et al. Bcl-xL-inhibitory BH3 mimetics can induce a transient thrombocytopathy that undermines the hemostatic function of platelets. Blood. 2011;118(6):1663-1674. 4. Konopleva M, Pollyea DA, Potluri J, et al. Efficacy and biological correlates of response in a phase II study of venetoclax monotherapy in patients with acute myelogenous leukemia. Cancer Discov. 2016;6(10):11061117. 5. DiNardo CD, Pratz K, Pullarkat V, et al. Venetoclax combined with decitabine or azacitidine in treatment-naive, elderly patients with acute myeloid leukemia. Blood. 2019;133(1):7-17. 6. DiNardo CD, Pratz KW, Letai A, et al. Safety and preliminary efficacy of venetoclax with decitabine or azacitidine in elderly patients with previously untreated acute myeloid leukaemia: a non-randomised, open-label, phase 1b study. Lancet Oncol. 2018;19(2):216-228. 7. Pollyea DA, Stevens BM, Jones CL, et al. Venetoclax with azacitidine disrupts energy metabolism and targets leukemia stem cells in patients with acute myeloid leukemia. Nat Med. 2018;24(12):1859-1866.

haematologica | 2022; 107(1)

Disclosures BZC and MA received research funding from AstraZeneca; JC and LD are employees of AstraZeneca. Contributions BZC conceptualized the study, designed the experiments, and wrote the manuscript; PYM and WT performed the experiments and analyzed the data; MW performed the experiments, analyzed the data, and wrote the paper; PLL analyzed the data and edited the paper; DM, VR, and LT performed experiments; JC and LD provided materials, supported study, and edited the paper; MA contributed to the concept development and the experimental design and edited the manuscript. Acknowledgments We thank Joe Munch and Numsen Hail for editing the manuscript, Natalia Baran for assisting with mitochondrial respiration experiments, and Munazza Noor and Jairo Matthews for providing patient clinical information. Funding This work was supported in part by research funding from AstraZeneca (to BZC and MA); and by the Paul and Mary Haas Chair in Genetics (to MA). This work used MD Anderson Cancer Center Flow Cytometry and Cell Imaging, Research Animal Support, Metabolomics, and Characterized Cell Line Core Facilities, supported by the National Institutes of Health Cancer Center Support Grant (P30CA016672). The Metabolomics Core is additionally supported by CPRIT grant RP130397.

8. De Blasio A, Vento R, Di Fiore R. Mcl-1 targeting could be an intriguing perspective to cure cancer. J Cell Physiol. 2018;233(11):8482-8498. 9. Adams JM, Cory S. The BCL-2 arbiters of apoptosis and their growing role as cancer targets. Cell Death Differ. 2018;25(1):27-36. 10. Glaser SP, Lee EF, Trounson E, et al. Antiapoptotic Mcl-1 is essential for the development and sustained growth of acute myeloid leukemia. Genes Dev. 2012;26(2): 120-125. 11. Campbell CJ, Lee JB, Levadoux-Martin M, et al. The human stem cell hierarchy is defined by a functional dependence on Mcl-1 for self-renewal capacity. Blood. 2010;116(9): 1433-1442. 12. Pan R, Ruvolo VR, Wei J, et al. Inhibition of Mcl-1 with the pan-Bcl-2 family inhibitor ()BI97D6 overcomes ABT-737 resistance in acute myeloid leukemia. Blood. 2015;126(3): 363-372. 13. Yoshimoto G, Miyamoto T, JabbarzadehTabrizi S, et al. FLT3-ITD up-regulates MCL1 to promote survival of stem cells in acute myeloid leukemia via FLT3-ITD-specific STAT5 activation. Blood. 2009;114 (24):5034-5043. 14. Kaufmann SH, Karp JE, Svingen PA, et al. Elevated expression of the apoptotic regulator Mcl-1 at the time of leukemic relapse. Blood. 1998;91(3):991-1000. 15. Bose P, Gandhi V, Konopleva M. Pathways and mechanisms of venetoclax resistance. Leuk Lymphoma. 2017;58(9):1-17. 16. Niu X, Zhao J, Ma J, et al. Binding of released Bim to Mcl-1 is a mechanism of intrinsic resistance to ABT-199 which can be overcome by combination with daunorubicin or

cytarabine in AML cells. Clin Cancer Res. 2016;22(17):4440-4451. 17. Huang H, Shah K, Bradbury NA, Li C, White C. Mcl-1 promotes lung cancer cell migration by directly interacting with VDAC to increase mitochondrial Ca2+ uptake and reactive oxygen species generation. Cell Death Dis. 2014;5:(10)e1482. 18. Lee KM, Giltnane JM, Balko JM, et al. MYC and MCL1 cooperatively promote chemotherapy-resistant breast cancer stem cells via regulation of mitochondrial oxidative phosphorylation. Cell Metab. 2017;26(4):633-647. 19. Farge T, Saland E, de Toni F, et al. Chemotherapy-resistant human acute myeloid leukemia cells are not enriched for leukemic stem cells but require oxidative metabolism. Cancer Discov. 2017;7(7):716735. 20. Liyanage SU, Hurren R, Voisin V, et al. Leveraging increased cytoplasmic nucleoside kinase activity to target mtDNA and oxidative phosphorylation in AML. Blood. 2017;129(19):2657-2666. 21. Testa U, Labbaye C, Castelli G, Pelosi E. Oxidative stress and hypoxia in normal and leukemic stem cells. Exp Hematol. 2016;44(7):540-560. 22. Kotschy A, Szlavik Z, Murray J, et al. The MCL1 inhibitor S63845 is tolerable and effective in diverse cancer models. Nature. 2016;538(7626):477-482. 23. Tron AE, Belmonte MA, Adam A, et al. Discovery of Mcl-1-specific inhibitor AZD5991 and preclinical activity in multiple myeloma and acute myeloid leukemia. Nat Commun. 2018;9(1):5341. 24. Caenepeel S, Brown SP, Belmontes B, et al.

B.Z. Carter et al. AMG 176, a selective MCL1 inhibitor, is effective in hematologic cancer models alone and in combination with established therapies. Cancer Discov. 2018;8(12):1582-1597. 25. Cidado J, Boiko S, Proia T, et al. AZD4573 is a highly selective CDK9 inhibitor that suppresses Mcl-1 and induces apoptosis in hematological cancer cells. Clin Cancer Res. 2019;26(4):922-934. 26. Teh TC, Nguyen NY, Moujalled DM, et al. Enhancing venetoclax activity in acute myeloid leukemia by co-targeting MCL1. Leukemia. 2018;32(2):303-312. 27. Pan R, Ruvolo V, Mu H, et al. Synthetic lethality of combined Bcl-2 inhibition and p53 activation in AML: mechanisms and superior antileukemic efficacy. Cancer Cell. 2017;32(6):748-760. 28. Daver NG, Garcia JS, Jonas BA, et al. Updated results from the venetoclax (Ven) in combination with idasanutlin (Idasa) arm of a phase 1b trial in elderly patients (Pts) with relapsed or refractory (R/R) AML ineligible for cytotoxic chemotherapy. Blood. 2019;134(Suppl 1):S229-229. 29. Moujalled DM, Pomilio G, Ghiurau C, et al. Combining BH3-mimetics to target both BCL-2 and MCL1 has potent activity in preclinical models of acute myeloid leukemia. Leukemia. 2019;33(4):905-917. 30. Ramsey HE, Fischer MA, Lee T, et al. A novel MCL1 inhibitor combined with venetoclax rescues venetoclax-resistant acaute myelogenous leukemia. Cancer Discov. 2018;8(12):1566-1581. 31. Konopleva M, Andreeff M. Targeting the leukemia microenvironment. Curr Drug Targets. 2007;8(6):685-701. 32. Carter BZ, Mak PY, Wang X, et al. An ARCregulated IL1b/Cox-2/PGE2/b-Catenin/ARC circuit controls leukemia-microenvironment interactions and confers drug resistance in AML. Cancer Res. 2019;79(6):1165-1177. 33. Studeny M, Marini FC, Champlin RE,

Zompetta C, Fidler IJ, Andreeff M. Bone marrow-derived mesenchymal stem cells as vehicles for interferon-beta delivery into tumors. Cancer Res. 2002;62(13):3603-3608. 34. Carter BZ, Mak PY, Mu H, et al. Combined targeting of BCL-2 and BCR-ABL tyrosine kinase eradicates chronic myeloid leukemia stem cells. Sci Transl Med. 2016;8(355): 355ra117. 35. Carter BZ, Mak PY, Wang X, et al. Focal adhesion kinase as a potential target in AML and MDS. Mol Cancer Ther. 2017;16(6):1133-1144. 36. Han L, Qiu P, Zeng Z, et al. Single-cell mass cytometry reveals intracellular survival/proliferative signaling in FLT3-ITD-mutated AML stem/progenitor cells. Cytometry A. 2015;87(4):346-356. 37. Zhou H, Mak PY, Mu H, et al. Combined inhibition of beta-catenin and Bcr-Abl synergistically targets tyrosine kinase inhibitorresistant blast crisis chronic myeloid leukemia blasts and progenitors in vitro and in vivo. Leukemia. 2017;31(10):2065-2074. 38. Chen H, Lau MC, Wong MT, Newell EW, Poidinger M, Chen J. Cytofkit: a bioconductor package for an integrated mass cytometry data analysis pipeline. PLoS Comput Biol. 2016;12(9):e1005112. 39. Chou TC, Talalay P. Quantitative analysis of dose-effect relationships: the combined effects of multiple drugs or enzyme inhibitors. Adv Enzyme Regul. 1984;22:2755. 40. Sun Y, Bandi M, Lofton T, et al. Functional genomics reveals synthetic lethality between phosphogluconate dehydrogenase and oxidative phosphorylation. Cell Rep. 2019;26(2):469-482. 41. Gelman SJ, Naser F, Mahieu NG, et al. Consumption of NADPH for 2-HG synthesis increases pentose phosphate pathway flux and sensitizes cells to oxidative stress. Cell Rep. 2018;22(2):512-522.

42. Molina JR, Sun Y, Protopopova M, et al. An inhibitor of oxidative phosphorylation exploits cancer vulnerability. Nat Med. 2018;24(7):1036-1046. 43. Abraham M, Biyder K, Begin M, et al. Enhanced unique pattern of hematopoietic cell mobilization induced by the CXCR4 antagonist 4F-benzoyl-TN14003. Stem Cells. 2007;25(9):2158-2166. 44. Virappane P, Gale R, Hills R, et al. Mutation of the Wilms' tumor 1 gene is a poor prognostic factor associated with chemotherapy resistance in normal karyotype acute myeloid leukemia: the United Kingdom Medical Research Council Adult Leukaemia Working Party. J Clin Oncol. 2008;26(33): 5429-5435. 45. Hwang AB, Lee SJ. Regulation of life span by mitochondrial respiration: the HIF-1 and ROS connection. Aging. 2011;3(3):304-310. 46. Mathieu J, Zhang Z, Zhou W, et al. HIF induces human embryonic stem cell markers in cancer cells. Cancer Res. 2011;71(13):4640-4652. 47. Blombery P, Anderson MA, Gong JN, et al. Acquisition of the recurrent Gly101Val mutation in BCL2 confers resistance to venetoclax in patients with progressive chronic lymphocytic leukemia. Cancer Discov. 2019;9(3):342-353. 48. Chen X, Glytsou C, Zhou H, et al. Targeting mitochondrial structure sensitizes acute myeloid leukemia to venetoclax treatment. Cancer Discov. 2019;9(7):890-909. 49. Nechiporuk T, Kurtz SE, Nikolova O, et al. The TP53 apoptotic network is a primary mediator of resistance to BCL2 inhibition in AML cells. Cancer Discov. 2019;9(7):910925. 50. Zhao X, Ren Y, Lawlor M, et al. BCL2 amplicon loss and transcriptional remodeling drives ABT-199 resistance in B cell lymphoma models. Cancer Cell. 2019;35(5):752766.

haematologica | 2022; 107(1)

ARTICLE

Acute Myeloid Leukemia

A genome-wide CRISPR screen identifies regulators of MAPK and MTOR pathways that mediate resistance to sorafenib in acute myeloid leukemia

Ferrata Storti Foundation

Alisa Damnernsawad,1,2 Daniel Bottomly,3 Stephen E. Kurtz,4 Christopher A. Eide,4 Shannon K. McWeeney,3 Jeffrey W. Tyner1,4 and Tamilla Nechiporuk1 Department of Cell, Developmental & Cancer Biology, Knight Cancer Institute, Oregon Health & Science University, Portland, OR, USA; 2Department of Biology, Faculty of Science, Mahidol University, Bangkok, Thailand; 3Division of Bioinformatics and Computational Biology, Knight Cancer Institute, Oregon Health & Science University, Portland, OR, USA and 4Division of Hematology and Medical Oncology, Knight Cancer institute, Oregon Health & Science University, Portland, OR, USA 1

Haematologica 2022 Volume 107(1):77-85

ABSTRACT

rug resistance impedes the long-term effect of targeted therapies in acute myeloid leukemia (AML), necessitating the identification of mechanisms underlying resistance. Approximately 25% of AML patients carry FLT3 mutations and develop post-treatment insensitivity to FLT3 inhibitors, including sorafenib. Using a genomewide CRISPR screen, we identified LZTR1, NF1, TSC1 and TSC2, negative regulators of the MAPK and MTOR pathways, as mediators of resistance to sorafenib. Analyses of ex vivo drug sensitivity assays in samples from patients with FLT3-ITD AML revealed that lower expression of LZTR1, NF1, and TSC2 correlated with sensitivity to sorafenib. Importantly, MAPK and/or MTOR complex 1 (MTORC1) activity was upregulated in AML cells made resistant to several FLT3 inhibitors, including crenolanib, quizartinib, and sorafenib. These cells were sensitive to MEK inhibitors, and the combination of FLT3 and MEK inhibitors showed enhanced efficacy, suggesting the effectiveness of such treatment in AML patients with FLT3 mutations and those with resistance to FLT3 inhibitors.

Introduction Acute myeloid leukemia (AML), a rapidly progressing hematologic malignancy, is caused by the impaired differentiation and subsequent proliferation of hematopoietic progenitor cells. AML is characterized by cytogenetic heterogeneity and numerous recurrent genetic lesions.1-5 The Fms-related tyrosine kinase 3 (FLT3) receptor tyrosine kinase is normally expressed on hematopoietic stem and progenitor cells and functions in promoting cell proliferation and survival as well as normal development of these cells.6-8 FLT3 activating mutations occur in approximately 25% of AML patients, either by internal tandem duplications (FLT3-ITD) or point mutations in the tyrosine kinase domain,4,9-12 stimulating AML cell proliferation and survival. These mutations are associated with poor outcomes including an enhanced risk of relapse.6,7,13 The high frequency and adverse effects of FLT3 mutations have prompted the development of small-molecule inhibitors targeting FLT3. Among the FLT3 tyrosine kinase inhibitors that have been developed, several have provided encouraging results in clinical trials,7,14,15 and two in particular, midostaurin and gilteritinib, have been approved for FLT3-mutant AML.16-18 Nevertheless, all the FLT3 inhibitors developed to date lack long-term, durable clinical efficacy because of the development of resistance. Point mutations within the kinase domain of FLT3, such as variants in residues D835 and F691, cause resistance to type II FLT3 inhibitors (quizartinib and sorafenib) in vitro as well as in relapsed/refractory patients.19-21 While patients with tyrosine kinase domain mutations develop resistance to type II inhibitors, they are sensitive to type I

haematologica | 2022; 107(1)

Correspondence: TAMILLA NECHIPORUK nechipot@ohsu.edu JEFFREY W. TYNER tynerj@ohsu.edu Received: May 3, 2020. Accepted: December 21, 2020. Pre-published: December 30, 2020. https://doi.org/10.3324/haematol.2020.257964

A. Damnernsawad et al.

inhibitors, even though the responses are transient.21 Diverse mutations underlie resistance in AML patients to the type I inhibitor, crenolanib, including rare mutations at the gatekeeper region of FLT3, as well as NRAS and IDH2 mutations in FLT3-independent subclones, and TET2 and IDH1 mutations in FLT3 mutant clones in nonresponding patients.22 Single-cell DNA sequencing of specimens from patients relapsing after treatment with a novel type I inhibitor, gilteritinib, revealed shifts in clonal architecture to select for secondary mutations in NRAS, KRAS, IDH2, or BCR-ABL1, either in the context of FLT3ITD or FLT3 wild-type clones.18 Aberrant activation of ERK either extrinsically through the bone marrow microenvironment or intrinsically in a cell-autonomous manner has been implicated in FLT3 resistance in AML.23,24 Upregulation of the RAS/RAF/ERK pathway has been observed after treatment with FLT3 tyrosine kinase inhibitors in AML cell lines and AML patients’ bone marrow samples.23,25 Signaling through JAK/STAT5 mediated by granulocyte-macrophage colony-stimulating factor and interleukin-3 allows AML cells to survive FLT3 inhibitor treatment.26 Activation of the phosphatidylinositol-3 kinase (PI3K)/mammalian target of rapamycin (MTOR) pathway has also been demonstrated to promote resistance to a FLT3 inhibitor.27 Sorafenib, a multi-kinase inhibitor targeting not only FLT3 but also RAF, VEGFR, FGFR, KIT and RET,28 has been evaluated in combination with azacytidine in AML patients with FLT3-ITD, who had an overall response rate of 46%.29 The combination of sorafenib and standard-ofcare chemotherapy extended event-free survival in patients younger than 60 years old.30 Data from a phase I trial showed that patients harboring FLT3-ITD who were treated with allogeneic hematopoietic stem cell transplantation had a 1-year progression-free survival rate of 85% and a 1-year overall survival rate of 95%.31 To identify mechanisms of resistance to sorafenib we used a genome-wide CRISPR (clusters of regularly interspaced short palindromic repeats) knockout screen to search for genes whose loss-of-function variants can promote FLT3 inhibitor-sensitive AML cells to survive in the presence of sorafenib. To confirm that aberrant signaling in the identified pathways renders cells insensitive to FLT3 inhibitors, we established AML cells resistant to both type I and type II FLT3 inhibitors. Our CRISPR screen identified genes in the MTOR and mitogen-activated protein kinase (MAPK) pathways that modulate sensitivity to sorafenib. Activities of MTOR and MAPK pathways were upregulated in cells with acquired resistance, and these cells were sensitive to MEK inhibitors supporting the role of aberrant downstream MAPK signaling in resistance to FLT3 inhibitors. We found the combination of FLT3 and MEK inhibitors had synergistic efficacy in both FLT3 inhibitor-sensitive and -resistant AML cells as well as in samples from AML patients. In summary, our work identified several negative regulators of MTOR and MAPK signaling pathways, LZTR1, TSC1/2, NPRL2, NF1, not previously associated with AML, as modulators of sensitivity to sorafenib. We show that aberrations in MTOR and MAPK pathways are important mechanisms of resistance to sorafenib as well as other FLT3 inhibitors in AML and suggest that the combination of FLT3 inhibitors and MEK inhibitors could be useful for the treatment of FLT3 inhibitor-resistant AML. 78

Methods Cell lines Human MOLM13 cells were obtained from the Sanger Institute Cancer Cell Line Panel. All cell lines used in this study were authenticated at the OHSU DNA Services Core facility. Cell lines were maintained in 20% fetal bovine serum, RPMI medium, supplemented with glutamine, penicillin/streptomycin and an antimycotic. All cell lines were tested for Mycoplasma on a monthly schedule.

Lentivirus production and transduction HEK293T cells were transfected using Lipofectamine-2000 (Invitrogen) with single transfer vectors in combination with packaging plasmids, psPax2 (Addgene, #12260) and VSVG (Invitrogen). Supernatants were collected, filtered through 0.45 mM filters and used for transduction as described previously.32

The CRISPR/Cas9 library screen and CRISPR/Cas9 gene inactivation by individual sgRNA Cas9-expressing cells were generated using Cas9Blst (Addgene, #52962). Loss-of-function screens were performed using pooled human genome-wide single-guide (sg)RNA libraries, the Y. Kosuke library,33 purchased from Addgene (#67989), as described previously,32 which targets 18,010 genes with 90,709 sgRNA (average of 5 guides per gene). High-titer lentivirus was generated using standard calcium phosphate precipitation procedures in HEK293T cells. Viral supernatant was concentrated and the titer determined using a viral titration kit (ABM good, Canada). One hundred million cells were used for viral transduction at a multiplicity of infection (MOI) of 0.3, selected with puromycin for 5-7 days to ensure stable viral integration. Individual genes were inactivated by cloning sgRNA into plentiCRISPRV2 (Addgene, #52961) according to the manufacturer’s suggestions. The following sgRNA were used in the study: LZTR1: 5’ CCCATAGACGACGGCCGAG 3’, NF1: 5’CATATCAGTCTGTGGGATC 3’, TSC1, 5’ACGTCGTTGTCCTCACAAC 3’, TSC2: 5’ TTGATGCGCACGGCGCCTC 3’, NPRL2: 5’ GAACCCATCAATGTAGGGC 3’, DEPDC 5’ GACTGTGACTCAAGTGTTCC and 5’ TGTTAATGTCGTAGACCCTA, TBC1D7 5’ GTATCGTASAGGAGCAGTACT. Sequencing data were deposited to GEO with, accession number GSE138343.

Drug sensitivity assay Small-molecule inhibitors, purchased from LC Laboratories Inc. (Woburn, MA, USA) and Selleck Chemicals (Houston, TX, USA), were reconstituted in dimethylsulfoxide (DMSO). Cells were seeded at 1,000 cells/well in a 384-well plate in 50 mL medium (RPMI-1640 supplemented with fetal bovine serum [15%], L-glutamine, penicillin-streptomycin and an antimycotic) with different concentrations of drugs and cultured for 72 h. For the drug sensitivity assay, 5 uL of MTS reagent (CellTiter96 AQ One; Promega Madison, WI, USA) were added to each well and incubated for 4 h. Optical density was measured at 490 nm. Relative cell viability was calculated by normalizing the readings to those of untreated control wells. Prism software (GraphPad) was used to produce non-linear fitting and determine the response to the drug, the half maximal inhibitory concentration (IC ) and the area under the curve (AUC). ueous

Immunoblot analysis Whole cell protein lysates were prepared using cell lysis buffer (Cell Signaling Technologies), 1 mM phenylmethylsulfonyl fluoride, proteasome (Roche) and a phosphatase inhibitor cocktail (Sigma-Aldrich). Proteins were resolved on 4 -15% gradient gels

haematologica | 2022; 107(1)

Mechanisms of sorafenib resistance in AML

(Biorad), transferred onto polyvinylidene fluoride membranes (Amersham), and subjected to immunoblotting using primary antibodies from Cell Signaling Technologies: p44/42 MAPK (ERK1/2; #9102), phospho-ERK1/2 (phospho-p44/42 MAPK Thr202/Tyr204; #4376), AKT (#9272), phospho-AKT (Ser473; #4060), TSC1 (#6935), phospho-TSC2 (Ser664; #40729), phosphoTSC2 (Tyr1571; #3614), TSC2 (#4308), phospho-mTOR (Ser2481; #2974), phospho-mTOR (Ser2448; #2971), mTOR (#2983), NF1 (#14623), MEK (#9122S), phosphor-MEK (#9154), vinculin (#4650); from ThermoFisher Scientific: GAPDH (#AM4300); from Millipore: anti-pan_Ras (clone RAS 10 MABS195) from Sigma: LZTR1 (HPA071248). Corresponding horseradish perodixase-conjugated secondary antibodies (Promega) were used for chemiluminescent detection.

Biostatistical analysis The bioinformatics pipeline for analyzing CRISPR library sequences was MAGeCK (model-based analysis of genome-wide CRISPR-Cas9 knockout).34 The hits were prioritized according to a previously described tiering structure.32 Briefly, tier 1 represents hits having a log fold change ≥2, 75% of sgRNA per gene present and concordance among sgRNA per gene ≥75%; tier 2 hits have a log fold change ≥2 and concordance among sgRNA per gene of 100%; tier 3 hits have a log fold change ≥1 and a concordance among sgRNA per gene of 100%. Singleton hits represent significantly enriched genes with log fold change ≥2, an adjusted sgRNA count of 1 and average control mean ≥100 reads. Enriched hits not satisfying these criteria were classified into the unassigned group. 2

included members of the GATOR1 complex, encoded by NPRL2 and DEPCD5.38 To prioritize candidates for validation, we developed a tiering structure that incorporates three key factors: evidence (determined by the number of sgRNA guide hits per gene), concordance (indicated by the agreement across the set of guides for a given gene) and discovery (based on effect size) to rank sgRNA hits and enable a progression to pathway analysis for lower scoring hits.38 Using the prioritization scheme, the tier 1 hits (n=16) included LZTR1, TSC2, and TBC1D7 and several genes implicated in RNA splicing and ribosome biogenesis, such as DHX15, EBNA1BP2, LSM5, PUS7, RPSA and ABCB1 transporter, linked to poor prognostic factors in AML (Online Supplementary Tables S1 and S2, Online Supplementary Figure S1). Our tier structure imposed additional constraints for ranking sgRNA hits into the more selective tiers, which generally preserved MAGeCK RRA rankings, although there were exceptions such as TSC1, which ranked as a tier 3 hit because of the variance in its log-fold change across the set of sgRNA for this gene. Using a false discovery rate cutoff, we decided to focus here on connecting the AKT/PI3K/MTOR and RAS/MAPK/MEK networks to verify candidates emerging from the screen (Figure 1B bottom panel, D).

Data availability. Raw data files for CRISPR screens have been deposited at GEO and can be found under the accession number GSE138343.

Results The MTOR and MAPK pathways are central components in resistance to sorafenib To identify genes whose loss-of-function variants contribute to resistance to sorafenib in AML, we selected MOLM13 cells, an AML cell line harboring an FLT3-ITD mutation resulting in sensitivity to several FLT3 inhibitors, including sorafenib. MOLM13 cells, engineered to express Cas9, were stably transduced with a genome-wide lentiviral sgRNA CRISPR knockout library33 and treated for 14 days with vehicle or 50 nM sorafenib, a concentration projected to kill 80% of the cells within 3 days of drug administration (IC ). Genomic DNA was harvested from control and sorafenib-treated cultures and evaluated for enriched sgRNA using MAGeCK robust rank aggregation (RRA) analyses34 (Figure 1A, B; Online Supplementary Table S1). A comparison of sequencing reads from sorafenibtreated cultures and vehicle-treated controls identified significant enrichment for sgRNA targeting negative regulators of the MAPK and AKT/MTOR pathways (Figure 1B-D). The screen uncovered negative RAS/RAF/MEK/ERK regulator, leucine zipper like transcription regulator 1 (LZTR1), which inhibits the MAPK pathway by regulating RAS ubiquitination and degradation.35,36 a negative regulator of RAS signaling, neurofibromin 1 (NF1); three components of the tuberous sclerosis (TSC) complex including TSC complex subunit 1 (TSC1), TSC complex subunit 2 (TSC2), and TBC1 domain family member 7 (TBC1D7).37 Top hits also 80

haematologica | 2022; 107(1)

Deficiency of top hit genes decreases sensitivity to sorafenib in acute myeloid leukemia cell lines To validate top hit genes belonging to the LZTR1-connected network, we transduced MOLM13 cells with lentivirus expressing Cas9 and individual sgRNA to generate cells deficient in single genes. Sensitivity to sorafenib was assessed in 72 h cell viability MTS assays. Cells in which LZTR1, NF1, TSC1, TSC2, or NPRL2 were inactivated showed reduced sensitivity to sorafenib (Figure 1E). The degree of resistance to sorafenib varied across targeted genes, with TSC1- and LZTR1-deficient cells demonstrating the strongest resistance to sorafenib (parental IC = 5.03 nM, NT (non-targeting control) IC = 6.31 nM, sgTSC1 IC = 97.34 nM, and sgLZTR1 IC = 22.37 nM), while targeting of TSC2 yielded comparably more modest resistance to sorafenib (IC = 14 nM) (Figure 1E). TBC1D7deficient cells had decreased sensitivity to sorafenib while DEPCD5-deficient cells were modestly resistant (Online Supplementary Figure S2). Deficiencies of LZRT1, NF1, TSC1, TSC2 and NPRL2 were evident by western blot analysis (Online Supplementary Figure S3A). The corresponding efficiencies of CRISPR knockouts were determined using Inference of CRISPR Edits (ICE) software (Synthego.com) (Online Supplementary Figure S3B). 50

Reduced expression levels of LZTR1, NF1, TSC1, and TSC2 correlate with reduced sensitivity to sorafenib in samples from patients with acute myeloid leukemia and deficiency results in hyperactivation of MAPK or MTOR pathways in acute myeloid leukemia cells We evaluated results from our CRISPR screen for relevance to drug sensitivity and gene expression profiles observed in patients’ samples in the Beat AML database.3 RNA expression levels of LZTR1, NF1, and TSC2 showed negative correlations with sensitivity to sorafenib in samples from AML patients harboring FLT3-ITD mutations (P<0.0001, P<0.001, and P<0.01, respectively) (Figure 2A). We did not observe a significant negative correlation between gene expression and sensitivity to sorafenib for 79

A. Damnernsawad et al.

other screen hits, including RASA2, TBC1D7, NPRL2, and DEPDC5. In MOLM13 cells, engineered to model LZTR1 and NF1 deficiencies, we observed elevated levels of phosphorylated ERK, suggesting increased activation of the MAPK signaling

pathway (Figure 2B). As MAPK can cross-activate MTOR signaling,39-41 we observed increased phosphorylation level of MTORC1, similar to results with inactivated inhibitory functions of TSC1 and TSC2, indicating common aberrancy in downstream signaling (Figure 2B). Elevated levels of

E D

Figure 1. A genome-wide CRISPR knockout screen identifies negative regulators of MAPK and MTOR pathways, LZTR1, NF1, TSC1/2 and NPRL2, attenuating sensitivity to sorafenib in acute myeloid leukemia cell lines. (A) Schematic representation of the CRISPR knockout screen in MOLM13 acute myeloid leukemia cells. (B) Scatter plots of differential enrichment of single guide (sg)RNA in sorafenib-treated versus dimethylsulfoxide (DMSO)-treated control cells, analyzed at day 14 of drug exposure. Median log fold changes are plotted versus P-values (top panel), and false discovery rate (FDR)-adjusted P-values (bottom panel), generated from MAGeCK robust ranked aggregation (RRA) analysis. Colors represent various tiers associated with the ranking system based on directional concordance among sgRNA and significance of fold increase (see Methods section). (C) Plots of normalized sgRNA read counts targeting the five top candidate genes from sorafenib- and DMSOtreated samples over time. (D) Diagram of the MAPK and MTOR pathways with top hit genes shown in red circles. (E) Dose response curve of the 72 h sorafenib sensitivity assay performed on parental MOLM13 cells and MOLM13 cells transduced with sgRNA targeting LZTR1, NF1, TSC1, TSC2, NPRL2, and non-targeting (NT) control. The percentages of viable cells were measured using the MTS assay in triplicate with seven-point escalating drug concentrations.

haematologica | 2022; 107(1)

Mechanisms of sorafenib resistance in AML

phospho-ERK were evident in TSC1-deficient, but not in TSC2-deficient cells, potentially reflecting different roles of TSC1 and TSC2 in the TSC complex. Levels of RAS protein were elevated in LZTR1-deficient cells, but not in those deficient in NF1 or TSC1 (Figure 2C).

MTOR and MAPK pathways are upregulated in acute myeloid cells resistant to FLT3 inhibitors In a parallel approach to understand mechanisms of resistance to sorafenib, we generated AML cell lines resistant to FLT3 inhibitors by gradually exposing MOLM13 cells to type I (crenolanib) or type II (quizartinib and sorafenib) FLT3 inhibitors. Crenolanib- and sorafenib- resistant MOLM13 cells showed reduced sensitivity, detected by higher IC and AUC values, to both type I and type II inhibitors, whereas quizartinib-resistant cells showed resistance only to type II inhibitors (Figure 3A). To investigate whether there is an overlap between the acquisition of resistance by CRISPR-derived knockout cells versus resistance generated by prolonged exposure to drugs, we evaluated the activity of MTORC1 and MAPK pathways. We detected an increase in levels of phospho-ERK and elevated RAS levels in FLT3 drug-resistant cells, indicating upregulation of the MAPK pathway (Figure 3B). Surprisingly, levels of TSC2 were increased in crenolanib- and sorafenib-resis50

tant cells; in contrast, loss of function for TSC2 was revealed in the CRISPR knockout resistance screen. Previous studies showed that the MAPK pathway can inhibit MTOR activity by phosphorylating TSC2 at S664 causing dissociation of the TSC complex observed in breast and colon carcinomas.41-43 This observation prompted us to test for levels of phospho-TSC2. We observed enhanced levels of phospho-TSC2 at S664 in these two cell lines, suggesting that inhibition of TSC complex formation by the MAPK pathway promotes resistance to FLT3 inhibitors. Moreover, we observed an elevated level of phospho-TSC2 at Y1571, which additionally impairs the TSC1-TSC2 interaction,44 supporting inactivation of TSC2 in these resistant cells, concordant with our CRISPR screen results.

MEK inhibitors resensitize FLT3-inhibitor-resistant cell lines Data from the CRISPR screen and resistant cell lines suggested that upregulation of the MAPK signaling pathway contributes to resistance to FLT3 inhibitors in AML. This prompted us to hypothesize that inhibitors targeting the MAPK pathway may resensitize FLT3-inhibitorresistant cells. Parental and resistant cells were tested for sensitivity to the MEK inhibitor trametinib and MTOR inhibitors PP242, PI-103 and rapamycin. Resistant cells

Figure 2. Low sensitivity to sorafenib correlates with low expression levels of CRISPR top hits LZTR1, NF1, and TSC2 in specimens from acute myeloid leukemia patients and inactivation in the MOLM13 acute myeloid leukemia cell line reveals aberrant activation of MAPK or MTOR pathways. (A) Scatter plots of RNA expression levels (shown as normalized reads per kilobase per million [RPKM]) plotted against drug sensitivity measured ex vivo as area under the curve (AUC) from samples from patients with acute myeloid leukemia harboring FLT3-ITD mutations (n=86). Pearson correlation coefficients are shown as (r) values Statistical significance is indicated. ****P<0.0001, ***P<0.001, **P<0.01, n.s., non-significant. (B, C) Immunoblot analyses of proteins from LZTR1-, NF1-, TSC1- and TSC2-deficient cells (LZTR1_sgRNA, NF1_sgRNA, TSC1_sgRNA and TSC2_sgRNA, correspondingly) in comparison to parental and non-targeting (NT_sgRNA) controls with indicated antibodies. p-ERK denotes phosphorylated-ERK (Thr202/Tyr204), p-MTOR denotes phosphorylated-MTOR at S2448. Vinculin and GAPDH immunoreactivity served as loading controls.

haematologica | 2022; 107(1)

A. Damnernsawad et al.

showed greater sensitivity to trametinib relative to the parental cells (Figure 3C; Online Supplementary Figure S4A). Combinations of each FLT3 inhibitor with trametinib revealed enhanced efficacy in both resistant and parental AML cells (Figure 4A; Online Supplementary Figure S4B). Moreover, combinations of sorafenib and trametinib showed high synergy scores at several concentrations as analyzed by R_SynergyFinder45 (average synergy [zero interaction potency] score for parental cells = 3.449 vs. 8.015 in sorafenib-resistant MOLM13 cells; synergy scores above 1 are significant). Similar synergy in sensitivity was obtained with trametinib in combination with either crenolanib or quizartinib (Online Supplementary Figure S4D). MTOR inhibitors PP242 and PI-103 did not exhibit substantial cell killing as single agents in any of the FLT3-inhibitor-resistant cell lines (Figure 3C, bottom panel; Online Supplementary Figure S4A) and a combination of MTOR inhibitor and a FLT3 inhibitor resulted in a marginal decrease in cell viability (Figure 4A bottom panels; Online Supplementary Figure S4B, C). In contrast, MTOR inhibitors appeared to resensitize TSC1- and NPRL2-deficient cells to sorafenib (Online Supplementary Figure S5). The effect of PI-103 was more pronounced than that of PP242 which may reflect its dual targeting of MTOR and PI3K.

Discussion Sorafenib, as well as other FLT3 inhibitors, in combination with standard-of-care chemotherapy prolongs the survival of AML patients with or without FLT3 mutations, although relapse caused by drug resistance remains a clinical challenge. To elucidate mechanisms of resistance to sorafenib, we subjected MOLM13 AML cells to genomewide CRISPR screening to identify genes whose loss-offunction contributes to reduced drug sensitivity. The top screen hits indicated that resistance to FLT3 inhibitors in AML can occur via aberrant activation of the AKT/PI3K/MTOR and RAS/MAPK signaling pathways. Our results are consistent with findings from previous studies on resistance to FLT3 inhibitors that revealed aberrant ERK and RAS signaling18,23 and extend these with the identification of a broad spectrum of genes regulating RAS/MAPK and, additionally, MTOR signaling pathways as modulators of resistance (Figure 4). Analysis of the screen using the MAGeCK pipeline in combination with a tiering system developed previously32 identified LZTR1, TSC1/2, NPRL2, NF1, and TBC1D7 as significant hits. The identification of LZTR1 is not unexpected as LZTR1 loss confers MAPK activation by dysregulating RAS signaling;35,36 it also facilitates degradation of

Figure 3. Acute myeloid leukemia cell lines made resistant to FLT3 inhibitors demonstrate activation of MAPK and/or MTOR pathways. (A) MOLM13 AML cell lines were made resistant to type I and type II FLT3 inhibitors by continuous exposure to crenolanib, quizartinib or sorafenib. Drug sensitivity assays were performed for 72 h on parental and FLT3-inhibitor-resistant MOLM13 cells. Cell viability was measured in triplicate using the MTS assay with seven-point escalating drug concentrations. (B) Immunoblot of whole cell lysates from parental MOLM13 and MOLM13 cells resistant to crenolanib and sorafenib treated with dimethlysulfoxide or FLT3 inhibitors at 20 nM concentration for 4 h performed with indicated antibodies. (C) Dose response curves of trametinib and PP242 on parental and FLT3-inhibitorresistant MOLM13 cells.

haematologica | 2022; 107(1)

Mechanisms of sorafenib resistance in AML

RAS-GTPases.46,47 Consistent with our results, loss of LZTR1 function has recently been shown to cause resistance to tyrosine kinase inhibitors including several FLT3 inhibitors (tandutinib, quizartinib, and ponatinib) in AML cell lines.36 LZTR1 loss-of-function mutations have been observed in other cancers including glioblastoma multiforme, adrenocortical cancer, and pancreatic cancer, and have also recently been reported in hepatocellular carcinoma, a cancer for which sorafenib is a first-line therapy.48,49 The screen hits included components of the TSC and the GATOR complexes, which have not been previously identified as modulators of resistance to FLT3 inhibitors. Our screen also identified two negative regulators of RAS: NF1 and RASA2. Loss-of-function mutations in NF1 have been associated with poor prognosis in AML patients,50 suggest-

ing that patients with NF1 mutations would have poor sensitivity to FLT3 inhibitors. TSC2 and TBC1D7 along with TSC1 form the TSC complex, which acts as a GTPase activating protein for RHEB, a small G-protein upstream of MTOR complex 1 (MTORC1).36,51-53 MTORC1 is activated by another small G-protein, RAG, which is negatively regulated by the GATOR1 complex, comprising NPRL2, NPRL3, and DEPDC5 proteins.37 Our screen identified TSC1 and components of the GATOR1 complex, DEPDC5 and NPRL2. Loss-of-function variants of TSC1 and TSC2 are found in ~16% of patients with hepatocarcinoma and are associated with an aggressive form of this malignancy.54 Our data suggest that the roles of TSC1 and TSC2 are complex; we note that TSC1-, but not TSC2-deficient MOLM13 cells

Figure 4. Evaluation of sensitivity to MTOR and MEK inhibitors in combination with FLT3 inhibitors. (A) Parental and FLT3inhibitor-resistant MOLM13 cell lines described in Figure 3A were tested for sensitivity to sorafenib, trametinib and PP242 by 72 h drug sensitivity assays. Cell viability was measured in triplicate using the MTS assay with seven-point escalating drug concentrations. Drug combinations were tested at equimolar concentrations. (B) Immunoblot of parental MOLM13 cells treated with sorafenib, trametinib, PP4242 or combinations at 10 nM for 24 h performed with the indicated antibodies. Vinculin immunoreactivity served as a loading control.

haematologica | 2022; 107(1)

A. Damnernsawad et al.

showed increased phospho-ERK activity. This observation may not be surprising given the structural difference between the two proteins; TSC1 lacks the kinase domain that is present in TSC2.36,51 It is possible that TSC1 executes functions outside of its interactions with TSC2 to regulate MAPK signaling. Consistent with results from the CRISPR screen, samples from AML patients harboring FLT3-ITD mutations with reduced RNA expression levels of LZTR1, NF1, and TSC2 exhibited less sensitivity to sorafenib. Upregulation of phospho-MTORC1 (Ser2481) is observed in two FLT3-inhibitor-resistant cell lines, made by gradual exposure to higher concentrations of FLT3 inhibitors, supporting a role for MTORC1 in resistance to FLT3 inhibitors. Increased phospho-ERK levels in FLT3resistant cells confirmed the importance of the MAPK pathway in FLT3 inhibitor resistance. Concordantly, the resistant cells demonstrated enhanced sensitivity to MEK inhibitors. We also showed that the connection between MAPK and MTOR pathways influenced FLT3 inhibitor resistance. We found enhanced levels of phospho-TSC2 at S664 and Y1571 in the resistant lines which are regulated by phosphorylated ERK and AKT, respectively. These phosphorylation events inhibit the formation of the TSC complex, mimicking TSC1 and TCS2 loss-of-function hits as revealed by the CRISPR resistance screen. The combination of FLT3 plus MEK inhibitors has synergistic effectiveness in both sensitive and resistant cells, a finding that is consistent with data from a recent study that demonstrated synergy for the combination of crenolanib with trametinib in Ba/F3 cells harboring PTPN11 A72D, FLT3 D835Y or double mutations of both.22 Similarly, the combination of sorafenib or pazopanib with trametinib showed strong synergy in MOLM13 cells.25 Moreover, the combination of sorafenib with the MEK inhibitor, PD0325901, showed synergy in MOLM14 and MV4;11 AML cell lines,23,55 and the combination of gilteritinib and trametinib had enhanced efficacy in MOLM14 cells with NRAS G12C and NRAS Q61K.18 Enhanced efficacy was observed in AML patients’ samples assayed ex vivo for sensitivity to a combination of an FLT3 inhibitor (quizartinib) and trametinib (Beat AML data, Online Supplementary Figure S4E). Evaluation of signaling in FLT3i-resistant cells and top hits from the CRISPR knockout screen underscore activation of the MTOR pathway

References 1. Papaemmanuil E, Gerstung M, Bullinger L, et al. Genomic classification and prognosis in acute myeloid leukemia. N Engl J Med. 2016;374(23):2209-2221. 2. Döhner H, Weisdorf DJ, Bloomfield CD. Acute myeloid leukemia. N Engl J Med. 2015;373(12):1136-1152. 3. Tyner JW, Tognon CT, Bottomly D, et al. Functional genomic landscape of acute myeloid leukaemia. Nature. 2018;562(7728): 526-531. 4. McCurdy SR, Levis MJ. Emerging molecular predictive and prognostic factors in acute myeloid leukemia. Leuk Lymphoma. 2018; 59(9):2021-2039. 5. Cancer Genome Atlas Research Network. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med. 2013;368(22):2059-2074.

in sorafenib resistance. Assessment of sensitivity to an MTOR inhibitor in combination with an FLT3 inhibitor showed a profound effect only in TSC1- and NPRL2-deficient cells. This result may indicate reliance on multiple pathways in FLT3-inhibitor-resistant cells. Increased activity of JAK/STAT5 has also been implicated in resistance to FLT3 inhibitors. Granulocytemacrophage colony-stimulating factor and interleukin-3 mediate FLT3 resistance in AML cells via JAK/STAT5 and PIM/cytokine-activated JAK/STAT5 signaling.26 Although, we did not evaluate JAK/STAT5 pathways in this study, our CRISPR screen did identify several negative regulators of JAK/STAT5 pathways, including PTPN1, SUMO3, and PTPN6. Additionally, our screen identified several tier 1 hits involved in RNA metabolism and splicing, including DHX15, EBNA1BP2, LSMR, PUS7 and RPSA. DHX15, for example, encodes an RNA helicase that is commonly mutated in AML patients with RUNX1RUNX1T1 fusions.56 It will be important to pursue these findings in subsequent studies given the roles of RNA metabolism in cancer pathogenesis.57-59 Disclosures JWT has received research support from Agios, Aptose, Array, AstraZeneca, Constellation, Genentech, Gilead, Incyte, Janssen, Petra, Seattle Genetics, Syros, and Takeda. The other authors have no conflicts of interest to disclose. Contributions AD and TN performed experimentation, DB and CE performed data analyses, AD, SK, TN, and JT, wrote the manuscript, JT, SM, and TN supervised the study. Acknowledgments We thank the OHSU Massively Parallel Sequencing Shared Resource and Flow Cytometry Core for technical support, and Sunil K. Joshi for reviewing the manuscript. Funding This study was supported by grants from the National Cancer Institute (1U01CA217862, 1U54CA224019, 3P30CA06953318S5). JWT received grants from the V Foundation for Cancer Research, the Gabrielle’s Angel Foundation for Cancer Research, and the National Cancer Institute (1R01CA183947). TN is supported by grant R50 CA251708-01.

6. Gilliland DG, Griffin JD. The roles of FLT3 in hematopoiesis and leukemia. Blood. 2002;100(5):1532-1542. 7. Daver N, Schlenk RF, Russel NH, et al. Targeting FLT3 mutations in AML: review of current knowledge and evidence. Leukemia. 2019;33(2):299-312. 8. Small D. FLT3 mutations: biology and treatment. Hematology Am Soc Hematol Educ Program. 2006(1):178-184. 9. Nakao M, Yokota S, Iwai T, et al. Internal tandem duplication of the flt3 gene found in acute myeloid leukemia. Leukemia. 1996;10(12):1911-918. 10. Yokota S, Kiyoi H, Nakao M, et al. Internal tandem duplication of the FLT3 gene is preferentially seen in acute myeloid leukemia and myelodysplastic syndrome among various hematological malignancies. A study on a large series of patients and cell lines. Leukemia. 1997;11(10):1605-1609.

11. Tse KF, Mukherjee G, Small D. Constitutive activation of FLT3 stimulates multiple intracellular signal transducers and results in transformation. Leukemia. 2000;14(10): 1766-1776. 12. Yamamoto Y, Nakano Y, Suzuki R, et al. Activating mutation of D835 within the activation loop of FLT3 in human hematologic malignancies. Blood. 2001;97(8):2434-2439. 13. Kiyoi H, Yanada M, Ozekia K. Clinical significance of FLT3 in leukemia. Int J Hematol. 2005;82(2):85-92. 14. Larrosa-Garcia M, Baer MR. FLT3 inhibitors in acute myeloid leukemia: current status and future directions. Mol Cancer Ther. 2017;16(6):991-1001. 15. Wu M, Li C, Zhu X. FLT3 inhibitors in acute myeloid leukemia. J Hematol Oncol. 2018;11(1):133. 16. Stone RM, Mandrekar S, Sanford SL, et al. Midostaurin plus chemotherapy for acute

haematologica | 2022; 107(1)

Mechanisms of sorafenib resistance in AML

myeloid leukemia with a FLT3 mutation. N Engl J Med. 2017;377(5):454-464. 17. Perl AE, Altman JK, Cortes J, et al. Selective inhibition of FLT3 by gilteritinib in relapsed or refractory acute myeloid leukaemia: a multicentre, first-in-human, open-label, phase 1–2 study. Lancet Oncol. 2017;18(8): 1061-1075. 18. McMahon CM, Ferng T, Canaani J, et al. Clonal selection with Ras pathway activation mediates secondary clinical resistance to selective FLT3 inhibition in acute myeloid leukemia. Cancer Discov. 2019;9(8):10501063. 19. Smith CC, Wang Q, Chin C-S, et al. Validation of ITD mutations in FLT3 as a therapeutic target in human acute myeloid leukaemia. Nature. 2012;485(7397):260-263. 20. Man CH, Fing TK, Ho C, et al. Sorafenib treatment of FLT3-ITD(+) acute myeloid leukemia: favorable initial outcome and mechanisms of subsequent nonresponsiveness associated with the emergence of a D835 mutation. Blood. 2012;119(22):51335143. 21. Albers C, Leischner H, Verbeek M, et al. The secondary FLT3-ITD F691L mutation induces resistance to AC220 in FLT3-ITD+ AML but retains in vitro sensitivity to PKC412 and Sunitinib. Leukemia. 2013;27 (6):1416-1418. 22. Zhang H, Savage S, Schultz AR, et al. Clinical resistance to crenolanib in acute myeloid leukemia due to diverse molecular mechanisms. Nature Commun. 2019;10(1): 244-257. 23. Bruner JK, Ma HS, Li L, et al. Adaptation to TKI treatment reactivates ERK signaling in tyrosine kinase-driven leukemias and other malignancies. Cancer Res. 2017;77(20):55545563. 24. Yang X, Sexauer A, Levis M. Bone marrow stroma-mediated resistance to FLT3 inhibitors in FLT3-ITD AML is mediated by persistent activation of extracellular regulated kinase. Br J Haematol. 2014;164(1):61-72. 25. Martínez-López J, Linares M, Morales ML, et al. Preclinical evidence that trametinib enhances the response to tyrosine kinase inhibitors in acute myeloid leukemia. Blood. 2016;128(22):1581. 26. Sung PJ, Sugita M, Koblish H, et al. Hematopoietic cytokines mediate resistance to targeted therapy in FLT3-ITD acute myeloid leukemia. Blood Adv. 2019;3(7): 1061-1072. 27. Lindblad O, Cordero E, Puissant A, et al. Aberrant activation of the PI3K/mTOR pathway promotes resistance to sorafenib in AML. Oncogene. 2016;35(39):5119-5131. 28. Wilhelm S, Carter C, Lynch M, et al. Discovery and development of sorafenib: a multikinase inhibitor for treating cancer. Nat Rev Drug Discov. 2006;5(10):835-844. 29. Ravandi F, Alattar ML, Grunwald MR, et al. Phase 2 study of azacytidine plus sorafenib in patients with acute myeloid leukemia and

haematologica | 2022; 107(1)

FLT-3 internal tandem duplication mutation. Blood. 2013;121(23):4655-4662. 30. Röllig C, Serve H, Huttmann A, et al. Addition of sorafenib versus placebo to standard therapy in patients aged 60 years or younger with newly diagnosed acute myeloid leukaemia (SORAML): a multicentre, phase 2, randomised controlled trial. Lancet Oncol. 2015;16(16):1691-1699. 31. Chen Y-B, Li S, Lane AA, et al. Phase I trial of maintenance sorafenib after allogeneic hematopoietic stem cell transplantation for Fms-like tyrosine kinase 3 internal tandem duplication acute myeloid leukemia. Biol Blood Marrow Transplant. 2014;20(12): 2042-2048. 32. Nechiporuk T, Kurtz SE, Nikolova O, et al. The TP53 apoptotic network is a primary mediator of resistance to BCL2 inhibition in AML cells. Cancer Discov. 2019;9(7):910925. 33. Tzelepis K, Koike-Yusa H, De Braekeleer E, et al. A CRISPR dropout screen identifies genetic vulnerabilities and therapeutic targets in acute myeloid leukemia. Cell Rep. 2016; 17(4):1193-1205. 34. Li W, Xu H, Xiao T, et al. MAGeCK enables robust identification of essential genes from genome-scale CRISPR/Cas9 knockout screens. Genome Biol. 2014;15(12):554-566. 35. Steklov M, Pandolfi S, Baietti MF, et al, Mutations in LZTR1 drive human disease by dysregulating RAS ubiquitination. Science. 2018;362(6419):1177-1182. 36. Bigenzahn J, Collu GM, Kartnig F, et al. LZTR1 is a regulator of RAS ubiquitination and signaling. Science. 2018;362(6419):11711177. 37. Dibble CC, Elis W, Menon S, et al. TBC1D7 is a third subunit of the TSC1-TSC2 complex upstream of mTORC1. Mol Cell. 2012;47(4):535-546. 38. Shen K, Huang RK, Brignole RG, et al. Architecture of the human GATOR1 and GATOR1–Rag GTPases complexes. Nature. 2018;556(7699):64-69. 39. Carriere A, Romea Y, Acosta-Jacquez JK, et al. ERK1/2 phosphorylate Raptor to promote Ras-dependent activation of mTOR complex 1 (mTORC1). J Biol Chem. 2011;286(1):567-577. 40. Mendoza MC, Er EE, Blenis J. The Ras-ERK and PI3K-mTOR pathways: cross-talk and compensation. Trends Biochem Sci. 2011;36 (6):320-328. 41. Ma L, Chen Z, Erdjument-Bromage H, et al. Phosphorylation and functional inactivation of TSC2 by Erk: implications for tuberous sclerosis and cancer pathogenesis. Cell. 2005;121(2):179-193. 42. Ma L, Teruya-Feldstein J, Bonner P, et al. Identification of S664 TSC2 phosphorylation as a marker for extracellular signal-regulated kinase-mediated mTOR activation in tuberous sclerosis and human cancer. Cancer Res. 2007;67(15):7106-7112. 43. Ballif BA, Roux PP, Gerber SA, et al.

Quantitative phosphorylation profiling of the ERK/p90 ribosomal S6 kinase-signaling cassette and its targets, the tuberous sclerosis tumor suppressors. Proc Natl Acad Sci U S A. 2005;102(3):667-672. 44. Aicher LD, Campbell JS, Yeung RS. Tuberin phosphorylation regulates its interaction with hamartin: two proteins involved in tuberous sclerosis. J Biol Chem. 2001;276 (24):21017-21021. 45. He L, Kulesskiy E, Saarela J, et al. Methods for high-throughput drug combination screening and synergy scoring. Methods Mol Biol. 2018;1711:351-398. 46. Castel P, Cheng A, Cuevas-Navarro A, et al. RIT1 oncoproteins escape LZTR1-mediated proteolysis. Science. 2019; 363(6432):12261230. 47. Abe T, Umeki I, Kanno S-I, et al. LZTR1 facilitates polyubiquitination and degradation of RAS-GTPases. Cell Death Differ. 2020;27(3):1023-1035. 48. Cancer Genome Atlas Research Network. Comprehensive and integrative genomic characterization of hepatocellular carcinoma. Cell. 2017;169(7):1327-1341. 49. Frattini V, Trifonov, V, Chan JM, et al. The integrated landscape of driver genomic alterations in glioblastoma. Nat Genet. 2013;45 (10): 1141-1149. 50. Eisfeld A-K, Kohlschmidt J, Mrozek K, et al. NF1 mutations are recurrent in adult acute myeloid leukemia and confer poor outcome. Leukemia. 2018;32(12):2536-2545. 51. Huang J, Manning BD. The TSC1–TSC2 complex: a molecular switchboard controlling cell growth. Biochem J. 2008;412(2):179190. 52. Laplante M, Sabatini DM. mTOR signaling in growth control and disease. Cell. 2012;149(2):274-293. 53. Manning BD, Toker A. AKT/PKB signaling: navigating the network. Cell. 2017;169(3): 381-405. 54. Ho DWH, Chan LK, Chui YT, et al. TSC1/2 mutations define a molecular subset of HCC with aggressive behaviour and treatment implication. Gut. 2017;66(8):1496-1506. 55. Morales ML, Arenas A, Ortiz-Ruiz A, et al. MEK inhibition enhances the response to tyrosine kinase inhibitors in acute myeloid leukemia. Sci Rep. 2019;9(1):18630-18641. 56. Opatz S, Bamopoulus SA, Metzeler KH, et al. The clinical mutatome of core binding factor leukemia. Leukemia. 2020;34(6): 1553-1562. 57. Wang BD, Lee NH. Aberrant RNA splicing in cancer and drug resistance. Cancers (Basel). 2018;10(11):458-482. 58. Desterro J, Bak-Gordon P, Carmo-Fonseca M. Targeting mRNA processing as an anticancer strategy. Nat Rev Drug Discov. 2020;19(2):112-129. 59. Di C, Syafrizayanti, Zhang Q, et al. Function, clinical application, and strategies of pre-mRNA splicing in cancer. Cell Death Differ. 2019;26(7):1181-1194.

ARTICLE Ferrata Storti Foundation

Acute Myeloid Leukemia

Epigenetic changes in human model KMT2A leukemias highlight early events during leukemogenesis Thomas Milan,1 Magalie Celton,1 Karine Lagacé,1 Élodie Roques,1 Safia Safa-Tahar-Henni,1 Eva Bresson,2,3 Anne Bergeron,2,3 Josée Hebert,4,5 Soheil Meshinchi,6 Sonia Cellot,7 Frédéric Barabé2,3,8 and Brian T. Wilhelm1,5 Laboratory for High Throughput Biology, Institute for Research in Immunology and Cancer, Montréal, Québec, Canada; 2Centre de Recherche en Infectiologie du CHUL, Centre de Recherche du CHU de Québec Université Laval, Québec City, Québec, Canada; 3CHU de Québec Université Laval Hôpital Enfant-Jésus, Québec City, Québec, Canada; 4Division of Hematology-Oncology and Leukemia Cell Bank of Quebec, Maisonneuve-Rosemont Hospital, Montréal, Québec, Canada; 5Department of Medicine, Université de Montréal, Montréal, Québec, Canada; 6Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, Washington, PA, USA; 7Department of Pediatrics, Division of Hematology, Ste-Justine Hospital, Montréal, Québec, Canada and 8 Department of Medicine, Université Laval, Quebec City, Québec, Canada 1

Haematologica 2022 Volume 107(1):86-99

ABSTRACT

Correspondence: BRIAN T. WILHELM brian.wilhelm@umontreal.ca Received: September 7, 2020. Accepted: December 21, 2020. Pre-published: December 30, 2020. https://doi.org/10.3324/haematol.2020.271619

hromosomal translocations involving the KMT2A gene are among the most common genetic alterations found in pediatric acute myeloid leukemias although the molecular mechanisms that initiate the disease remain incompletely defined. To elucidate these initiating events we used a human model system of acute myeloid leukemia driven by the KMT2A-MLLT3 (KM3) fusion. More specifically, we investigated changes in DNA methylation, histone modifications, and chromatin accessibility at each stage of our model system and correlated these with expression changes. We observed the development of a pronounced hypomethylation phenotype in the early stages of leukemic transformation after KM3 addition along with loss of expression of stem-cell-associated genes and skewed expression of other genes, such as S100A8/9, implicated in leukemogenesis. In addition, early increases in the expression of the lysine demethylase KDM4B was functionally linked to these expression changes as well as other key transcription factors. Remarkably, our ATAC-sequencing data showed that there were relatively few leukemia-specific changes and that the vast majority corresponded to open chromatin regions and transcription factor clusters previously observed in other cell types. Integration of the gene expression and epigenetic changes revealed that the adenylate cyclase gene ADCY9 is an essential gene in KM3-acute myeloid leukemia, and suggested the potential for autocrine signaling through the chemokine receptor CCR1 and CCL23 ligand. Collectively, our results suggest that KM3 induces subtle changes in the epigenome while co-opting the normal transcriptional machinery to drive leukemogenesis.

Introduction Leukemias are a class of blood disorders characterized by the uncontrolled proliferation of hematopoietic stem and progenitor cells (HSPC) that have acquired a block in the normal process of differentiation.1-3 Recent large-scale studies have highlighted the age-related genetic differences between pediatric and adult acute myeloid leukemia (AML) and have also revealed the high level of genetic heterogeneity in this disease.4 Not only are chromosomal translocations more frequent in younger patients, but recurrent mutations in specific genes implicated in the disease show a bias in either pediatric patients (e.g., NRAS, KIT, KRAS) or adults (DNMT3A, NPM1, IDH1/2).5 As a result, defining the combined role of these mutations in each patient, especially in pediatric patients in whom predisposition

haematologica | 2022; 107(1)

Early epigenetic changes in KM3-AML

variants may play a greater role, is extremely challenging and increases the complexity of identifying novel potential targeted therapeutics. Studies underscore the fact that even though the mutational burden in AML is one of the lowest of any cancer4 there is still tremendous heterogeneity between leukemic patients which creates challenges when trying to define the genetic determinates of the disease. As noted, chromosomal fusions are common in pediatric AML, particularly those involving the lysine specific methyl transferase 2A gene (KMT2A; also known as the mixed lineage leukemia [MLL] gene). KMT2A is rearranged in approximately 10% of all leukemias6 but the frequency in infant acute lymphocytic leukemia (ALL, >70%7) and AML (>35%7) is much higher. The KMT2A gene encodes a large (500 kDa) and complex 38-exon protein8,9 that can be cleaved by taspase-1 into two separate fragments (KMT2A-C and KMT2A-N).10 The conserved SET domain located at the C-terminus of the protein is responsible for the methylation of the histone H3 at lysine 4 (H3K4). Previous studies have shown that KMT2A is part of a large macromolecular complex composed of different proteins11 that function to improve the stability of KMT2A allowing the complex to regulate the transcriptional activation of HOX genes. The KMT2A gene can be fused in-frame with more than 120 different partners, creating fusion proteins that are typically associated with poor prognosis leukemias.6 One of the most common fusion partners is the MLLT3 gene (also known as AF9), which is found in 30% and 13% of KMT2Arearranged AML and ALL,12 respectively; this fusion is associated with an intermediate risk. The KMT2AMLLT3 (KM3) fusion protein is part of the DOT1L complex (DOTCOM) and leads to aberrant expression of specific target genes marked by H3K79 methylation.13 Despite the identification of several key gene targets required for transformation, the complete molecular mechanisms utilized by this oncogenic fusion are still unclear. The strong oncogenic potential of KMT2A fusions makes them ideal drivers for in vivo experimental model systems to explore molecular mechanisms involved in leukemogenesis.14 To specifically overcome the problems of patients’ genetic heterogeneity and scarcity of samples, we have developed a single-donor, human model leukemia system using healthy CD34+ cord blood HSPC.15 This adaptation of a prior human model16 uses a retrovirus to deliver the human KM3 fusion gene which we have shown is sufficient to generate a human leukemia.15 Transduced cells are cultured in vitro for 30-40 days prior to injection into immune-deficient mice which develop a leukemia 24-30 weeks later. Because we can sequence the transcriptome/genome/exome of the initial CD34+ cells used to generate the model AML, the genetic background (e.g., single nucleotide variants) of the initial donor is well defined, allowing the potential role of acquired mutations to be assessed precisely. In addition to these advantages, this model also allows the study of genetic mechanisms involved in the initiation of the disease, when the fusion is first introduced but before the complete transformation of the cells. Our previous genomic analyses demonstrated that the oncogenic fusion alone is sufficient for the development of leukemias, without the requirement for any recurrent secondary mutations,15 despite the presence of these in pediatric patients with KMT2A-mutated AML. haematologica | 2022; 107(1)

We have also used these data to identify a number of genes that represent novel biomarkers in patients with KMT2A fusions.17 Interestingly, while approximately onethird of these biomarker genes were expressed in our model leukemia system shortly after cells were transduced with the KM3 fusion, the majority were only expressed after xenotransplantation. This observation, coupled with the limited potential for in vitro growth of these cells, suggests that cells partially transformed (or “primed”) by KM3 may require additional in vivo signals (e.g., from the bone marrow niche) to complete their leukemic transformation. The single-donor model leukemias therefore not only recapitulate the behavior and phenotype of the disease, but also provide an experimental system to study genetic and epigenetic mechanisms involved in the disease, and a unique insight into early stages of transformation. In the present study, we leverage these advantages to perform a detailed epigenetic analysis of the various stages of the model and, through correlation with model and expression data from patients with leukemia, define the epigenetic changes driven by the oncogenic fusion that contribute to leukemia development. Analysis of changes in the patterns of DNA methylation confirmed previous observations of profound hypomethylation in KMT2A-translocated AML. Interestingly, however, B-cell ALL driven by the same fusion show a smaller number of differentially methylated cytosines relative to HSPC methylation, which are predominately hypermethylated. Despite the generally poor global correlation between DNA methylation and gene expression, we identified ADCY9, a member of the adenylyl cyclase family, as a gene essential for KM3-AML growth that also exhibits coordinated changes in DNA methylation and gene expression. We further characterized changes in chromatin accessibility through an assay for transposaseaccessible chromatin with high-throughput sequencing (ATAC-sequencing), along with alterations in histone modifications including H3K4me3 and H3K79me2. Interestingly, our ATAC-sequencing analysis revealed that the vast majority of regions of open chromatin are shared with normal HSPC and monocytes, with very few being specific to the leukemic cells. Even those regions that are unique to the leukemia show a high degree of overlap with regions of open chromatin seen in other cell types and are also characterized by the presence of known clusters of transcription factor binding sites. Integration of the epigenetic data with expression data highlighted a role for the histone demethylase KDM4B during the initial stages of leukemic transformation through its impact on S100A8/S100A9 expression levels, whose relative ratios have been demonstrated to be critical for blocking differentiation of leukemic cells.18 Collectively, these results suggest that the leukemic transformation of normal HSPC by KM3 involves a very subtle epigenetic shift that also implicates co-option of normal transcriptional networks.

Methods Patients’ samples and model leukemia generation All pediatric and adult AML patients’ samples and additional clinical information used in this study were collected by the Banque de Cellules Leucémiques du Québec (BCLQ) with

T. Milan et al.

informed consent and project approval from the Research Ethics Boards of CHU Ste-Justine, Maisonneuve-Rosemont Hospital and Université de Montréal. RNA was extracted from samples preserved in TRIzol (Invitrogen) according to standard protocols, while DNA for methylation-sequencing (methyl-sequencing) analysis was obtained from viably conserved samples. The characteristics of the patients’ AML samples are shown in Online Supplementary Table S1. The model leukemias used in this study were generated as previously described.15

Human CD34+ hematopoietic stem and progenitor cell isolation and cell culture CD34+ human HSPC were isolated from fresh umbilical cord blood, collected under informed consent at Sainte-Justine pediatric hospital (distributed by HémaQuébec’s Public Cord Blood Bank [Montreal, Quebec, Canada]) or CHU de Quebec and Hotel Dieu de Levis. CD34+ cord blood cells were positively selected as previously described15 and processed immediately without in vitro culture. Briefly, cord blood was diluted with phosphate-buffered saline (PBS) + citrate 0.6% (1:2) and mixed with Ficoll. After spinning the mixture at 2,100 rpm, for 30 min at room temperature with the lowest deceleration speed, white blood cells found at the interphase were collected and diluted with PBS (1:2). Cells were then spun at 2,100 rpm for 10 min and resuspended in PBS + citrate 0.6% + bovine serum albumin 0.5%. An isolation kit (Miltenyi Biotec; #130-100-453) composed of beads recognizing CD34+ cells was used according to the manufacturer’s protocol. The Posseld method on an AutoMACS® machine was then used to isolate CD34+ cells and the final yield was quantified using a hemocytometer. Leukemia cells lines (KG1a, THP-1, NOMO1 and MOLM-13) were grown in RPMI medium supplemented with 10% heat-inactivated fetal bovine serum, penicillin (100 U/mL) and streptomycin (100 mg/mL) at 37°C in 5% CO . Cell lines were transfected using a lentiviral protocol as previously described.15 The shRNA sequences used are shown in Table 1. 2

Chromatin immunoprecipitation, RNA-sequencing and ATAC-sequencing Cells were cross-linked using formaldehyde 1% (for 7 min, room temperature) before glycine quenching (0.125 nM, 5 min, room temperature), cell lysis and sonication (Covaris S2 sonicator). Protein G-coupled beads (Protein G Dynabeads™) were incubated overnight on a rotator at 4°C with 3 mg of antibodies against different histone marks (H3: #ab1791; H3K4me3: #ab8580-100; H3K79me2: #ab3594). Sonicated chromatin was incubated with 250 mL of beads for 4 h on a rotator at 4°C before

sequential washing with: ‘Low Salt’ buffer (0.5% NP40, 15 mM KCl, 10 mM Tris pH 8.0, 1 mM EDTA), three different ‘High Salt’ buffers (0.5% Triton, 10 mM Tris pH 8.0, 100 mM [2] or 400 mM [3] or 500 mM [4] NaCl), twice in LiCl buffer (0.5% NP40, 250 mM LiCl, 10 mM Tris pH 8.0, 1 mM EDTA). ChIPed material was then eluted (buffer: 10 mM Tris pH 8.0, 1 mM EDTA) and incubated overnight at 65°C with shaking (1,200 rpm) to reversecrosslink. An RNaseA was added (for 1 h at 37°C) and Proteinase K digestion was performed (1 h at 37°C) before DNA was purified by phenol-chloroform. A TruSeq ChIP Library Preparation Kit (Illumina, #IP-202-9001) was used to construct DNA libraries as described in the manufacturer’s instructions, with polymerase chain reaction (PCR) for a total of 12 cycles using the Illumina indexed library primers. An Illumina NextSeq 500 was used to sequence samples (75 bp paired-end) according to standard Illumina protocols. For all RNA-sequencing experiments, RNA was purified from cells (1x106) in TRIzol reagent, purified on RNeasy columns (Qiagen) and then used as input for Illumina stranded Tru-seq protocols. For CD34+ cells transduced with the KM3 fusion and cultured in vitro, RNA was collected after 30 days from GFP+ cells while for all shRNA knockdown experiments RNA was collected 4 days after lentiviral transduction. Lastly, RNA from the xenografted model AML was collected from the bone marrow of mice sacrificed 24-30 weeks after injection. All sequencing in this study was performed by the genomics core facility at the Institute for Research in Immunology and Cancer. The ATAC-sequencing protocol used was based on a method published by Buenrostro et al.19 and was performed using 50,000 cells. Samples were sequenced (75 bp paired-end) according to Illumina protocols. The primers used are shown in Table 2.

DNA methylation capture sequencing Genomic DNA from cell pellets from either KM3 patients’ samples, pooled cord blood donors (freshly isolated), cultured cells (collected after 30 days) or model AML cells (collected 24 weeks after injection) was isolated using the PureLink Genomic DNA Mini Kit according to the manufacturer’s protocol (Life Technologies, cat. #K1820-01). Three micrograms of DNA from each sample were fragmented to 140-180 bp DNA fragments using Covaris S2 (parameters: 10% duty cycle at intensity 5 for 6 cycles of 60 s with 200 cycles set at sweeping mode). Methylsequencing libraries for DNA methylation analysis were prepared using the SureSelectXT Human Methyl-Seq Target enrichment system (Agilent Technologies, cat. #5190-4661) according to the manufacturer’s instructions. Briefly, DNA was fragmented, purified and used for end-repair and adenylated prior to ligation

Table 1. List of primers used for shRNA-mediated knockdown of ADCY9 expression.

Target

Sequence (5’->3’)

Scramble ADCY9 – shRNA1 ADCY9 – shRNA2

TGCTGTTGACAGTGAGCGCCCGCCTGAAGTCTCTGATTAATAGTGAAGCCACAGATGTATTAATCAGAGACTTCAGGCGGTTGCCTACTGCCTCGGA TGCTGTTGACAGTGAGCGCGACTGTCAAAACCTTTGATAATAGTGAAGCCACAGATGTATTATCAAAGGTTTTGACAGTCTTGCCTACTGCCTCGGA TGCTGTTGACAGTGAGCGCGGGTATTATTTGACTTTTAGATAGTGAAGCCACAGATGTATCTAAAAGTCAAATAATACCCATGCCTACTGCCTCGGA

Table 2. List of primers used to prepare assays for transposase-accessible chromatin (ATAC)-sequencing libraries.

Primer name

Primer sequences (5’->3’)

Ad1_noMX Ad2.1_TAAGGCGA Ad2.2_CGTACTAG Ad2.3_AGGCAGAA Ad2.4_TCCTGAGC Ad2.5_GGACTCCT

AATGATACGGCGACCACCGAGATCTACACTCGTCGGCAGCGTCAGATGTG CAAGCAGAAGACGGCATACGAGATTCGCCTTAGTCTCGTGGGCTCGGAGATGT CAAGCAGAAGACGGCATACGAGATCTAGTACGGTCTCGTGGGCTCGGAGATGT CAAGCAGAAGACGGCATACGAGATTTCTGCCTGTCTCGTGGGCTCGGAGATGT CAAGCAGAAGACGGCATACGAGATGCTCAGGAGTCTCGTGGGCTCGGAGATGT CAAGCAGAAGACGGCATACGAGATAGGAGTCCGTCTCGTGGGCTCGGAGATGT haematologica | 2022; 107(1)

Early epigenetic changes in KM3-AML

of methylation adaptors (synthesized with 5’-methylcytosine instead of cytosine). The adaptor-ligated DNA samples were purified using AMPure XP beads (Beckman Coulter Genomics, cat. #A63880). Nine hundred nanograms of each DNA sample were hybridized with Agilent SureSelect Methyl-Seq biotinyled RNA baits for 24 h at 65°C. The libraries were captured using Dynal MyOne Streptavidin T1 magnetic beads (Invitrogen, cat. #65601). The captured DNA was subjected to bisulfite conversion with the EZ DNA Methylation Gold Kit (Zymo Research, cat. #D5005) as described in the manufacturer’s instructions, followed by PCR for a total of 16 cycles using the Illumina indexed library primers. The final libraries were quantified by 2100 Bioanalyzer chips (Agilent Technologies) with quality control on a MiSeq (Illumina) sequencer. Libraries were sequenced on multiple lanes of a HiSeq2000 flow cell at the IRIC genomics platform and the FastQC R package was used to assess the quality of the reads.

Bioinformatics analysis of ChIP-sequencing and ATACsequencing The quality of the raw sequence data was checked by FastQC (version 0.11.4). Trimmomatic (version 0.32) was used to remove adapters and keep reads with quality scores >30. Paired-end sequencing reads were then mapped to the human reference genome (GRCH37/hg19) using BWA (version 0.7.10) with the default parameters. PCR duplicates were removed by Samtools (version 1.9). For ATAC-sequencing data, reads from mitochondrial DNA were removed (samtools idxstats input.bam | cut -f 1 | grep -v MT | xargs samtools view -b input.bam > output.bam). The EaSeq software package (version 1.04)20 was used to analyze data and integrate gene expression and data from epigenetic marks from chromatin immunoprecipitation (ChIP)-sequencing experiments. To calculate the enrichment of each histone mark, the analysis focused on specific regions in the genome with a distance from the transcription start site of -1,000 to +3,000 bp for H3K79me2, and -2,000 to +2,000 bp for H3K4me3. MACS2 (version 2.2.4)21 was the peak calling algorithm used to determine the list of significant enrichment peaks, following the default parameters (callpeak –f BAMPE –g hs –q 0.05). Intervene22 was used to find peaks in common in our datasets (bedtools intersect –a – b –f 0,50 –r) and generate Venn diagrams (Pybedtools). A ngs.plot23 was used to visualize global repartition of mapped reads across the genome. Previously published data for lineagespecific ATAC-sequencing in a range of normal blood cells24 were accessed through the UCSC track hub (https://s3-us-west-1.amazonaws.com/chang-public-data/2016_NatGen_ATAC-AML/hub.txt) DNAse hypersensitive sites and transcription factor clusters generated from the ENCODE consortium V3 were analyzed by custom Perl Scripts. Public ATAC-sequencing peaks for all cell types were identified using a threshold of a normalized score >60 for all tracks for starts/ends of peaks. Fastq files for BLUEPRINT ATAC-sequencing data (accessions EGAD00001002710 and EGAD00001002709) were downloaded, remapped and analyzed with MACS2 as described above. Gene set enrichment analysis was performed using software (version 3.0) from the Broad Institute25 and a pre-ranked list based on significantly differentially expressed genes between CD34+ and CD34+KM3 cells using the R package DESeq2.26

Mapping methyl-sequencing reads and methylation analysis Paired end sequencing reads (2x100 bp) were mapped to the human reference genome (GRCH37/hg19) using Bismark with a directional alignment mode and Bowtie2 as the alignment process. All putative duplicate reads were removed by previous-

haematologica | 2022; 107(1)

ly described Perl scripts.27 The methylation calls were extracted with Bismark-methylation-extractor and the analysis was performed using custom python scripts, R package MethylKit 0.9.2 and SeqMonk 1.45.2 as the visualization tool (www.bioinformatics.babraham.ac.uk/projects/seqmonk/). Hierarchical clustering of the samples based on the differentially methylated cytosines was performed using the computed Pearson correlation distance between samples. For each cell type (CD34+, CD34+KM3, KM3AML model) the DNA methylation data corresponded to the combination of methyl-sequencing libraries from two biological replicates. Analysis of methylated cytosines was restricted to positions covered by a minimum of eight reads coverage for each biological replicate. To identify differentially methylated cytosines (DMC) a logistic regression model was applied to compare the fraction of methylated cytosines across the three types of cells and a sliding linear model (SLIM) method was used to correct P-values for multiple testing and derive q-values.28 To define the DMC, we specified two parameters: q-values below 0.01 and a threshold of percentage methylation difference ≤40%. All functional annotations for genomic regions were defined based on GRCh37/hg19 assembly downloaded from the UCSC website. Gene bodies were defined as the transcribed regions from the start to the end of transcription sites for each RefSeq entry and promoter regions were defined as 1 kb of sequence upstream of the transcription start site for each RefSeq transcript.

Results Transcriptional consequences of KM3 expression To characterize the short-term transcriptional changes induced by the KM3 fusion, we analyzed the RNAsequencing data from the initial unmanipulated cord blood cells and these cells after their transduction with the KM3 fusion. As previously described,16 untransduced cells are largely lost in culture after about 1 month of in vitro culture and the remaining cells show induction of genes associated with a myeloid phenotype (e.g., CCR1, CD14) (Figure 1A). Interestingly, transduced cells also display the loss of normal or leukemia stem cell markers (e.g., CD34 and GPR56, respectively) and the loss of expression of other genes previously associated with AML induction in other subtypes (e.g., MN130). This observation is supported by expression data from patients with KMT2A-translocated AML in the TARGET study,4 which showed an equivalent loss of MN1 and GPR564 (Figure 1B) and a similar loss of expression in studies of GPR56 seen in adult AML.31 A global view of the expression changes through gene set enrichment analysis confirmed the loss of stem cell phenotype and the gain in expression of HOXA9-MEIS1 targets consistent with leukemic transformation (Online Supplementary Figure S1). To examine the regulation underlying the transcriptional changes seen, we studied the transcription factor binding motifs upstream of the differentially expressed genes. Analysis of enriched motifs using several different software (including HOMER,32 MAGICTRICKS,33 and ShinyGO34) produced differing lists of potential transcription factors (Figure 1C), but factors in the CEBP and FOS families, along with NFE2, were highlighted consistently, at differing ranks. Analysis of RNA-sequencing data confirmed that the expression of all members of the CEBP and FOS families increased during in vitro culture (Figure 89

T. Milan et al.

1D), along with transient changes in ETS, MAF, and JUN families (Online Supplementary Figure S2). We also observed an increase in the expression of S100A9 and S100A8 (Figure 1E), which have been described as direct targets of CEBPb.35 It has been recently demonstrated that the ratio of these calcium-binding proteins is critical for blocking AML differentiation,18 with S100A8 opposing the differentiation promoted by S100A9, which acts through TLR4. Interestingly, and in agreement with these

findings, while the expression of both S100A8 and S100A9 increased when KM3 was added to CD34+ cells, there was a larger increase in S100A8, changing the ratio of S100A8:S100A9 from 1.1:1 to 3.2:1, favoring inhibition of differentiation. Other CEBP members, including CEBPb, have also previously been implicated in the process of normal myeloid differentiation,36 which suggests that transcription factors normally present are coopted to activate genes that are part of a leukemic tran-

Figure 1. Continued on following page.

haematologica | 2022; 107(1)

Early epigenetic changes in KM3-AML

Figure 1. Gene expression changes in the KMT2A-MLLT3 leukemia model. (A) Volcano plot of differentially expressed genes (red dots are genes with a P value <10-5, and a log fold change >3) from RNA-sequencing data of CD34+ cells and CD34+KM3 cells. Highlighted genes include known markers for stem cells, myeloid cells or previously reported biomarkers for KM3-AML. (B) Scatterplot of MN1 and GPR56 expression in approximately 1,000 cases of pediatric AML in the TARGET cohort. Red dots represent patients with KMT2A-rearranged AML. (C) Summary of transcription factor binding motifs found to be enriched in promoters of upregulated genes in (A) with either ShinyGO v0.6134 or MAGICTRICKS.33 (D) Gene expression levels of transcription factor families highlighted in (C) in either CD34+ cells (blue) or CD34+KM3 cells (red). (E) High levels of expression of TLR4, S100A8 and S100A9 are seen after introduction of the KMT2A-MLLT3 fusion and are maintained in the model AML. (F) Loss of KDM4B expression through shRNA knockdown in the KMT2A-MLLT3+ cell line (THP-1) induces growth inhibition (*P<0.005, **P<0.001) and (G) reduced expression of S100A8/A9, CEBPb and other cell-cycle related genes. (H) shRNA knockdown of KDM4B in CD34+ cells transduced with the KM3 fusion show gene expression decreases very similar to those in THP-1 cells. FPKM: fragments per kilobase of exon per million mapped reads 2

scriptional program. At the same time, these factors and others upregulated are also involved in immune and inflammatory pathways, suggesting that there may be additional transcriptional programs that are activated and superimposed on the myeloid differentiation that we observe. To determine whether changes in the expression of chromatin-modifying enzymes might be correlated with our RNA-sequencing data, we looked for enzymes whose expression increased at least 2-fold during in vitro culture and whose expression was maintained in both the model and patients’ AML (Online Supplementary Table S2). This analysis highlighted a single gene with this profile, KDM4B, which has been implicated in murine AML, albeit with functional redundancy.37 The shRNA-mediathaematologica | 2022; 107(1)

ed knockdown of KDM4B in THP-1 cells showed a significant impact on growth (Figure 1F) and RNA-sequencing from these cells revealed a profound downregulation of S100A8 and S100A9, CEBPb, P2RY2 (a KM3-AML biomarker17), and other cell cycle regulators (Figure 1G). We next performed the same experiment but with CD34+ cells transduced with the KM3 fusion to validate the changes in our model system. The data showed very similar gene expression changes in these genes with the exception of CEBPb, although other CEBP family members were downregulated (Figure 1H). Together, these data support the hypothesis that the early epigenetic changes, at least partially mediated through KDM4B, lead to activation of S100A8/9 as well as other genes required 91

T. Milan et al.

for sustained proliferation. Given the similarity of their DNA binding consensus sequences, defining the precise contribution of different CEBP family members in this process will require further focused studies.

Sequential changes in DNA methylation are not globally correlated with expression changes Given the known role of KM3 as an epigenetic regulator and the demonstrated importance of DNA methylation changes in other subtypes of AML,38,39 we next sought to investigate changes in DNA methylation in our model leukemias and genetically matched patients’ samples. Using methyl-sequencing, capture-based, bisulfite sequencing,40 we assessed DNA methylation levels genome-wide at 84 Mbp of target sequence regions including CpG islands, promoters and published differentially methylated regions. Biological duplicates at each step of our model leukemia system were compared to bisulfite-treated DNA from three adult KM3-AML patients. In line with expression data from our model leukemias,15 the methylation data between replicates were highly consistent (Online Supplementary Figure S3) and were also consistent with published data regarding global methylation levels (Online Supplementary Figure S4).41 Hierarchal clustering of DMC (CpG coverage ≥8, qvalue ≤0.1 and methylation differences ≥40%) demonstrated that relatively small numbers of changes occurred after introduction of the KM3 fusion gene and in vitro culture (Figure 2A). The number of genes with DMC present varied from 7,645 (CD34+ vs. CD34+KM3) to 13,947 (CD34+ vs. model AML); however a large percentage of these genes (40% and 20%, respectively) contained only a single DMC. Interestingly, relative to the initial CD34+ cells with the KM3 fusion, there was a striking increase (>6 fold) in the number of DMC detected in the final model leukemias and patients’ samples. This suggests that the initial epigenetic changes induced by the fusion gene are relatively limited, with the majority of changes occurring only during in vivo growth. Along with the previous stepwise changes in gene expression patterns we have reported,15 these results again support the importance of microenvironmental cues received in vivo. With respect to the direction of change, in agreement with recent observations in patients with KMT2A-AML,41 DMC observed in both model and patient KM3-AML were profoundly biased (~89%) towards demethylation (Figure 2B). In opposition to this trend, model leukemias generated from the same donor cells but which developed a lymphoid phenotype (B-cell ALL) had primarily hyper-methylated cytosines, and fewer DMC compared to the AML samples. The majority (>75%) of DMC were located in gene bodies and particularly in promoter regions and putative functional elements at various stages of the model AML and in patients (Online Supplementary Figures S5 and S6). To clarify the importance of the localization of these changes and their potential impact, we next examined the correlation between DNA methylation changes and gene expression. Despite the general view regarding the impact of DNA methylation changes in the promoter regions and transcription levels, we observed only a weak global correlation between the two (Figure 2C). While we could find examples of expected changes (e.g., in previously described KMT2A target genes such as GATA2, 92

HOXA7 and HOXA942,43) (Figure 2D), overall, there was a very poor global correlation. This observation mirrors those made in recent comprehensive single-cell studies of DNA methylation changes during normal hematopoiesis in humans44 as well as previous bulk studies.45 We were able to identify new examples of genes with evidence for epigenetic regulation such as the atypical adenylate cyclase gene ADCY9. The promoter region of ADCY9 exhibits hypomethylation specifically in the model AML compared to normal CD34+ cells and model B-cell ALL (Figure 2E) and the gene is only expressed in the KMT2Arearranged model and patients’ leukemias (Figure 2F). ADCY9 is a poorly characterized and distantly related member of the adenylate cyclase gene family and no specific inhibitors have yet been described, although it was recently identified as a fusion partner of the KMT2D gene in a case of ALL.46 We therefore performed shRNA knockdowns in several KM3-AML cell lines to determine whether ADCY9 expression was relevant to these leukemias. We found that all the KM3-AML lines showed impaired proliferation with multiple shRNA in contrast to no effect in a non-KM3-AML control line (Figure 2G), demonstrating that ADCY9 is required in these cells. Beyond regulating specific genes, we next wanted to examine whether global methylation changes might have an impact on gene expression through their localization within the DNA binding sites of transcription factors. To do this, we selected all the DMC with coverage that exhibited consistently higher (≥40%) methylation levels in CD34+ cells than in either CD34+KM3 or model leukemias. Transcription factors in the ENCODE ChIPsequencing dataset (covering 338 factors in 130 cell lines) that overlapped the positions of these cytosines were then used to generate a network (using GOnet47) based on shared gene ontology terms (Online Supplementary Figure S7). This network contains a large collection of factors that: (i) are highly expressed in normal bone marrow, (ii) exhibit a large loss of DNA methylation directly in their binding sites after KM3 addition, and (iii) are enriched for activity in regulating hematopoietic and myeloid differentiation. The results indicate that despite the poor overall correlation between DNA methylation and the expression of individual genes, the changes that occur are globally relevant for leukemic transformation.

Epigenetic analysis reveals potential CCR1-CCL23 autocrine signaling in KM3-acute myeloid leukemia Given that the KM3 fusion is a well-known epigenetic regulator, we next asked whether the presence of the gene fusion had a specific impact on the chromatin organization in our model leukemias. To investigate this, we performed ChIP-sequencing on relevant histone marks (H3K4me3, H3K79me211) and ATAC-sequencing on samples of each step of our leukemia model (CD34+, CD34+KM3, KM3-AML). Enrichment of the different histone marks correlated with gene expression levels (Online Supplementary Figure S7) and expected spatial distribution, and the majority of peaks seen in the model leukemias were common to both the initial and transduced cells (Figure 3A). In addition, we analyzed the combined expression and epigenetic data to compare the initial CD34+ cells and CD34+KM3 cells to highlight critical early events potentially implicated in the disease. Using the EaSeq package20 we identified a set of genes (n=331) (Online Supplementary Table S2) with coordinated epigehaematologica | 2022; 107(1)

Early epigenetic changes in KM3-AML

Figure 2. DNA methylation changes in model leukemias. (A) Hierarchical clustering of differentially methylated cytosines (DMC) identified by methyl-sequencing analysis of model leukemias at different stages and KM3 patients’ samples (pAML). Numbers and arrows represent the number of DMC relative to the initial CD34+ cells. (B) The KM3 fusion induces a strong hypomethylation phenotype in both patient and model acute myeloid leukemias (pAML and mAML, respectively), whereas model B-cell acute lymphoblastic leukemias (mB-ALL) predominately exhibit hypermethylation. (C) As noted in other global studies, the overall correlation between DNA methylation and expression is relatively limited, likely due in part to the complexity of the localization and size of the changes. (D) Differentially methylated regions around genes known to be suppressed (GATA2) or activated (HOXA9) in AML can be seen comparing CD34+ cells versus AML patients’ cells. (E, F) ADCY9 exhibits hypomethylation in its promoter region specifically in the model AML, compared to CD34+ cells, which is correlated with gene expression levels, specifically in KMT2A-MLLT3 AML models as well as patients (E) and external data from TARGET (F). (G) Knockdown of ADCY9 using multiple shRNA shows a significant impact specifically on KM3-AML cell proliferation but not on AML cells lacking the fusion (KG1a; *P<0.005, **P<0.001).

haematologica | 2022; 107(1)

T. Milan et al.

netic and expression changes (Figure 3B). Among the genes with the most striking changes was the C-C motif chemokine receptor 1 (CCR1) gene, a G-protein coupled receptor family member. CCR1 is expressed in different types of blood cells, can bind a number of different ligands

and is induced during normal monocyte development.48 This gene was of specific interest because our previous work had shown that CCL23, a ligand only recognized by CCR1, is also a biomarker for KMT2A-translocated AML.15 RNA-sequencing data from our models and genetically

Figure 3. Legend on following page.

haematologica | 2022; 107(1)

Early epigenetic changes in KM3-AML

Figure 3. Epigenetic analysis changes in KMT2A-MLLT3 acute myeloid leukemia. (A) Venn diagram of peaks determined by MACS2 in different stages of the model system (CD34+, CD34+KM3, KM3 AML model) for H3K4Me3 and H3K79Me2 (n=3 for chromatin immumoprecipitation [ChIP]-sequencing of each mark). (B) A scatterplot of ChIP-sequencing enrichments for each histone mark in an area ± 2,000 bp (H3K4me3) or -1,000 to +3000 bp (H3K79me2) around the transcription start site for each gene is shown, with individual genes colored by the log ratio (CD34+/CD34+KM3) of expression for each gene. (C) Boxplots show the expression of CCR1 and its ligands in different stages of the model leukemias and in pediatric KMT2A-AML patients. The dotted green line represents an arbitrary threshold for an expressed gene (FPKM >1) (D) Annotation of ATAC-sequencing peaks unique to model leukemias (mAML) show that the majority are either in intronic or intergenic regions. (E) Hierarchical clustering of ATAC-sequencing peaks in published data from different normal blood cell types compared to leukemic cells. NK: natural killer cells; CD4T: CD4+ T lymphocytes; CD8T: CD8+ T lymphocytes; ERY: erythrocytes; AML: acute myeloid leukemia (model); Mono: monocytes; CD34 HSPC: CD34+ hematopoietic stem and progenitor cells; GMP: granulocyte-monocyte progenitor cells; B: B lymphocytes; CLP: common lymphoid progenitor; MEP: megakaryocyteerythroid progenitor cells; HSC: hematopoietic stem cells; MPP: multipotential progenitor cells; LMPP: lymphoid-primed multipotent progenitors cells; CMP: common myeloid progenitor cells. (F) A list of transcription factors is shown ranked by their abundance in the mAML and with their corresponding rank in the entire dataset and their expression level in the model system stages in the columns to the right. (G) UCSC snapshot showing an example of a gene with ATAC-sequencing peaks in CD34+KM3, mAML, CD34+ HSPC, hematopoietic stem cells and monocytes. The blue vertical line shows a peak that is unique to the mAML samples, whereas red vertical lines highlight common peaks that also overlap ENCODE transcription factor clusters. 2

haematologica | 2022; 107(1)

T. Milan et al.

matched patients’ samples confirmed that the expression of both CCR1 and CCL23 is increased dramatically upon the integration of the KM3 fusion, while neither is expressed in normal CD34+ cells (Figure 3C). External RNA-sequencing data from the BloodSpot database (Online Supplementary Figure S9A) or TARGET cohort4 also confirm the KMT2A-specific expression of CCL23 (Online Supplementary Figure S9B) and there is a small but significant difference in survival of pediatric AML patients with high levels of CCL23 (Online Supplementary Figure S9C). Although previous studies have highlighted the importance of CCR1 in different cancers (e.g., breast49 and prostate50 cancer) no studies have looked at its role in AML. Although we saw an inhibition of growth specifically in KM3-AML cell lines using a potent inhibitor of CCR1 (data not shown), this inhibition was only observed at levels (1020 mM) much higher than the reported half maximal inhibitory concentration (IC ) (9 nM) of the inhibitor.51 The very high expression levels of both the receptor and ligand may explain this observation; however, specific genetic models will be required to define the biological importance of this potential autocrine-signaling axis. 50

KM3-mediated epigenetic changes are minimal relative to normal blood cells Location analysis of the annotated ATAC-sequencing peaks unique to the model AML cells revealed the expected peaks at promoters of active genes, although the majority of open chromatin regions were either intergenic or intronic (Figure 3D). As a point of comparison for our ATAC-sequencing data, we downloaded previously published and normalized ATAC-sequencing data generated from normal blood cells24 along with raw ATACsequencing data from normal CD34+ cells and monocytes from the BLUEPRINT project.52 Using peaks identified via MACS2 in our ATAC-sequencing data or remapped published data, or through supervised thresholding of normalized UCSC tracks, we compared regions of open chromatin in normal and leukemic blood cells. As with the histone modifications, the majority of peaks identified in the model leukemia samples were shared with one or more normal types of blood cell. Considering the 10,415 peaks identified by MACS2 that were specific to the model AML (Online Supplementary Figure S10), analysis of potential transcription factor binding motifs once again identified factors, such as PU.1, that are critical for normal myeloid cell differentiation (Online Supplementary Figure S11). Hierarchical clustering of all ATAC-sequencing peaks from the CD34+KM3 and model AML revealed that the chromatin landscape in the model AML is close to that of normal monocytes and CD34+ HSPC (Figure 3E). The clear separation of CD34+KM3 cells, despite the small absolute number of unique peaks (and more peaks common to CD34+ HPSC than to monocytes), suggests that the chromatin of these cells is subtly affected by their culture in vitro. Using data from the ENCODE consortium, we further examined the transcription factors that have been experimentally observed to bind to leukemia-specific open chromatin regions. For this, we ranked the abundance of individual factors in the entire dataset, and in the leukemia-specific peaks, to account for the difference in frequencies of binding sites for individual factors (Figure 3F). This analysis revealed an enrichment of factors in the leukemia-specific peaks that did not necessarily correlate 96

with an increase in expression of the factor. Given the ChIP-sequencing data, this suggests that the novel, leukemia-specific peaks may create competition for the binding of a range of expressed transcription factors, potentially co-opting their activity. In other cases, while there was no difference in the ranking, the factor itself was more highly expressed; in cases in which such factors have been demonstrated to be essential for leukemia development (e.g., FOS53), the higher expression level may be important to ensure a high/continuous occupancy at the open regions in order to drive the leukemic transcriptional program. Across the genome, we observed many examples of genomic regions in which ATAC-peaks common to both leukemic and normal cells were seen in the vicinity of other leukemia-specific peaks (Figure 3G). Given our observations, we questioned whether all of the leukemia-specific peaks we observed were potentially functional or whether they might represent stochastic changes in accessibility. To address this question, we compared data for chromatin accessibility (DNaseI hypersensitive sites) and ChIP-sequencing data from the ENCODE consortium to the ATAC-sequencing peaks that were unique to the model leukemias. Interestingly, the peaks determined by either MACS2 or supervised thresholding showed high levels of overlap (65% and 75%, respectively) with both previously defined transcription factor clusters, DNaseI hypersensitive sites and overlapped chromatin marks associated with enhancers seen in other cell types. These data argue that despite their unique presence in leukemic cells, the large majority of leukemia-specific open chromatin regions likely represent enhancers.

Discussion The KMT2A-MLLT3 gene fusion is one of the most common genetic anomalies found in pediatric AML and is an epigenetic regulator that is essential, directly or indirectly, for the alterations of chromatin structure13,54,55 and DNA methylation seen in the disease.56 Despite its extensive characterization, less is known regarding its role in the initial transformation of HSPC into leukemic cells. Our current study leverages our previously described human model leukemia system to integrate epigenetic and expression changes to better characterize these changes. With respect to DNA methylation changes, our data confirm previously published results57 showing that an aberrant hypomethylation profile is found in KM3 leukemias. Interestingly, relative to the initial CD34+ cells, CD34+KM3 and B-cell ALL have a hypermethylation phenotype, whereas the model leukemia and patients’ AML show the reverse. While the significance of all of these global differences is not yet clear, previous singlecell studies showed that differentially methylated promoter regions in normal blood cells were enriched in binding sites for transcription factors promoting myeloid differentiation.44 Our data (e.g., Online Supplementary Figure S7) agree with this finding, which may reflect a default lymphoid differentiation pathway in the absence of demethylation of these regions. Recent studies have highlighted the importance of DNA methyltransferase in KM3-mediated leukemogenesis in which DNMT3A and DNMT3B loss-of-function mutations accelerate leukemia haematologica | 2022; 107(1)

Early epigenetic changes in KM3-AML

progression by enhancing cell cycle progression.58 Forced expression of DNMT3B delayed KM3 leukemia development59 whereas the deletion of DNMT1 prevented KM3 AML progression.60 These findings and our own data demonstrate that the loss of de novo methylation, but not necessarily maintenance methylation, is required for the induction of AML. Analysis of the correlated epigenetic and gene expression changes highlighted a number of genes of interest, including ADCY9, an atypical member of the adenylate cyclase family. ADCY9 shows low but relatively broad expression in most tissues except blood (of all lineages61) (Online Supplementary Figure S12), and is capable of inducing cAMP accumulation,62 although its activation with G protein subunits appears to be cell-type specific. Our functional studies demonstrate that ADCY9 function is critical for the growth of KM3-AML cells, and levels of cAMP in leukemias63,64 have been shown to be important. Nevertheless, whether the growth defects seen in response to the loss of ADCY9 expression are mediated through its ability to produce cAMP remains unclear and will require further study. The integration of our data also highlighted the potential importance of the chemokine receptor gene CCR1 and one of its ligands, CCL23. This receptor-ligand pair were noteworthy in part because of their strong level of upregulation, but also because the expression of both increased immediately after the addition of the gene fusion, prior to the complete transformation of the cells. The CCR1 receptor can recognize at least ten different ligands including CCL3 and CCL5, which are both ubiquitously expressed in normal blood cells. In contrast, CCL23 is only expressed at very low levels in promyelocytes65 (LinFSChiSSCintCD34-CD15intCD49dhiCD33hiCD11b-CD16-) (Linand granulocyte-monocyte progenitors65 + + + + CD34 CD38 CD45RA CD123 ) but is highly expressed in KMT2A-AML according to a number of independent datasets. The impact of CCR1 signaling in AML remains poorly defined, although one consequence appears to be activation of the Ga pathway resulting in the inhibition of cAMP production.66 In this context, it is interesting to note that ADCY9 is specifically upregulated in KMT2AAML (Figure 2F), potentially as counterbalance to the activity of CCR1 and Ga and to maintain cAMP levels. Perhaps most surprisingly with respect to chromatin structure and accessibility, while we expected to see widespread changes, in fact there were relatively few modifications specific to the model leukemia cells compared to normal blood cells. This observation agrees with those of a recent study showing that KMT2A-MLLT3 helps to preserve gene expression of the cellular states in which it is expressed.67 As noted, most of the leukemiaspecific regions overlap open chromatin regions and transcript factor clusters seen in large-scale ENCODE datasets. Although speculative, these observations suggest the interesting hypothesis that there may be a finite number of regions in the genome that are competent to be open in vivo across all cell types. The addition of the KM3 fusion, while activating the expression of a subset of genes, may also cause re-localization of chromatin, leading to the opening of regions not normally seen in blood cells. Thus, while chromatin remodeling is required to give access to endogenously expressed transcription factors in some of the novel locations that we characterized in our study, others may simply be a response to this i

haematologica | 2022; 107(1)

remodeling. The distribution of ATAC peaks that we observed, largely in introns and intergenic regions (Figure 3D), also raises the question of the potential biological role as novel enhancers. Similar distributions in ATACsequencing data have been noted68 and intronic enhancers have been shown to regulate the expression of genes in a tissue-specific manner,69 which may explain why many of the ATAC-sequencing peaks not present in blood cells nevertheless corresponded to open chromatin regions in other cell types. These results open a new avenue to look at some of these regions of potential interest in the context of AML, to better understand their role in enabling access to genes which may be essential for leukemic growth. In conclusion, here we have leveraged the advantages of our model system to characterize the immediate impact of the KM3 fusion without the complicating factor of additional mutations and other genetic abnormalities typically found in patients’ genomes. While our results highlight the subtlety of the changes that occur once the oncogene is introduced, they also support the role of previous genes implicated in leukemogenesis and highlight a novel receptor/ligand pair with potential functional relevance for KMT2A-AML. The novel insight into the nature and extent of epigenetic reprogramming that occurs in the disease will help to guide future studies to identify novel potential therapeutic targets. Disclosures No conflicts of interest to disclose. Contributions BTW and TM designed the study, analyzed data and drafted the manuscript. TM and KL performed ChIP-sequencing and ATAC-sequencing, analyzed the data and generated figures. MC performed methyl-sequencing experiments, analyzed data, generated figures and drafted the manuscript. SS-T-H assisted with the ATAC-sequencing analysis and generation of figures. KL, ER, EB, and AB helped with molecular biology work, cloning and cell cultures. EB, AB and FB generated the model on mice from CD34+ cells. SM provided expression and clinical data for patients’ samples and JH and SC collected, characterized, and banked the local patients’ samples used in this study. All authors discussed the results and participated in the revision of the manuscript. Acknowledgments We would like to thank the clinicians and nurses at CHU Sainte-Justine, CHU de Québec, Hotel Dieu de Lévis for cord blood collection and we also wish to acknowledge the contribution of all of the courageous patients who provided samples used in this study. Finally, we would like to acknowledge Jennifer Huber, Sarah Boissel, Raphaëlle Lambert, Danièle Gagné and Annie Gosselin for their expert technical assistance with experimental work. Funding This work was supported by grants from the Cole Foundation (to TM, fellowship), the Impact Grant of the Canadian Cancer Society in collaboration with The Cole Foundation, Molson Foundation, R. Howard Webster Foundation, Mirella and Lino Saputo Foundation, Fonds de Recherche du Quebec - Santé, Faculté de Médicine, Université de Montréal, Letko Brosseau, Birks Family Foundation, Maryse and William Brock, CHU Sainte-Justine Foundation, Montreal Children's Hospital 97

T. Milan et al.

Foundation, Morris and Rosalind Goodman Family Foundation, Zeller Family Foundation, David H. Laidley Foundation, Drummond Foundation, and the Henry and Berenice Kaufmann Foundation (grant 705047-IMP-17; to BW, FB, SC), and the Leukemia & Lymphoma Society of Canada (to FB). Human leukemia specimens were collected and analyzed by the Banque de Cellules Leucémiques du Québec, supported by the Cancer

References 1. Andersson AK, Ma J, Wang J, et al. The landscape of somatic mutations in infant MLLrearranged acute lymphoblastic leukemias. Nat Genet. 2015;47(4):330-337. 2. Saultz JN, Garzon R. Acute myeloid leukemia: a concise review. J Clin Med. 2016;5(3):33. 3. De Kouchkovsky I, Abdul-Hay M. Acute myeloid leukemia: a comprehensive review and 2016 update. Blood Cancer J. 2016;6(7):e441. 4. Bolouri H, Farrar JE, Triche T Jr, et al. The molecular landscape of pediatric acute myeloid leukemia reveals recurrent structural alterations and age-specific mutational interactions. Nat Med. 2018;24(1):103-112. 5. Cancer Genome Atlas Research Network. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med. 2013;368(22):2059-2074. 6. Winters AC, Bernt KM. MLL-Rearranged leukemias-an update on science and clinical approaches. Front Pediatr. 2017;5:4. 7. Sam TN, Kersey JH, Linabery AM, et al. MLL gene rearrangements in infant leukemia vary with age at diagnosis and selected demographic factors: a Children's Oncology Group (COG) study. Pediatr Blood Cancer. 2012;58(6):836-839. 8. Gu Y, Nakamura T, Alder H, et al. The t(4;11) chromosome translocation of human acute leukemias fuses the ALL-1 gene, related to Drosophila trithorax, to the AF-4 gene. Cell. 1992;71(4):701-708. 9. Tkachuk DC, Kohler S, Cleary ML. Involvement of a homolog of Drosophila trithorax by 11q23 chromosomal translocations in acute leukemias. Cell. 1992;71(4): 691-700. 10. Hsieh JJ, Ernst P, Erdjument-Bromage H, Tempst P, Korsmeyer SJ. Proteolytic cleavage of MLL generates a complex of N- and C-terminal fragments that confers protein stability and subnuclear localization. Mol Cell Biol. 2003;23(1):186-194. 11. Slany RK. The molecular biology of mixed lineage leukemia. Haematologica. 2009;94 (7):984-993. 12. Meyer C, Burmeister T, Groger D, et al. The MLL recombinome of acute leukemias in 2017. Leukemia. 2018;32(2):273-284. 13. Bernt KM, Zhu N, Sinha AU, et al. MLLrearranged leukemia is dependent on aberrant H3K79 methylation by DOT1L. Cancer Cell. 2011;20(1):66-78. 14. Milne TA. Mouse models of MLL leukemia: recapitulating the human disease. Blood. 2017;129(16):2217-2223. 15. Barabe F, Gil L, Celton M, et al. Modeling human MLL-AF9 translocated acute myeloid leukemia from single donors reveals RET as a potential therapeutic target. Leukemia. 2017;31(5):1166-1176. 16. Barabe F, Kennedy JA, Hope KJ, Dick JE. Modeling the initiation and progression of human acute leukemia in mice. Science. 2007;316(5824):600-604.

Research Network of the Fonds de Recherche du Québec en Santé. Data-sharing statement All sequencing data for this manuscript has been deposited in the Gene Expression Omnibus and is accessible through the following accession numbers: GSE142579 and GSE70755.

17. Lagace K, Barabe F, Hebert J, Cellot S, Wilhelm BT. Identification of novel biomarkers for MLL-translocated acute myeloid leukemia. Exp Hematol. 2017;56:58-63. 18. Laouedj M, Tardif MR, Gil L, et al. S100A9 induces differentiation of acute myeloid leukemia cells through TLR4. Blood. 2017;129(14):1980-1990. 19. Buenrostro JD, Wu B, Chang HY, Greenleaf WJ. ATAC-seq: a method for assaying chromatin accessibility genome-wide. Curr Protoc Mol Biol. 2015;109:21. 20. Lerdrup M, Johansen JV, Agrawal-Singh S, Hansen K. An interactive environment for agile analysis and visualization of ChIPsequencing data. Nat Struct Mol Biol. 2016;23(4):349-357. 21. Thomas R, Thomas S, Holloway AK, Pollard KS. Features that define the best ChIP-seq peak calling algorithms. Brief Bioinform. 2017;18(3):441-450. 22. Khan A, Mathelier A. Intervene: a tool for intersection and visualization of multiple gene or genomic region sets. BMC Bioinformatics. 2017;18(1):287. 23. Shen L, Shao N, Liu X, Nestler E. ngs.plot: quick mining and visualization of next-generation sequencing data by integrating genomic databases. BMC Genomics. 2014;15:284. 24. Corces MR, Buenrostro JD, Wu B, et al. Lineage-specific and single-cell chromatin accessibility charts human hematopoiesis and leukemia evolution. Nat Genet. 2016;48(10):1193-1203. 25. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A. 2005;102(43):15545-15550. 26. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. 27. Krueger F, Andrews SR. Bismark: a flexible aligner and methylation caller for bisulfiteSeq applications. Bioinformatics. 2011;27 (11):1571-1572. 28. Wang HQ, Tuominen LK, Tsai CJ. SLIM: a sliding linear model for estimating the proportion of true null hypotheses in datasets with dependence structures. Bioinformatics. 2011;27(2):225-231. 29. Pennacchio LA, Ahituv N, Moses AM, et al. In vivo enhancer analysis of human conserved non-coding sequences. Nature. 2006;444(7118):499-502. 30. Sharma A, Jyotsana N, Gabdoulline R, et al. MN1 is indispensable for MLL-rearranged acute myeloid leukemia. Haematologica. 2020;105(5):1294-1305. 31. Pabst C, Bergeron A, Lavallee VP, et al. GPR56 identifies primary human acute myeloid leukemia cells with high repopulating potential in vivo. Blood. 2016;127(16): 2018-2027. 32. Heinz S, Benner C, Spann N, et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory ele-

ments required for macrophage and B cell identities. Mol Cell. 2010;38(4):576-589. 33. Roopra A. MAGICTRICKS: a tool for predicting transcription factors and cofactors that drive gene lists. bioRxiv. 2019:492744. 34. Ge SX, Jung D. ShinyGO: a graphical enrichment tool for ani-mals and plants. bioRxiv. 2018:315150. 35. Endoh Y, Chung YM, Clark IA, Geczy CL, Hsu K. IL-10-dependent S100A8 gene induction in monocytes/macrophages by doublestranded RNA. J Immunol. 2009;182(4): 2258-2268. 36. Tsukada J, Yoshida Y, Kominato Y, Auron PE. The CCAAT/enhancer (C/EBP) family of basic-leucine zipper (bZIP) transcription factors is a multifaceted highly-regulated system for gene regulation. Cytokine. 2011;54(1):6-19. 37. Agger K, Miyagi S, Pedersen MT, et al. Jmjd2/Kdm4 demethylases are required for expression of Il3ra and survival of acute myeloid leukemia cells. Genes Dev. 2016;30(11):1278-1288. 38. Sandoval JE, Huang YH, Muise A, Goodell MA, Reich NO. Mutations in the DNMT3A DNA methyltransferase in acute myeloid leukemia patients cause both loss and gain of function and differential regulation by protein partners. J Biol Chem. 2019;294 (13):4898-4910. 39. Ley TJ, Ding L, Walter MJ, et al. DNMT3A mutations in acute myeloid leukemia. N Engl J Med. 2010;363(25):2424-2433. 40. Lee EJ, Luo J, Wilson JM, Shi H. Analyzing the cancer methylome through targeted bisulfite sequencing. Cancer Lett. 2013;340 (2):171-178. 41. Akalin A, Garrett-Bakelman FE, Kormaksson M, et al. Base-pair resolution DNA methylation sequencing reveals profoundly divergent epigenetic landscapes in acute myeloid leukemia. PLoS Genet. 2012;8(6):e1002781. 42. Celton M, Forest A, Gosse G, et al. Epigenetic regulation of GATA2 and its impact on normal karyotype acute myeloid leukemia. Leukemia. 2014;28(8):1617-1626. 43. Jung N, Dai B, Gentles AJ, Majeti R, Feinberg AP. An LSC epigenetic signature is largely mutation independent and implicates the HOXA cluster in AML pathogenesis. Nat Commun. 2015;6:8489. 44. Farlik M, Halbritter F, Muller F, et al. DNA methylation dynamics of human hematopoietic stem cell differentiation. Cell Stem Cell. 2016;19(6):808-822. 45. Hodges E, Molaro A, Dos Santos CO, et al. Directional DNA methylation changes and complex intermediate states accompany lineage specificity in the adult hematopoietic compartment. Mol Cell. 2011;44(1):17-28. 46. Mankuzhy NP, Walling E, Anderson B, Mody R. Cryptic ETV6-ABL1 fusion and MLL2 truncation revealed by integrative clinical sequencing in multiply relapsed acute lymphoblastic leukemia. J Pediatr Hematol Oncol. 2019;41(8):653-656. 47. Pomaznoy M, Ha B, Peters B. GOnet: a tool

haematologica | 2022; 107(1)

Early epigenetic changes in KM3-AML

for interactive gene ontology analysis. BMC Bioinformatics. 2018;19(1):470. 48. Schaller MA, Kallal LE, Lukacs NW. A key role for CC chemokine receptor 1 in T-cellmediated respiratory inflammation. Am J Pathol. 2008;172(2):386-394. 49. Shin SY, Lee DH, Lee J, et al. C-C motif chemokine receptor 1 (CCR1) is a target of the EGF-AKT-mTOR-STAT3 signaling axis in breast cancer cells. Oncotarget. 2017;8 (55):94591-94605. 50. Kato T, Fujita Y, Nakane K, et al. CCR1/CCL5 interaction promotes invasion of taxane-resistant PC3 prostate cancer cells by increasing secretion of MMPs 2/9 and by activating ERK and Rac signaling. Cytokine. 2013;64(1):251-257. 51. Naya A, Sagara Y, Ohwaki K, et al. Design, synthesis, and discovery of a novel CCR1 antagonist. J Med Chem. 2001;44(9):14291435. 52. Stunnenberg HG, International Human Epigenome Consortium, Hirst M. The International Human Epigenome Consortium: a blueprint for scientific collaboration and discovery. Cell. 2016;167(7): 1897. 53. Assi SA, Imperato MR, Coleman DJL, et al. Subtype-specific regulatory network rewiring in acute myeloid leukemia. Nat Genet. 2019;51(1):151-162. 54. Garcia-Cuellar MP, Buttner C, Bartenhagen C, Dugas M, Slany RK. Leukemogenic MLLENL fusions induce alternative chromatin states to drive a functionally dichotomous group of target genes. Cell Rep. 2016;15(2):

haematologica | 2022; 107(1)

310-322. 55. Guenther MG, Lawton LN, Rozovskaia T, et al. Aberrant chromatin at genes encoding stem cell regulators in human mixed-lineage leukemia. Genes Dev. 2008;22(24):34033408. 56. Erfurth FE, Popovic R, Grembecka J, et al. MLL protects CpG clusters from methylation within the Hoxa9 gene, maintaining transcript expression. Proc Natl Acad Sci U S A. 2008;105(21):7517-7522. 57. Alvarez S, Suela J, Valencia A, et al. DNA methylation profiles and their relationship with cytogenetic status in adult acute myeloid leukemia. PLoS One. 2010;5(8): e12197. 58. Zheng Y, Zhang H, Wang Y, et al. Loss of Dnmt3b accelerates MLL-AF9 leukemia progression. Leukemia. 2016;30(12):2373-2384. 59. Schulze I, Rohde C, Scheller-Wendorff M, et al. Increased DNA methylation of Dnmt3b targets impairs leukemogenesis. Blood. 2016;127(12):1575-1586. 60. Trowbridge JJ, Sinha AU, Zhu N, et al. Haploinsufficiency of Dnmt1 impairs leukemia stem cell function through derepression of bivalent chromatin domains. Genes Dev. 2012;26(4):344-349. 61. Hay SB, Ferchen K, Chetal K, Grimes HL, Salomonis N. The Human Cell Atlas bone marrow single-cell interactive web portal. Exp Hematol. 2018;68:51-61. 62. Cumbay MG, Watts VJ. Novel regulatory properties of human type 9 adenylate cyclase. J Pharmacol Exp Ther. 2004;310

(1):108-115. 63. Perez DR, Smagley Y, Garcia M, et al. Cyclic AMP efflux inhibitors as potential therapeutic agents for leukemia. Oncotarget. 2016;7(23):33960-33982. 64. Gausdal G, Wergeland A, Skavland J, et al. Cyclic AMP can promote APL progression and protect myeloid leukemia cells against anthracycline-induced apoptosis. Cell Death Dis. 2013;4(2):e516. 65. Bagger FO, Kinalis S, Rapin N. BloodSpot: a database of healthy and malignant haematopoiesis updated with purified and single cell mRNA sequencing profiles. Nucleic Acids Res. 2019;47(D1):D881-D885. 66. Gilliland CT, Salanga CL, Kawamura T, Trejo J, Handel TM. The chemokine receptor CCR1 is constitutively active, which leads to G protein-independent, betaarrestin-mediated internalization. J Biol Chem. 2013;288(45):32194-32210. 67. Chen X, Burkhardt DB, Hartman AA, et al. MLL-AF9 initiates transformation from fastproliferating myeloid progenitors. Nat Commun. 2019;10(1):5767. 68. Yan F, Powell DR, Curtis DJ, Wong NC. From reads to insight: a hitchhiker's guide to ATAC-seq data analysis. Genome Biol. 2020;21(1):22. 69. Borsari B, Villegas-Mirón P, Laayouni H, et al. Intronic enhancers regulate the expression of genes involved in tissue-specific functions and homeostasis. bioRxiv. 2020 Aug 22. doi: 10.1101/2020.08.21.260836 [preprint, not peer-reviewed]

ARTICLE Ferrata Storti Foundation

Acute Myeloid Leukemia

Fusion gene detection by RNA-sequencing complements diagnostics of acute myeloid leukemia and identifies recurring NRIP1-MIR99AHG rearrangements Paul Kerbs,1,2,3 Sebastian Vosberg,1,2,3 Stefan Krebs,4 Alexander Graf,4 Helmut Blum,4 Anja Swoboda,1 Aarif M. N. Batcha,5 Ulrich Mansmann,5 Dirk Metzler,6 Caroline A. Heckman,7 Tobias Herold1,2,3 and Philipp A. Greif1,2,3

Haematologica 2022 Volume 107(1):100-111

Department of Medicine III, University Hospital, LMU Munich, Munich, Germany; German Cancer Consortium (DKTK), partner site Munich, Germany; 3German Cancer Research Center (DKFZ), Heidelberg, Germany; 4Laboratory for Functional Genome Analysis (LAFUGA), Gene Center, LMU Munich, Munich, Germany; 5Department of Medical Data Processing, Biometry and Epidemiology, LMU Munich, Munich, Germany; 6 Division of Evolutionary Biology, Faculty of Biology, LMU Munich, Planegg-Martinsried, Germany and 7Institute for Molecular Medicine Finland (FIMM), University of Helsinki, Helsinki, Finland 1 2

ABSTRACT

Presented in part at the 61st ASH Annual Meeting Kerbs P, et al. Blood. 2019;134(Suppl_1):4655.

Correspondence: PHILIPP A. GREIF pgreif@med.lmu.de Received: January 26, 2021. Accepted: May 3, 2021. Pre-published: June 17, 2021. https://doi.org/10.3324/haematol.2021.278436

100

dentification of fusion genes in clinical routine is mostly based on cytogenetics and targeted molecular genetics, such as metaphase karyotyping, fluorescence in situ hybridization and reverse-transcriptase polymerase chain reaction. However, sequencing technologies are becoming more important in clinical routine as processing time and costs per sample decrease. To evaluate the performance of fusion gene detection by RNAsequencing compared to standard diagnostic techniques, we analyzed 806 RNA-sequencing samples from patients with acute myeloid leukemia using two state-of-the-art software tools, namely Arriba and FusionCatcher. RNA-sequencing detected 90% of fusion events that were reported by routine with high evidence, while samples in which RNA-sequencing failed to detect fusion genes had overall lower and inhomogeneous sequence coverage. Based on properties of known and unknown fusion events, we developed a workflow with integrated filtering strategies for the identification of robust fusion gene candidates by RNA-sequencing. Thereby, we detected known recurrent fusion events in 26 cases that were not reported by routine and found discrepancies in evidence for known fusion events between routine and RNA-sequencing in three cases. Moreover, we identified 157 fusion genes as novel robust candidates and comparison to entries from ChimerDB or Mitelman Database showed novel recurrence of fusion genes in 14 cases. Finally, we detected the novel recurrent fusion gene NRIP1MIR99AHG resulting from inv(21)(q11.2;q21.1) in nine patients (1.1%) and LTN1-MX1 resulting from inv(21)(q21.3;q22.3) in two patients (0.25%). We demonstrated that NRIP1-MIR99AHG results in overexpression of the 3' region of MIR99AHG and the disruption of the tricistronic miRNA cluster miR-99a/let-7c/miR-125b-2. Interestingly, upregulation of MIR99AHG and deregulation of the miRNA cluster, residing in the MIR99AHG locus, are known mechanisms of leukemogenesis in acute megakaryoblastic leukemia. Our findings demonstrate that RNA-sequencing has a strong potential to improve the systematic detection of fusion genes in clinical applications and provides a valuable tool for fusion discovery.

Introduction Fusion genes result from chromosomal aberrations, such as translocations, duplications, inversions or small interstitial deletions. On the transcript level, fusion genes may not only reflect underlying genomic rearrangements but may also arise due to

haematologica | 2022; 107(1)

RNA-sequencing of fusion genes in AML

aberrant transcription or trans-splicing events. Many fusion genes have been described as drivers across multiple human cancer entities.1 Hematopoietic malignancies are well-characterized regarding the abundance of fusion genes, e.g, chronic myeloid leukemia harbors the BCR-ABL1 fusion in more than 95% of cases, and acute promyelocytic leukemia is characterized by PML-RARA in 90% of cases. In acute myeloid leukemia (AML), fusion genes are found in about 30% of patients2 and are often regarded as major markers, defining clinically relevant subtypes.3-6 Their identification is crucial for risk assessment and deciding treatment strategy. During the initial diagnosis of AML, fusion genes are detected using conventional metaphase karyotyping (hereafter referred to as Karyotyping) and/or targeted molecular assays (hereafter referred to as molecular diagnostics) such as fluorescence in situ hybridization (FISH) or reverse-transcriptase polymerase chain reaction (hereafter referred to as PCR). On a chromosomal level, Karyotyping detects abnormalities by light microscopy of metaphase spreads, whereas FISH labels chromosomal alterations using specifically designed probes that bind to particular genomic regions of interest. On a molecular level, PCR may confirm the presence of a specific genomic or transcriptomic sequence by targeted amplification. However, these methods are laborious and time-consuming, depend on the experience of the analyst and might be subject to erroneous assessments. Furthermore, the resolution of Karyotyping is limited to the microscopic level of chromosomal arms/bands and PCR/FISH can only be used to analyze predefined targets. Small inversions, duplications or short interstitial deletions as well as cryptic fusions are difficult to detect with these procedures. Although FISH and PCR are suitable for the targeted detection of submicroscopic lesions, they are not routinely applied to the systematic identification of previously uncharacterized aberrations and usually serve as complementary validation methods. Over the last decade, next-generation sequencing techniques have evolved tremendously and are being increasingly used in clinical diagnostics.7-9 Next-generation sequencing methods enable scalable genomic analyses, ranging from single genes and gene sets of interest up to

genome-wide analyses, covering the entire genome at single base pair resolution. Furthermore, RNA-sequencing enables transcriptome-wide studies, covering all transcribed genes present in a cell. Recently, a study proposed a single bioinformatic pipeline for AML diagnostics which integrates detection of fusion genes, small variants, tandem duplications and gene expression from RNA-sequencing data.10 Thus, DNA and RNA-sequencing allow for the examination of a wide range of genetic lesions, including the discovery of novel aberrations. Sequencing technologies are improving quickly and innovation in this field continuously reduces time and costs for genomic analyses, which enables the processing of even more samples in parallel with even greater precision. Simultaneously, developments in computational biology can exploit these advancements for accurate detection of genetic aberrations. To date, more than 20 algorithms for fusion gene detection by RNA-sequencing have been published9,11,12 but identification of fusions using RNA-sequencing remains challenging and a high rate of false positives is common. Therefore, careful evaluation of fusion calls and appropriate filtering strategies are needed to enable reliable application of this technology in diagnostics. In AML, no comprehensive comparison of fusion gene detection by RNA-sequencing and clinical routine has been reported so far. In this study, we set out to assess the potential of RNA-sequencing for the detection of clinically relevant fusion genes in comparison to standard diagnostic methods. Additionally, we developed several filters for robust fusion gene identification and the discovery of novel rearrangements in AML patients.

Methods Patients’ samples A total of 806 AML patients’ samples were subjected to wholetranscriptome sequencing. The samples were collected from within four different cohorts: (i) the German AML cooperative group (AMLCG 2008 and AMLCG 1999, n=257);13,14 (ii) the German Cancer Consortium (DKTK, n=69);15,16 (iii) the Beat AML program (n=423);17 and (iv) the Institute for Molecular Medicine Finland

Table 1. Summary of the patients’ characteristics.

Median age (range)

Sex, n (%)

ELN risk group, n (%)

AMLCG (n=257)

Cohort

58 (18-79)

Females = 131 (51.0) Males = 126 (49.0)

DKTK (n=69)

61 (21-85)

Females = 31 (44.9)

Beat AML (n=423)

61 (2-87)

Females = 186 (44.0) Males = 237 (56.0)

58.5 (21-77)

Females = 29 (50.0) Males = 29 (50.0)

Favorable = 75 (29.2) Intermediate = 61 (23.7) Adverse = 107 (41.6) NA = 14 (5.5) Favorable = 33 (47.8) Males = 38 (55.1) ntermediate-I = 25 (36.2) Intermediate-II = 7 (10.1) Adverse = 3 (4.4) NA = 1 (1.5) Favorable = 112 (26.5) Intermediate = 141 (33.3) Adverse = 148 (35.0) Favorable or Intermediate = 13 (3.1) Intermediate or Adverse = 7 (1.6) NA = 2 (0.5) Favorable = 9 (15.8) Intermediate = 19 (33.3) Adverse = 18 (31.6) NA = 11 (19.3)

FIMM (n=57)

ELN: European LeukemiaNet; NA: not available.

haematologica | 2022; 107(1)

101

P. Kerbs et al.

(FIMM, n=57).18 RNA-sequencing was performed as described previously.13-18 The patients’ characteristics are summarized in Table 1 and Online Supplementary Table S1. Sequencing metrics are summarized in Table 2. In addition, RNA-sequencing data of healthy samples were obtained from the Gene Expression Omnibus Database (Online Supplementary Table S2; n=36) and the FIMM cohort (n=3). All study protocols were approved by the institutional review boards of the participating centers. All patients provided written informed consent for scientific use of surplus samples in accordance with the Declaration of Helsinki.

Definitions and metrics for evaluating performance in fusion gene detection Comprehensive definitions and metrics are provided in the Online Supplementary Methods. In brief, recurrent and reliably detected fusion genes that were reported by public databases were defined as known fusions. Furthermore, fusion genes that were found with high evidence by at least one method in routine diagnostics were defined as a benchmark (true fusions). High and low evidence were defined separately for Karyotyping, molecular diagnostics and RNA-sequencing (Online Supplementary Methods, Online Supplementary Figure S1).

Filtering strategies Initially, built-in filters of the callers were applied and fusions were filtered by a custom-generated blacklist (Online Supplementary Methods). The Promiscuity Score (PS), developed in this study, excluded fusion events whose respective partner genes were frequently called in other distinct fusion events, since these are likely artifacts. Furthermore, low read coverage of a fusion event relative to the read coverage of its partner genes indicates an artifact. Our custom Fusion Transcript Score (FTS) measures, in transcripts per million, the expression of a fusion relative to the expression of its partner genes. Fusion events with a low FTS were excluded. The Robustness Score (RS) of a fusion gene is defined as the ratio between the number of samples in which this fusion gene passed all applied filters and the total number of samples in which this fusion gene was called. Only fusion genes passing all filters in at least half of the reported samples were considered. A comprehensive description of the filtering is enclosed in the Online Supplementary Methods.

Results Close correlation of fusion detection by routine diagnostics and RNA-sequencing In 806 patients’ samples, we identified 138 true fusions which provided the benchmark for the comparison of performance in fusion gene detection between routine diagnostics (Karyotyping, molecular diagnostics) and RNA-

sequencing (Figure 1, Online Supplementary Table S3). Of 138 true fusions, Karyotyping identified 121 (87.7%) and molecular diagnostics identified 107 (77.5%) with high evidence. In addition, Karyotyping identified 11 (8%) and molecular diagnostics identified five (3.6%) true fusions with low evidence. Fusion gene detection by RNAsequencing resulted in 124 (89.9%) positive findings (high evidence: 115, low evidence: 9), thereby missing 14 true fusions (AMLCG: n=10; Beat AML: n=4). Notably, samples from the AMLCG cohort showed less overall coverage and mappability of sequencing reads as compared to other samples (Table 2). In particular, CBFB and KMT2A showed poor coverage (Online Supplementary Figure S2), both involved in eight of ten undetected true fusions by RNA-sequencing in those samples. Further fusions missed by RNA-sequencing were DEK-NUP214 and GTF2I-RARA. Overall, in samples from the AMLCG cohort, substantially fewer fusion events were detected by FusionCatcher while Arriba detected twice as many compared to samples from other cohorts (Table 2). In the Beat AML cohort, we observed discrepancies in reported fusions between RNA-sequencing and clinical routine in three of four cases of true fusions missed by RNAsequencing: (i) Karyotyping reported t(6;11)(q27;q23) resulting in KMT2A-AFDN, while RNA-sequencing detected KMT2A-MLLT10 resulting from t(10;11)(p12;q23); (ii) Karyotyping reported del(2)(p21p23) resulting in EML4ALK, while RNA-sequencing detected KMT2A-MLLT3 resulting from t(9;11)(p21;q23); (iii) Karyotyping reported der(17)t(15;17)(q24;q21) and inv(17)(q21q21) resulting in PML-RARA and STAT5B-RARA, respectively, while RNAsequencing detected PML-RARA but not STAT5B-RARA. In the fourth case, a PML-RARA fusion was found by FISH while Karyotyping indicated a normal karyotype in this sample.

RNA-sequencing identifies known fusions missed by routine and yields additional candidates Before filtering, a total of 25,817 and 56,594 fusion events were detected in 806 samples by Arriba and FusionCatcher, respectively (mean 32 and 70 per sample, respectively) (Table 2). We applied filtering strategies as shown in Figure 2A. PS filter cutoffs for individual cohorts were set at 8.25 for AMLCG, 3.5 for DKTK, 16.5 for Beat AML and 3.5 for FIMM (Online Supplementary Figure S3A, Online Supplementary Methods). In addition to our previously described cutoffs for FTS and FTS (Online Supplementary Methods), we set a minimum FTS for unknown fusion events based on the median FTS of all detected unknown fusions (FTS ≥0.1) (Online Supplementary Figure S3B). Finally, we defined highly reliable fusion gene candidates based on 5’

3’

Table 2. Statistics for RNA-sequencing, mapping and fusion calling. RNA selection Avg. library size in millions (range) Avg. % uniquely mapped reads (range) Avg. % reads mapped to exons (range) Avg. insert size (range) Avg. fusion events called by Arriba Avg. fusion events called by FusionCatcher

AMLCG

DKTK

Beat AML

FIMM

poly(A)

30.6 (19.1-97.8) 80 (44.2-94.1) 72.4 (40.5-87.6) 248.1 (97-455) 48.3 12.9

33.7 (23.4-43.3) 90.7 (82-93.7) 81.5 (75.6-85.7) 257.1 (217-289) 23.2 113.1

55.1 (24.7-126.8) 91.4 (80.9-94.3) 76.8 (60.1-86.8) 186.7 (131-246) 24.1 97.8

Total RNA (rRNA depleted) 57.4 (23.9-119.9) 86.3 (70.4-93.3) 51 (20.2-67.9) 219.5 (141-289) 27.8 71.4

Avg: average.

102

haematologica | 2022; 107(1)

RNA-sequencing of fusion genes in AML

the overlap of filtered fusion calls from Arriba and FusionCatcher. The built-in filter of Arriba excluded, on average, more putative false fusion events (74.8%) as compared to the built-in filter of FusionCatcher (62.3%). By applying our additional filtering strategies, we further reduced the amount of putative false fusion events substantially, resulting in an average of around 94% excluded fusion events from Arriba calls and around 96% from

FusionCatcher calls (Figure 2B). Besides detected true fusions (n=115), we also found 187 fusion events as robust candidates. Thirty of these 187 events have been described before, while 157 were putative novel fusion events (Online Supplementary Table S4). Clinical routine showed only low evidence for four of the 30 known events (Figure 1B, Online Supplementary Table S5), while 26 candidates were not reported by routine diagnostics in our cohorts. In two of the

Figure 1. Evidence for fusions by routine clinical diagnostics and RNA-sequencing. (A) True fusions detected by Karyotyping, molecular diagnostics (MDx) and RNAsequencing (RNA-seq) in the AMLCG, DKTK, Beat AML and FIMM cohorts. Dark green boxes indicate high evidence, light green boxes indicate low evidence. Gray boxes represent no evidence although the respective method was performed. White boxes indicate that the respective method was not performed, or information was missing. (B) Known fusions detected with high evidence by RNA-seq that were missed or detected with low evidence only by Karyotyping/MDx. (C) Venn diagram summarizing fusions detected with the different methods.

haematologica | 2022; 107(1)

103

P. Kerbs et al. A Figure 2. Detection workflow and filtering of fusion events. (A) Detection workflow and number of filtered fusion events by filtering strategies. (B) Ratios of fusion events excluded by Arriba and FusionCatcher in each filter step and cohort. AML: acute myeloid leukemia; FTS: Fusion Transcript Score; PS: Promiscuity Score; RS: Robustness Score.

104

haematologica | 2022; 107(1)

RNA-sequencing of fusion genes in AML

four events described by clinical routine, a rearrangement of KMT2A was detected using FISH without any evidence from analyses by Karyotyping and in the other two events, Karyotyping reported rearranged chromosomes matching the chromosomal location of the fusion partner genes but different chromosomal bands. The 30 known fusion events having no or only low evidence by routine diagnostics

include recurrent fusions NUP98-NSD1 (n=8), KMT2AMLLT10 (n=4), DEK-NUP214 (n=3), KAT6A-CREBBP (n=2), KMT2A-MLLT3 (n=2) and RUNX1-CBFA2T3 (n=2). Based on the newly identified fusion genes, patients would be assigned to a different European LeukemiaNet risk group in six of the 30 cases (Online Supplementary Table S5). Chromosomal locations of detected true, known and puta-

A Figure 3. Genomic origin of fusion events detected by RNA-sequencing. Circos plots of (A) known and (B) unknown fusion gene candidates found in the AMLCG, DKTK, Beat AML and FIMM cohorts, illustrating chromosomal origin of the fusion events. Lines connect the positions of fusion partners. Thickness of lines indicates recurrence. Recurrent fusions are labeled with gene symbols of the partner genes. Blue lines indicate known fusion events, red lines indicate recurrent novel and gray lines show non-recurrent novel fusion events.

haematologica | 2022; 107(1)

105

P. Kerbs et al.

Figure 4. Detection and validation of the novel NRIP1-MIR99AHG fusion gene. Evidence for the NRIP1-MIR99AHG fusion gene in sample AM-0028-DX determined by various methods. (A) Schematic representation of the fusion transcript as predicted by RNA-sequencing. (B) Gel-electrophoresis of reverse transcriptase polymerase chain reaction analysis of fusion breakpoint and NRIP1 exon 4. Three samples from cytogenetically normal patients with acute myeloid leukemia were used as negative controls. (C) A trace from Sanger sequencing of the fusion breakpoint. (D) Mapping of long reads from Nanopore sequencing of genomic DNA. Each line represents one read, which can be divided at the breakpoints of the fusion. Single parts of the read can be mapped to the positive strand (blue) at one locus with the other part mapped to the negative strand (red) at the other locus. The consensus inversed region is indicated by orange. The mapping structure of a highlighted read at the bottom shows that one part of the read was inversely mapped to the NRIP1 locus, while the other part was mapped to the MIR99AHG locus.

106

haematologica | 2022; 107(1)

RNA-sequencing of fusion genes in AML

NRIP1-MIR99AHG is a novel recurrent fusion gene resulting from inv(21)(q11.2;q21.1)

tive novel fusion events are presented as Circos plots19 in Figure 3. Based on sample availability, we validated known fusion genes by PCR analysis that were exclusively found by RNA-sequencing in one sample (AM-0292-DX: DEKNUP214) (Online Supplementary Figure S4). Moreover, 14 out of the 157 putative novel fusion events had an entry in ChimerDB or Mitelman Database but were not classified as known based on the criteria in the present study.

Beyond the detection of known rearrangements, we sought to identify novel recurrent fusion genes. Among our 157 putative novel fusion genes, we found NRIP1MIR99AHG (Figure 4A) resulting from inv(21)(q11.2;q21.1) in six and LTN1-MX1 (Online Supplementary Figure S5) resulting from inv(21)(q21.3;q22.3) in two patients’ sam-

Figure 5. Gene expression of genes involved in fusions. Gene expression of the 5' and 3' partner genes of the respective fusion. Red dots indicate samples positive for the respective fusion, gray dots represent samples negative for the respective fusion. TPM: transcripts per million.

haematologica | 2022; 107(1)

107

P. Kerbs et al.

ples. Notably, LTN1-MX1 was only found in co-occurrence with NRIP1-MIR99AHG. Further recurrence of NRIP1MIR99AHG was reported by FusionCatcher alone in two patients’ samples (AM-0013-DX, FI-1216-RE). Based on cDNA availability, we validated the junction of the NRIP1-MIR99AHG fusion transcript by PCR in sample AM-0028-DX. Three cytogenetically normal samples (AM0044-DX, AM-0054-DX, AM-0069-DX) were used as negative controls (Figure 4B). Sanger sequencing of the PCR product confirmed a junction spanning sequence which matched the prediction of the RNA-sequencing fusion callers (Figure 4C). Nanopore sequencing of available gDNA from NRIP1-MIR99AHG-positive samples AM0028-DX (Figure 4D) and AM-0013-DX (Online Supplementary Figure S6) identified the breakpoints (Online Supplementary Table S6) and confirmed an inversion on the genomic level. With the aim of determining the complete fusion transcript, we generated a customized reference sequence of the inversion based on the identified breakpoints. Reads from Nanopore cDNA sequencing (median length: 883 bp) of the two NRIP1-MIR99AHG positive samples were mapped to this reference. Only unique mappings were considered to obtain reads spanning the junction of the fusion. We observed high coverage of the custom reference by junction-spanning reads in the two fusion-positive patients (Online Supplementary Figure S7), while there was no coverage in negative controls. NRIP1 includes a consensus coding sequence with an open reading frame starting in exon 4, whereas MIR99AHG is non-coding. The identified breakpoint in the NRIP1 locus in AM-0028-DX was located between exons 3 and 4, while the breakpoint in AM-0013DX was located between exons 1 and 2, consistent with reports from RNA-sequencing fusion callers. In both cases, no annotated open reading frame was included in the putative fusion transcripts. A validation in samples from the Beat AML cohort was not possible because of lack of access to the patients' material. Literature research yielded the report20 of a chronic myelomonocytic leukemia (CMML) patient with trisomy 21. The authors identified an inversion of chromosome 21 with breakpoints in the NRIP1 locus and in a region upstream of MIR125B2 (overlapping with an intronic region of MIR99AHG). We analyzed RNAsequencing data from this patient (FI-0564-RE) with our fusion detection workflow and found high evidence for a NRIP1-MIR99AHG fusion. In total, NRIP1-MIR99AHG was found in nine (1.1%) of 806 AML patients (AMLCG, n=2; Beat AML, n=5; FIMM, n=1) and one CMML patient.

Increased expression of the 3’ partner gene in NRIP1-MIR99AHG and other fusions In addition to the detection of fusion transcripts, we examined the expression rate of the single partner genes of a fusion and compared it between samples with and without this specific fusion. Sequence coverage of a gene as obtained from mapping but not read coverage of the fusion junction was considered as expression of this gene. Samples harboring a fusion, whose 3' partner gene is usually not expressed or expressed at low levels only, showed increased expression of the 3' partner gene up to the levels of the 5' partner gene, which is expressed at reasonable levels regardless of the fusion (Figure 5A, B). We did not observe an increase in the expression of the 3’ partner in fusion events with similar expression rates between the 5’ and 3’ partner genes (Figure 5C, D). Accordingly, MIR99AHG, which is usually not expressed or expressed at 108

low levels only, showed increased expression levels in NRIP1-MIR99AHG positive samples (Figure 5E). On the other hand, MX1, which is inherently fairly expressed, only showed a slight elevation of expression levels in LTN1MX1-positive samples (Figure 5F).

Clinical and genetic characteristics of patients with NRIP1-MIR99AHG fusion All patients found to harbor NRIP1-MIR99AHG had poor survival with a median of 296 days (range, 36-1650 days). Interestingly, most of the patients were male (6/9) and had a median age of 59 years (Online Supplementary Table S7). Karyotyping showed a complex karyotype in four patients and five patients were refractory to intensive induction therapy. Furthermore, three patients showed a gain, and one patient showed a loss of chromosome 21. Unfortunately, we have no information about whether these patients had a constitutional or somatic monosomy/trisomy 21. Cytomorphology was available for three of the nine patients without there being any evidence of megakaryoblastic leukemia (French-American-British classification, M7). Mutational status was available for six of the nine patients, but no apparent pattern was observed. However, recurrently mutated genes among those patients were NRAS (n=2) and ASXL1 (n=2) (Online Supplementary Table S7).

Discussion The aim of this study was to test the potential of fusion gene detection by RNA-sequencing in several cohorts of AML patients’ samples and to assess its diagnostic applicability by comparison to current standard techniques used in clinical routine. Based on our benchmark, the vast majority of true fusions reported by routine diagnostics was also detected by RNA-sequencing, underscoring the high sensitivity of this method. Notably, most of the samples in which a true fusion could not be detected by RNA-sequencing had a low read depth (median = 24 million mapped reads), while a minimum of 30 million mapped reads is recommended by the ENCODE consortium21 for general expression analyses and even deeper sequencing for transcript discovery (e.g., fusion transcripts). Therefore, fusion gene detection was most likely hampered by the low read depth of these samples. Limitations of fusion gene detection by RNA-sequencing are governed by library preparation steps, read depth, expression rates of the affected genes and the applied bioinformatic algorithms. On the other hand, Karyotyping is limited to a resolution of 5-10 Mb,22 which hampers the identification of small or cryptic rearrangements as well as rearrangements in specific locations (e.g., centromeric, telomeric).23 Furthermore, break-apart FISH probes identify genomic rearrangements in targeted regions through the visual separation of fluorescent labels. Although this can indicate the rearrangement of a targeted locus, the detection of a specific aberration is still limited by the resolution of microscopic inspection, and the identification of the involved partner locus requires additional assays. In contrast to break-apart FISH, dual fusion probes target two partner loci and thereby can detect specific rearrangements but are restricted to the candidate loci of interest. In analogy, targeted PCR amplification of fusion transcripts requires prior knowledge of the affected genes and the corresponhaematologica | 2022; 107(1)

RNA-sequencing of fusion genes in AML

ding break-point regions. Diagnostic application of RNAsequencing has the potential to overcome these limitations through systematic detection of fusion genes on a transcriptome-wide level, as demonstrated in these three examples: (i) NUP98-NSD1 is a biomarker for poor prognosis and NUP98 fusions in general were found to define a clinically relevant distinct subgroup in AML24-26 but reliable detection of the underlying cryptic translocation t(5;11)(q35.2;p15.4) by Karyotyping is not possible.27 Of note, we identified NUP98-NSD1 in eight samples using RNA-sequencing, as well as further known fusion genes in 22 samples that showed no or only low evidence for these fusions by either Karyotyping or molecular diagnostics. (ii) We observed discrepancies between results from routine and RNA-sequencing, i.e., one sample showing a translocation t(6;11)(q27;q23), according to Karyotyping. This translocation results in a KMT2A-AFDN fusion but RNA-sequencing reported a KMT2A-MLLT10 fusion with high evidence, corresponding to translocation t(10;11)(p12;q23). Furthermore, KMT2A rearrangements were reported by break-apart FISH without any evidence for a rearrangement by Karyotyping in two cases. Fusion detection by RNA-sequencing identified a KMT2A-MLLT10 fusion in these samples. Since various KMT2A fusions may reflect different risk assessments based on the European LeukemiaNet classification,6 the correct description of the fusion may have therapeutic consequences. (iii) In another sample, Karyotyping and FISH identified a t(15;17)(q24;q21) translocation, typically resulting in a PML-RARA fusion transcript (no information on PCR status was available), while RNA-sequencing identified a PML-CASC3 fusion, with CASC3 being located ~170 kb upstream of RARA. Unfortunately, no information was available on this patient’s response to all-trans retinoic acid treatment. In addition to standard diagnostic methods that are used in clinical routine, targeted RNA-sequencing panels are becoming increasingly popular for high-throughput detection of annotated fusion genes and were shown to be more sensitive than classical approaches.28 Admittedly, RNA-sequencing-based fusion callers report many false positive events due to technical and biological properties, such as sequencing errors, false mapping, homologous genomic regions, polymorphic genes, or exceptionally high gene expression.29 Some genes are therefore prone to be reported in fusion gene artifacts, requiring reasonable filtering to maintain sensitivity while increasing specificity of the fusion detection analysis. Current callers integrate blacklists of fusion events into their built-in filters, which are compiled from public databases. However, technical differences between sequencing protocols and fusion calling algorithms may result in specific fusion artifacts that are not covered by those blacklists. Therefore, the generation of an additional customized blacklist further improves the specificity in RNA-sequencing-based fusion analyses. Furthermore, we found genes which form fusions with many distinct partners indicating that these events are likely artifacts. The PS, developed in the present study, evaluates fusion events using this characteristic and filters events based on scores obtained from known fusions. However, the PS depends on the sequencing properties and the number of samples from which the score was derived. Thus, we defined cutoffs for the individual cohorts separately. Furthermore, the amount of fusion supporting reads correlates with the number of reads supporting the expression of the individual partner haematologica | 2022; 107(1)

genes. The FTS, also developed in this study, measures the abundance of fusion transcripts relative to their respective partner gene transcripts. Most known fusions had an FTS around 0.3, but fusions present in subclones only, or fusions found in samples with lower tumor load will yield lower scores. As a tradeoff between specificity and sensitivity, we defined the median of all FTS detected in unknown fusion events as a cutoff. Besides, we observed unknown fusion events with high recurrence that passed all preceding filter steps in some samples, while these fusion events were filtered out in most other samples. This may indicate transcript artifacts of error-prone genes. The RS filter therefore excludes fusion events that failed at least one preceding filter in most of the identified cases. The integration of our PS, FTS, customized blacklist and RS Filter into our detection strategy substantially reduced the fusion calls that were most likely false or irrelevant. Selection of fusion events consistently found between Arriba and FusionCatcher further increased the evidence of fusion candidates. As an additional source of evidence for fusion events, we utilized individual gene expression values of the partner genes. The expression of a fusion transcript is mostly driven by the promoter of the 5' partner gene and the expression of the 3' partner should therefore adjust to the levels of the 5' partner. Although this simplified assumption neglects the influence of 3' enhancers and other regulatory elements, we observed substantially elevated expression of the 3' partner if it is usually not expressed or expressed at low levels only. Consistently, 3' partner genes with inherently similar expression as the 5' partner showed no or only marginal adjustments in expression levels. However, genomic rearrangements do not necessarily result in a fusion transcript but may have other effects, e.g., the reallocation of the 3' enhancer of GATA2 in inv(3)(q21.3q26.2)/t(3;3)(q21.3;q26.2)-positive leukemia, causing overexpression of MECOM and GATA2 haploinsufficiency.30,31 Although, there is usually no fusion transcript in these patients, we found evidence for the transposition of MECOM by chimeric reads found in several affected samples (data not shown). Among our fusion candidates, we identified the novel recurrent fusion gene NRIP1-MIR99AHG, which results from inversion inv(21)(q11.2;q21.1). Interestingly, both Nanopore sequencing and RNA-sequencing revealed different breakpoint positions in NRIP1-MIR99AHG-positive samples. None of the identified fusion transcripts included an annotated consensus coding sequence, and therefore translation to a protein product is rather unlikely. NRIP1 was described as a transcriptional repressor,32 playing a role in hematologic malignancies,33,34 and was found to be involved in other fusions.35 A disruption of the corresponding gene by the NRIP1-MIR99AHG rearrangement might therefore contribute to leukemogenesis. On the other hand, overexpression of MIR99AHG and accompanying enhanced proliferation were previously demonstrated in acute megakaryoblastic leukemia cell lines (with MIR99AHG referred to as MONC).36 Furthermore, MIR99AHG is the host gene of miR-99a/let-7c/miR-125b-2, a microRNA cluster, also shown to influence homeostasis of hematopoietic stem and progenitor cells.37 Interestingly, the identified fusion breakpoint in the MIR99AHG locus was located between let-7c and miR-125b-2, thereby disrupting the tricistronic gene cluster. This aberration as well as fusion-induced transcription of the 3’ region of MIR99AHG may constitute a mechanism of leukemogen109

P. Kerbs et al.

esis. In the present study, NRIP1-MIR99AHG was found in eight AML patients, as well as in one CMML patient, all of whom had poor survival and were mostly refractory to intensive induction treatment. However, this might also be related to the complex karyotype in several patients. Of note, a recent whole-transcriptome study of 572 patients with AML and 630 with myelodysplastic syndromes did not find any NRIP1-MIR99AHG fusions.38 An extended analysis by the same authors39 of overlapping cohorts, presented at the recent Annual Meeting of the American Society of Hematology, identified recurring NRIP1MIR99AHG in AML and myelodysplastic syndromes but not in lymphoid malignancies (with MIR99AHG referred to as LINC00478). Further studies are needed to gain more insight into the pathogenic, diagnostic and prognostic significance of the NRIP1-MIR99AHG fusion in AML and other hematologic malignancies. In conclusion, RNA-sequencing allows for accurate and more exhaustive identification of fusion transcripts as compared to classical cytogenetics or molecular diagnostics alone. We demonstrated that crucial AML-related fusions can be reliably identified by RNA-sequencing, but low sequence coverage limited sensitivity in a subset of samples. These findings underscore the need for stringent quality metrics in diagnostic RNA-sequencing applications. Nevertheless, we found several AML-related fusions that are difficult to detect by clinical routine. Furthermore, our workflow allowed for the identification of novel recurrent fusion transcripts such as NRIP1-MIR99AHG, which results from the chromosomal rearrangement inv(21)(q11.2;q21.1).

References 1. Gao Q, Liang W-W, Foltz SM, et al. Driver fusions and their implications in the development and treatment of human cancers. Cell Rep. 2018;23(1):227-238. 2. Grimwade D, Hills RK, Moorman AV, et al. Refinement of cytogenetic classification in acute myeloid leukemia: determination of prognostic significance of rare recurring chromosomal abnormalities among 5876 younger adult patients treated in the United Kingdom Medical Research Council trials. Blood. 2010;116(3):354-365. 3. Döhner H, Estey EH, Amadori S, et al. Diagnosis and management of acute myeloid leukemia in adults: recommendations from an international expert panel, on behalf of the European LeukemiaNet. Blood. 2010;115(3):453-474. 4. Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood 2016;127(20):2391-405. 5. Wang Y, Wu N, Liu D, Jin Y. Recurrent fusion genes in leukemia: an attractive target for diagnosis and treatment. Curr Genomics. 2017;18(5):378-384. 6. Döhner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017;129(4):424-447. 7. Mack E, Langer D, Marquardt A, et al. Comprehensive genetic diagnostics of acute myeloid leukemia by next generation sequencing. Blood. 2016;128(22):1665. 8. Bacher U, Shumilov E, Flach J, et al. Challenges in the introduction of next-gener-

110

This study presents RNA-sequencing as a valuable complementary method to current standard techniques for the detection of fusion genes and we recommend the integration of RNA-sequencing applications into clinical routine for more comprehensive and precise diagnostics of hematologic malignancies. Disclosures No conflicts of interest to disclose. Contributions PK, SV, SK and AS performed research and analyzed data. AMNB and AG provided computational support. CH, PAG and TH provided RNA-sequencing data and clinical annotations. SV, HB, UM, DM and PAG supervised the research. PK, SV and PAG wrote the manuscript. All authors approved the final manuscript. Acknowledgments We thank all participants and recruiting centers of the AMLCG and Beat AML trials. We also thank Bianka Ksienzyk and William Keay for technical support. Funding This study was supported by the Deutsche Forschungsgemeinschaft (DFG) within the Collaborative Research Center (SFB) 1243 "Cancer Evolution" (projects A08 and A16). PAG acknowledges support from the Munich Clinician Scientist Program (MCSP). SV was supported by the Deutsche José Carreras Leukämie-Stiftung.

ation sequencing (NGS) for diagnostics of myeloid malignancies into clinical routine use. Blood Cancer J. 2018;8(11):113. 9. Liu S, Tsai W-H, Ding Y, et al. Comprehensive evaluation of fusion transcript detection algorithms and a meta-caller to combine top performing methods in paired-end RNA-seq data. Nucleic Acids Res. 2016;44(5):e47. 10. Arindrarto W, Borràs DM, de Groen RAL, et al. Comprehensive diagnostics of acute myeloid leukemia by whole transcriptome RNA sequencing. Leukemia. 2020;35(1):4761. 11. Kumar S, Vo AD, Qin F, Li H. Comparative assessment of methods for the fusion transcripts detection from RNA-seq data. Sci Rep. 2016;6:21597. 12. Haas BJ, Dobin A, Li B, Stransky N, Pochet N, Regev A. Accuracy assessment of fusion transcript detection via read-mapping and de novo fusion transcript assembly-based methods. Genome Biol. 2019;20(1):213. 13. Braess J, Amler S, Kreuzer KA, et al. Sequential high-dose cytarabine and mitoxantrone (S-HAM) versus standard double induction in acute myeloid leukemia—a phase 3 study. Leukemia. 2018;32(12):25582571. 14. Büchner T, Berdel WE, Schoch C, et al. Double induction containing either two courses or one course of high-dose cytarabine plus mitoxantrone and postremission therapy by either autologous stem-cell transplantation or by prolonged maintenance for acute myeloid leukemia. J Clin Oncol. 2006;24(16):2480-2489. 15. Hartmann L, Dutta S, Opatz S, et al. ZBTB7A mutations in acute myeloid leukaemia with t(8;21) translocation. Nat Commun.

2016;7(1):1-7. 16. Greif PA, Hartmann L, Vosberg S, et al. Evolution of cytogenetically normal acute myeloid leukemia during therapy and relapse: An exome sequencing study of 50 patients. Clin Cancer Res. 2018;24(7):17161726. 17. Tyner JW, Tognon CE, Bottomly D, et al. Functional genomic landscape of acute myeloid leukaemia. Nature. 2018;562(7728): 526-531. 18. Pemovska T, Kontro M, Yadav B, et al. Individualized systems medicine strategy to tailor treatments for patients with chemorefractory acute myeloid leukemia. Cancer Discov. 2013;3(12):1416-1429. 19. Krzywinski M, Schein J, Birol I, et al. Circos: An information aesthetic for comparative genomics. Genome Res. 2009;19(9):16391645. 20. Majumder MM, Kontro M, Edgren H, et al. Genomic and transcriptomic data integration in chronic myelomonocytic leukemia reveals a novel fusion gene involving onco-miR125b-2. Cancer Res. 2012;72(8 Suppl):3175. 21. Davis CA, Hitz BC, Sloan CA, et al. The Encyclopedia of DNA Elements (ENCODE): data portal update. Nucleic Acids Res. 2018;46(D1):D794-D801. 22. Gelehrter TD, Collins FS, Ginsburg D. Principles of Medical Genetics. Williams & Wilkins. pp 153-194. 23. De Braekeleer E, Meyer C, Douet-Guilbert N, et al. Complex and cryptic chromosomal rearrangements involving the MLL gene in acute leukemia: a study of 7 patients and review of the literature. Blood Cells Mol Dis. 2010;44(4):268-274. 24. Kivioja JL, Lopez Martí JM, Kumar A, et al. Chimeric NUP98-NSD1 transcripts from the

haematologica | 2022; 107(1)

RNA-sequencing of fusion genes in AML

cryptic t(5;11)(q35.2;p15.4) in adult de novo acute myeloid leukemia. Leuk Lymphoma. 2018;59(3):725-732. 25. Hollink IHIM, van den Heuvel-Eibrink MM, Arentsen-Peters STCJM, et al. NUP98/NSD1 characterizes a novel poor prognostic group in acute myeloid leukemia with a distinct HOX gene expression pattern. Blood. 2011;118(13):3645-3656. 26. Bisio V, Zampini M, Tregnago C, et al. NUP98-fusion transcripts characterize different biological entities within acute myeloid leukemia: a report from the AIEOPAML group. Leukemia. 2017;31(4):974-977. 27. Kearney L. t(5;11)(q35;p15.5) NUP98/NSD1. Atlas Genet Cytogenet Oncol Haematol. 2002;6(3):209-211. http://atlasgeneticsoncology.org/Anomalies/t0511q35p15ID1209.ht ml (2002, accessed April 28, 2020). 28. Heyer EE, Deveson IW, Wooi D, et al. Diagnosis of fusion genes using targeted RNA sequencing. Nat Commun. 2020;11(1): 1810. 29. Conesa A, Madrigal P, Tarazona S, et al. A survey of best practices for RNA-seq data

haematologica | 2022; 107(1)

analysis. Genome Biol. 2016;17(1):13. 30. Gröschel S, Sanders MA, Hoogenboezem R, et al. A single oncogenic enhancer rearrangement causes concomitant EVI1 and GATA2 deregulation in Leukemia. Cell. 2014;157(2): 369-381. 31. Yamazaki H, Suzuki M, Otsuki A, et al. A remote GATA2 hematopoietic enhancer drives leukemogenesis in inv(3)(q21;q26) by activating EVI1 expression. Cancer Cell. 2014;25(4):415-427. 32. Castet A, Boulahtouf A, Versini G, et al. Multiple domains of the ReceptorInteracting Protein 140 contribute to transcription inhibition. Nucleic Acids Res. 2004;32(6):1957-1966. 33. Lapierre M, Castet-Nicolas A, Gitenay D, et al. Expression and role of RIP140/NRIP1 in chronic lymphocytic leukemia. J Hematol Oncol. 2015;8:20. 34. Herold T, Jurinovic V, Metzeler KH, et al. An eight-gene expression signature for the prediction of survival and time to treatment in chronic lymphocytic leukemia. Leukemia. 2011;25(10):1639-1645.

35. Zhang R, Kim YM, Yang X, Li Y, Li S, Lee JY. A possible 5’-NRIP1/UHRF1-3’ fusion gene detected by array CGH analysis in a Ph+ ALL patient. Cancer Genet. 2011;204(12): 687-691. 36. Emmrich S, Streltsov A, Schmidt F, Thangapandi VR, Reinhardt D, Klusmann JH. LincRNAs MONC and MIR100HG act as oncogenes in acute megakaryoblastic leukemia. Mol Cancer. 2014;13(1):171. 37. Emmrich S, Rasche M, Schöning J, et al. miR-99a/100~125b tricistrons regulate hematopoietic stem and progenitor cell homeostasis by shifting the balance between TGFb and Wnt signaling. Genes Dev. 2014;28(8):858-874. 38. Stengel A, Shahswar R, Haferlach T, et al. Whole transcriptome sequencing detects a large number of novel fusion transcripts in patients with AML and MDS. Blood Adv. 2020;4(21):5393-5401. 39. Haferlach C, Walter W, Meggendorfer M, et al. The diverse landscape of fusion transcripts in 25 different hematological entities. Blood. 2020;136(Suppl 1):16-17.

111

ARTICLE Ferrata Storti Foundation

Blood Transfusion

Beta thalassemia minor is a beneficial determinant of red blood cell storage lesion Vassilis L. Tzounakas,1* Alkmini T. Anastasiadi,1* Davide Stefanoni,2 Francesca Cendali,2 Lorenzo Bertolone,2 Fabia Gamboni,2 Monika Dzieciatkowska,2 Pantelis Rousakis,3 Athina Vergaki,4 Vassilis Soulakis,4 Ourania E. Tsitsilonis,3 Konstantinos Stamoulis,5 Issidora S. Papassideri,1 Anastasios G. Kriebardis,6 Angelo D’Alessandro2 and Marianna H. Antonelou1 Department of Biology, Section of Cell Biology and Biophysics, School of Science, National and Kapodistrian University of Athens (NKUA), Athens, Greece; 2Department of Biochemistry and Molecular Genetics, University of Colorado, School of Medicine– Anschutz Medical Campus, Aurora, CO, USA; 3Department of Biology, Section of Animal and Human Physiology, School of Science (NKUA), Athens, Greece; 4Regional Blood Transfusion Center, “Agios Panteleimon” General Hospital of Nikea, Piraeus, Greece; 5 Hellenic National Blood Transfusion Center, Acharnes, Athens, Greece and 6Department of Biomedical Science, School of Health & Caring Science, University of West Attica (UniWA), Egaleo, Greece. 1

Haematologica 2022 Volume 107(1):112-125

*VLT and ATA contributed equally as co-first authors.

ABSTRACT

Correspondence: MARIANNA H. ANTONELOU manton@biol.uoa.gr ANGELO D’ALESSANDRO angelo.dalessandro@cuanschutz.edu Received: October 6, 2020. Accepted: December 4, 2020. Pre-published: March 18, 2021. https://doi.org/10.3324/haematol.2020.273946

112

lood donor genetics and lifestyle affect the quality of red blood cell (RBC) storage. Heterozygotes for beta thalassemia (bThal+) constitute a non-negligible proportion of blood donors in the Mediterranean and other geographical areas. The unique hematological profile of bThal+ could affect the capacity of enduring storage stress, however, the storability of bThal+ RBC is largely unknown. In this study, RBC from 18 bThal+ donors were stored in the cold and profiled for primary (hemolysis) and secondary (phosphatidylserine exposure, potassium leakage, oxidative stress) quality measures, and metabolomics, versus sex- and age-matched controls. The bThal+ units exhibited better levels of storage hemolysis and susceptibility to lysis following osmotic, oxidative and mechanical insults. Moreover, bThal+ RBC had a lower percentage of surface removal signaling, reactive oxygen species and oxidative defects to membrane components at late stages of storage. Lower potassium accumulation and higher uratedependent antioxidant capacity were noted in the bThal+ supernatant. Full metabolomics analyses revealed alterations in purine and arginine pathways at baseline, along with activation of the pentose phosphate pathway and glycolysis upstream to pyruvate kinase in bThal+ RBC. Upon storage, substantial changes were observed in arginine, purine and vitamin B6 metabolism, as well as in the hexosamine pathway. A high degree of glutamate generation in bThal+ RBC was accompanied by low levels of purine oxidation products (IMP, hypoxanthine, allantoin). The bThal mutations impact the metabolism and the susceptibility to hemolysis of stored RBC, suggesting good post-transfusion recovery. However, hemoglobin increment and other clinical outcomes of bThal+ RBC transfusion deserve elucidation by future studies.

Introduction Inter-donor heterogeneity significantly impacts the two “gold standards” of red blood cell (RBC) storage quality, namely end of storage hemolysis and in vivo 24hour post-transfusion recovery.1 Donor age, sex, ethnicity2 and lifestyle (smoking, drinking, caffeine consumption3) all impact stored RBC energy and redox metabolism, and thereby RBC capacity to cope with oxidant and other insults. These factors ultimately affect transfusion efficacy, as gleaned by outcomes like hemoglobin (Hb) increments upon transfusion.4 Genetic factors impacting RBC redox status and antioxidant capacity have been linked to alteration of the metabolic age

haematologica | 2022; 107(1)

The favorable storability of bThal+ RBC

of stored RBC,5 such as deficiency in the activity of glucose 6-phosphate dehydrogenase (G6PD) resulting both in increased oxidant stress6 and lower recovery in healthy blood donors.7 Like G6PD deficiency, genetic polymorphisms associated with non-canonical hemoglobin traits are enriched in certain donor populations. Carrier state for beta thalassemia (bThal+) is characterized by mild effects on globin synthesis and RBC survival, and as such, many bThal+ subjects are eligible blood donors.8 Despite their unique hematological profile, that could affect the capacity of enduring storage stress in either a negative or a positive way, little is known about the storability and recovery of bThal+ RBC. Apart from RBC indices and minor Hb variants,9 bThal+ RBC may exhibit differences in membrane structure,10 deformability and ion exchange,11 among others. Free heme and iron reactions bring about mild but sustained oxidative stress that when combined with decreased plasma antioxidant capacity,12 may lead to oxidative defects in skeletal proteins13 and membrane lipids. Augmented protein phosphorylation14 and proteolytic cleavage of band 315 have been also observed as a probable result of caspase-3 activation. Since the N-terminus of band 3 can regulate metabolic fluxes through glycolysis by means of inhibitory binding to glycolytic enzymes, the aforementioned alterations could trigger excessive consumption of glucose through the glycolytic pathway at the expense of NADPH production by the pentose phosphate pathway (PPP),11 with a consequent deficit in the capacity to fuel several antioxidant systems that rely on this cofactor. Many of these bThal+ RBC distortions are typical storage lesion aspects and some of them have already been linked to poor recovery or adverse transfusion effects. On the other hand, the geometry16 and membrane cation permeability of bThal+ RBC render them osmotically resistant, a probably advantageous feature for RBC at storage conditions.17 Of note, baseline adult hemoglobin A (HbA ) and fetal hemoglobin (HbF) levels in the general donor population have been found to be positively associated with resistance of stored RBC to stress hemolysis.18 Moreover, bThal+ RBC exhibit low aggregability,19 probably rendering them less susceptible to the storage-induced tendency to cell aggregation. Improved metabolic20 and total cardiovascular risk profiles,21 along with survival following malaria infection22 have been also reported thalassemia traits. The aim of the present study was to clarify whether the homeostasis of these unique RBC acts positively or negatively towards the challenges of blood banking. 2

Methods Biological samples and blood unit preparation Venous blood from n=204 regular blood donors was collected into EDTA, citrate and serum vacutainer tubes (in vivo study). Thirty-eight donors (18 bThal+ and 20 controls) were selected to evaluate RBC storability in citrate-phosphate-dextrose (CPD)/ saline-adenine-glucose-mannitol (SAGM) for 42 days at 4°C. The two donor groups exhibited typical hematological differences between them but minimal baseline variation in sex, age, donation frequency and other demographics (Online Supplementary Table S1). bThal+ trait was confirmed by Hb electrophoresis and molecular identification of mutations. The study was approved by the Ethics Committee of the Department of

haematologica | 2022; 107(1)

Biology, School of Science, NKUA. Investigations were carried out upon donor consent, in accordance with the principles of the Declaration of Helsinki.

Physiological parameters Free Hb levels were measured in plasma/supernatant through spectrophotometry,23 followed by the Allen correction. In order to examine the osmotically induced hemolysis, RBC were exposed to solutions of increasing saline (NaCl) concentration and the mean corpuscular fragility (MCF, concentration of NaCl at 50% hemolysis) was calculated. The mechanical fragility index (MFI) was determined by measuring the amount of Hb released in the supernatant of RBC rocked with stainless steel beads for 1 hour (h). Oxidative hemolysis levels were evaluated following treatment of RBC with 17 mM phenylhydrazine (PHZ) for 1 h at 37°C. Reactive oxygen species (ROS) and calcium accumulation were measured by fluorometry; the extracellular antioxidant activity and lipid peroxidation were determined spectrophotometrically; phosphatidylserine (PS) exposure and RBC membrane protein carbonylation were estimated by multicolor flow cytometry and western blotting, respectively (see the Online Supplementary Methods for details).

Omics analyses Preliminary proteomics analyses were performed in stored samples of RBC membranes through a filter-aided sample preparation digestion prior to analysis via nano-ultra performance liquid chromatography - tandem mass spectrometer (nanoUHPLC-MS/MS) (Evosep One system coupled to timsTOF Pro mass spectrometer - Bruker Daltonics, Bremen, Germany), as extensively described in prior technical notes.24 Metabolomics analyses were performed as previously reported.25 100 mL of stored RBC were collected on a weekly basis, extracted at 1:6 dilution in methanol:acetonitrile:water (5:3:2) and analyzed by UHPLC-MS (Ultimate 3000 RSLC-Q Exactive, Thermo Fisher) (see the Online Supplementary Methods for details). Metabolite assignment was performed against an in house standard library, as reported,26 through the freely available software Maven (Princeton University, USA). No data pre-processing (neither normalization nor log-transformation) was performed.

Statistical analysis For statistical analysis (SPSS Version 22.0, IBM Corp, NKUA) Shapiro-Wilk test and detrended normal Q–Q plots (for testing normal distribution and outliers), independent t-test or repeated measures analysis of variance (ANOVA) with Bonferroni-like adjustment (for intergroup differences) and Pearson's or Spearman's tests (for correlation analysis) were used. Variables that exhibited repeatable correlations between each individual storage day (six time points) and fresh blood were used for the construction of in vivo/ex vivo biological networks (Cytoscape 3.7.2). Receiver operating characteristic (ROC) curves were used to find out parameters strongly indicative of bThal+ status in stored RBC. Significance was accepted at P<0.05.

Results Baseline features and storability of beta thalassemia red blood cells In a cohort of 204 eligible blood donors the bThal+ subgroup was approximately 9%. As expected, lower Hb concentration and RBC indices but higher RBC count and HbA concentration were measured in bThal+ donors (ROC curve: area under the curve [AUC] for mean cor2

113

V.L. Tzounakas et al.

puscular volume [MCV] =0.965, for HbA = 1,000), who carried an array of b-Hb mutations commonly found in Greece (IVS I-1, IVS I-6, IVS I-110, IVS II-1 and IVS II745). While osmotic fragility was significantly lower in bThal+ RBC, free Hb levels were similar to control. A trend for high total antioxidant capacity (TAC) and low plasma protein carbonylation was also observed in bThal+ (Table 1). For RBC storability analysis, the subgroup of bThal+ donors (n=18) was compared to an equivalent group of average controls (n=20). The bThal+ units exhibited lower levels of free Hb (storage, oxidative, mechanical and osmotic hemolysis) either throughout the storage period (mechanically and osmotically induced hemolysis) or for the last 2 weeks of it (storage and oxidative hemolysis) (Figure 1A). According to ROC curve analysis, the osmotic, mechanical and storage hemolysis have very good potential to predict the bThal+ status at every time point of storage (Online Supplementary Figure S1A). Of note, only osmotic fragility (which follows a logarithmic increase during storage) was lower in the bThal+ versus control RBC in vivo. Apart from hemolysis, the extracellular K+ was also lower in bThal+ plasma and day 42-supernatant (Figure 1B). Preliminary proteomics analysis revealed differences between the two groups in the abundance of membrane proteins physiologically related to the RBC volume control/cation homeostasis, including piezo-1, Na+/K+ ATPase and aquaporin-1 (Figure 1C). Higher TAC and uric acid-dependent antioxidant capacities (UAdAC) were measured in the plasma/supernatant of the bThal+ samples compared to control (Figure 2A). ROC curve analysis revealed variations in both TAC and UAdAC, strongly indicative of the bThal+ status in stored RBC (Online Supplementary Figure S1B), in contrast to the uric acid-independent antioxidant capacity (UAiAC) levels that were very similar to control (Figure 2A). Membrane lipid peroxidation and protein carbonylation were significantly lower in the bThal+ versus control RBC throughout the storage period or from middle storage onwards, respectively (Figure 2B). Spontaneous intracellular ROS levels were quite similar between the two donor groups at early storage, but lower levels were detected in bThal+ versus control RBC in the late period (Figure 2C). Stress-induced (e.g., by tBHP, diamide, 2

phenylhydrazine) ROS generation basically resulted in no between-group differences. Although significantly higher at baseline, the intracellular Ca2+ of bThal+ RBC exhibited only a trend toward higher levels during storage compared to controls (Figure 2D). In contrast, while similar at baseline (and for the first weeks of storage), PS externalization was significantly lower in bThal+ RBC for the last 2 weeks (Figure 2E). According to a preliminary proteomics analysis, the membrane expression of calcium related proteins (such as calmodulin, calpain and annexin A7/synexin) and of the lipid remodeling protein scramblase differed significantly between the two groups of stored RBC (Figure 2F).

Metabolic profile of beta thalassemia red blood cells in vivo and during storage in CPD-SAGM Full metabolomics analyses were performed in paired fresh and stored samples from the control and bThal+ groups (n=15, Online Supplementary Figures S2 to S11). In the case of fresh RBC (Figure 3A) partial least square-discriminant analysis (PLS-DA) separated the two groups across principal component 1 (PC1), explaining approximately 11% of the total variance (Figure 3B). The top 25 metabolic changes between the two groups – as determined by t-test – are highlighted in the heat map in Figure 3C. Increases were observed in bThal+ RBC with respect to metabolites derived from glycolysis or branching pathways (2,3-diphosphoglycerate, phosphoglycerate isomers, phosphoenolpyruvate) and PPP (6-phosphogluconate, ribose mono and diphosphate, NADPH) (Figure 3C). Overall, these steady state analyses are suggestive of activation of PPP and glycolysis upstream to pyruvate kinase in fresh bThal+ RBC, in the absence of significant alterations of glutathione pools and turn-over (Figure 4A to C, respectively). Indeed, bThal+ RBC were characterized by decreases in pyruvate, arginine, creatine, glycine, inosine and monophosphate (Figure 3C), suggestive of alterations in purine deamination/oxidation and arginine/polyamine metabolism (Figure 4D to E). Metabolomics analyses were thus performed on stored RBC units from the same donors on a weekly basis (Figure 3D). Unsupervised principal component analysis (PCA) and hierarchical clustering analysis of significant metabolites by repeated measures ANOVA are shown in

Table 1. Selected characteristics of beta thalassemia blood minor in vivo compared to control. Red blood cells (x10 /mL) Hematocrit (%) Total Hemoglobin (g/dL) Mean corpuscular volume (fL) Mean corpuscular hemoglobin (pg) Mean corpuscular hemoglobin concentration (g/dL) Red cell distribution width (%) Mean platelet volume (fL) Mean corpuscular fragility (% [NaCl]) Free hemoglobin (mg/dL) Total antioxidant capacity of plasma (TAC) (mM Fe2+) Carbonylated proteins of plasma (nmol/mg) Total plasma proteins (g/dL) 6

Control (n=186)

bThal+ (n=18)

4.94 ± 0.33 44.07 ± 2.58 15.00 ± 1.07 89.38 ± 4.17 30.44 ± 2.08 34.00 ± 1.09 12.48 ± 0.57 7.47 ± 1.57 0.438 ± 0.025 3.61 ± 1.83 622 ± 94.2 0.350 ± 0.107 7.19 ± 0.46

5.88 ± 0.34* 39.84 ± 2.20$ 13.36 ± 0.71* 67.07 ± 6.72* 22.51 ± 2.42* 32.69 ± 0.67$ 13.64 ± 0.88 9.24 ± 2.17$ 0.389 ± 0.028* 2.73 ± 1.24 735 ± 142$ 0.296 ± 0.111$ 6.78 ± 0.56$

Data are presented as mean ± standard deviation. *P<0.05; ($) marginal significance. bThal+: beta thalassemia minor.

114

haematologica | 2022; 107(1)

The favorable storability of bThal+ RBC

Figure 3E and F, respectively. Of note, several of the metabolic differences between the two groups at baseline were further exacerbated during storage, with significantly lower levels of IMP, hypoxanthine and adenosine monophosphate (AMP) in the stored bThal+ RBC (Figure 5A). On the other hand, the bThal+ RBC had higher levels of urate throughout storage. Since human RBC lack a functional uricase, chemical oxidation of urate promotes further catabolism to allantoin and allantoate, that were found significantly lower and higher in the bThal+ RBC, respectively, throughout storage (Figure 5A). These

observations were accompanied by a higher degree of glutamine consumption and glutamate generation in bThal+ RBC after storage day 7 (Figure 5A; Online Supplementary Figure S8, respectively), suggestive of ongoing glutaminolysis. Interestingly, metabolites of purine metabolism such as urate, allantoate and glutamine represent very good predictors of bThal+ status in stored RBC (ROC curve analysis, Online Supplementary Figure S1C). Intertwined with purine (especially adenosine) metabolism, S-adenosylmethionine (SAM) synthesis and consumption were not significantly altered in

Figure 1. Variation in hemolysis variables of fresh and stored red blood cells in beta thalassemia minor and control donors. (A) Time course evaluation of storage, osmotic, mechanical and oxidative hemolysis. Dashed lines represent trend lines of exponential or logarithmic models for storage and osmotic hemolysis, respectively. (B) Potassium leakage. (C) Storage levels of selected proteins related to osmotic/mechanical hemolysis. Data are presented as mean ± standard deviation. *P<0.05; F: fresh blood; A.U.: arbitrary units; bThal+: beta thalassemia minor.

haematologica | 2022; 107(1)

115

V.L. Tzounakas et al.

Figure 2. Redox status and calcium signaling of red blood cells from beta thalassemia minor and control donors. (A) Time course variation of total (TAC), uric acidindependent (UAiAC) and uric acid- dependent (UAdAC) antioxidant capacity. (B) Membrane lipid peroxidation and protein carbonylation (n=8 per group; 4.1R protein was used for internal control). (C) Variation of endogenous/exogenously induced (tert-butyl-hyperoxide; tBHP, diamide and phenylhydrazine; PHZ) reactive oxygen species (ROS) levels. (D) Intracellular calcium accumulation. (E) Phosphatidylserine (PS) exposure. (F) Storage levels of selected proteins related to calcium homeostasis and lipid remodeling. Dashed lines represent trendlines of logarithmic models. Data are presented as mean ± standard deviation. *P<0.05; F: fresh blood; TBA: thiobarbituric acid; bThal+: beta thalassemia minor.

116

haematologica | 2022; 107(1)

The favorable storability of bThal+ RBC

RBC from the bThal+ group (Figure 5B). Cysteine levels were higher and cysteate levels lower in the bThal+ group, especially past storage day 28 and at the end of storage (Figure 5B). Significantly higher levels of sphingosine 1-phosphate (S1P) – a marker of systemic hypoxemia27 - were observed in RBC from the bThal+ group on day 0 and throughout storage (Figure 5C). Like in fresh RBC, changes were observed in arginine metabolism between the two groups as a result of storage duration – with RBC from the bThal+ group being charac-

terized by significantly lower arginine and creatine and higher asymmetric dimethyl-arginine and acetyl-spermidine levels throughout storage (Figure 6A). Of all the pathways assessed in this study, the most notable betweengroup changes were unexpectedly found in vitamin B6 metabolism (Figure 6B) and in the hexosamine pathway (Figure 6C). bThal+ RBC were characterized by lower pyridoxate and pyridoxal and higher pyridoxamine levels throughout storage. On the other hand, all intermediates of the hexosamine pathway (uridine monophosphate -

Figure 3. Metabolomics analyses of fresh and stored red blood cells from control and beta thalassemia minor donors. (A) Fresh blood analysis of eligible blood donors (n=13 for controls and n=12 for beta thalassemia minor [bthal+] donors). (B) Principal component and discriminant analysis in fresh blood. (C) Heat map of top 25 metabolites differing between bthal+ and control fresh red blood cells (RBC). (D) Metabolomics analysis was performed in RBC stored in leukoreduced CPD/SAGM units (n=15 per group). (E) Principal component and discriminant analysis in stored RBC. (F) Heat map of variation in RBC metabolites throughout storage. CPD/SAGM: citrate-phosphate-dextrose/saline-adenine-glucose-mannitol.

haematologica | 2022; 107(1)

117

V.L. Tzounakas et al.

UMP, acetyl-glucosamine, acetyl-glucosamine phosphate) were significantly higher in bThal+ RBC at all storage time points, while the final product of this pathway – uridine diphosphate (UDP) N-acetyl-glucosamine was significantly lower in bThal+ RBC (Figure 6C), suggestive of either a

blockade at the metabolic step catalyzed by UDP-Nacetylglucosamine phosphorylase or an increased consumption of its product for protein O-GlcNAcylation (post-translational attachment of O-linked N-acetylglucosamine moieties to serine and threonine residues).

A B

Figure 4. Main metabolic differences between control and beta thalassemia minor donors in circulating red blood cells. (A) Glycolysis. (B) Pentose phosphate pathway (PPP) metabolites. (C) Glutathione oxidation. (D) Purine metabolism. (E) Arginine metabolism. SAM: S-adenosyl methionine. *P<0.05, **P<0.01, #P<0.001, †P<0.0001.

118

haematologica | 2022; 107(1)

The favorable storability of bThal+ RBC

Again, ROC curve analysis highlighted the potential of glucosamines to distinguish stored bThal+ RBC from their control counterparts (Online Supplementary Figure S1D). Finally, RBC from bThal+ donors were characterized by higher levels of carnosine and indole-acetate – metabolites

with either antioxidant activity or associated microbial metabolism, respectively, from baseline throughout storage, with no storage-dependent fluctuations, suggestive of differences in the circulating blood metabolome between the two groups at the time of donation (Figure 6D).

B C

Figure 5. Distinct metabolic pathways in beta thalassemia minor donors during storage I. (A) Purine metabolism. (B) Methionine metabolism. (C) Hypoxia modulated hemoglobin loading via a sphingosine-1-phosphate effect. SAM: S-adenosyl methionine; SAH: S-adenosyl homocysteine; RibP: ribose phosphate. *P<0.05, **P<0.01, #P<0.001, †P<0.0001.

haematologica | 2022; 107(1)

119

V.L. Tzounakas et al.

Intertwining biological networks in stored red blood cells and supernatants In order to find out major aspects of bThal+ RBC physiology, able to interconnect the in vivo/ex vivo states, and thus, to reveal the “effect” (if any) of baseline variables

upon storage, we proceeded to comparative bThal+ versus control in vivo/ex vivo network analysis. Our findings highlighted osmotic hemolysis as a “hub” parameter in bThal+ networks, with significant differences in the type and the amount of connections between

Figure 6. Distinct metabolic pathways in beta thalassemia minor donors during storage II. (A) Arginine metabolism. (B) Vitamin B6 metabolism. (C) Hexosamine pathway. (D) Carnosine and indole-acetate. SAM: S-adenosyl methionine; ADMA: asymmetric dimethylarginine; *P<0.05, **P<0.01, #P<0.001, †P<0.0001.

120

haematologica | 2022; 107(1)

The favorable storability of bThal+ RBC

Figure 7. Network analysis of statistically significant repeated correlations between parameters of fresh blood and stored red blood cells. (A) Osmotic fragility hub. (B) Plasma/supernatant antioxidant activity hub. Solid black lines: positive correlation; dashed red lines: negative correlation. MCF: mean corpuscular fragility; TAC/UAdAC/UAiAC: total, uric acid dependent and uric acid independent antioxidant capacities. Numbers correspond to the definitions provided in the Online Supplementary Table S2.

haematologica | 2022; 107(1)

121

V.L. Tzounakas et al.

the groups. The osmotic fragility of stored bThal+ RBC exhibited strong correlations with several biochemical (e.g., transferrin, r=0.560, P<0.01), cellular (e.g., RBC mechanical hemolysis, r=0.771, P<0.01), protein and metabolic (e.g., L-arginine, r=0.697, sphingosine-1-phosphate, r=-0.636, P<0.01) parameters of bThal+ RBC and plasma, in striking contrast to the controls (Figure 7A). Notably, in the case of bThal+ RBC, fragility variation during storage was less correlated to the in vivo levels of fatty acids but more correlated to those of carboxylic acids, compared to control. Despite the above-mentioned differences, osmotic hemolysis during storage was found interrelated with the in vivo levels of RBC aging markers and energy metabolism in both donor groups. In similarity to the MCF network, the extracellular antioxidant capacity of the bThal+ units presented higher connectivity with baseline parameters compared to control (twice the number of connections) including several metabolites of energy metabolism (e.g., Dglucose/UAdAC, r=-0.705, P<0.01), purine oxidation (e.g., TAC/5-hydroxyisourate, r=0.756, P<0.01), urea cycle (e.g., ornithine/UAdAC, r=-0.769, P<0.01) and carnitines (Figure 7B). Those findings suggested that that the basic physiology of bThal+ is strongly and intrinsically related with the unique resistance to osmotic hemolysis and the high antioxidant capacity of the RBC units throughout storage.

Discussion According to an increasing number of studies, genetic diversity among donors substantially affects RBC storage stability. Results of the REDS-III-Omics study showed that female sex and African American or Asian race-ethnicity2 predispose to low susceptibility to storage and stress hemolysis. RBC from G6PD-deficient donors exhibit control levels of storage hemolysis, opposite to sickle cell trait RBC that are more susceptible to spontaneous hemolysis during storage.28 Our study reports for the first time that bThal+ RBC are superior to those of the general population in terms of all hemolysis measures (including the less studied mechanical one), either throughout the storage period or for the more susceptible last 2 weeks of it. Superior secondary metrics of storage quality (including removal signaling and extracellular potassium) were also detected in bThal+. As expected, the unique storage capacity of bThal+ RBC was found mechanistically linked to changes in specific metabolic flows and membrane protein expression profiles. A tentative explanation for resistance to hemolysis lies in the appreciation of decreased HCT, MCH and Hb levels in bThal+. Similar findings in high frequency blood donors have been associated with decreases in ferritin and total iron pools and improved hemolytic parameters,29 in keeping with a negative role of iron levels and metabolism with RBC storage quality and post-transfusion efficacy.30 Moreover, bThal+ RBC are intrinsically different from control, with reduced cellular volume and increased surface:volume ratio.31 Altered ion transport homeostasis leading to efflux of osmotically active components32 underlies to great extent this phenotype. In support, serum levels of osmotically active ions in bThal+ exhibit significant correlation with RBC osmotic fragility.33 Under normal membrane and skeletal networks 122

(manuscript in preparation) these features contribute to increased resistance to osmotic (and probably to mechanical) stress. The biological networks suggested that the resistance of stored RBC to osmotic stress is a multifactorial phenotype exhibiting numerous correlations with the baseline RBC physiology (such as the energy metabolism) especially in the case of bThal+ RBC. Indeed, the levels of several biochemical and metabolic features of bThal+ RBC in vivo exhibited strong correlations with those of osmotic hemolysis throughout storage, including L-arginine34 and sphingosine-1-phosphate, which significantly differed between the two groups under examination at baseline. Of note, similar correlations were also detected in G6PDdefficient donors.35 This observation suggests a genetically determined aspect of osmotic hemolysis phenotype in stored RBC that deserves further examination. Since osmotic fragility is strongly donor-specific (storage levels are proportional to baseline levels in vivo),17 the superior osmotic hemolysis (and consequently part of storage hemolysis) of bThal+ RBC was rather anticipated. Resistance to oxidative hemolysis at late storage seemed to follow the fluctuation in ROS levels at the same period, as well as the modulation of purine and arginine metabolism, previously reported in control donor groups.36 Osmotic stress and additional hemolysis triggers may be further associated with the specific protein composition of RBC membrane in bThal+, since our preliminary analysis revealed variations in proteins critically involved in cell volume regulation and transmembrane water/cation flows responsive to osmotic and mechanical stimuli, including aquaporin and piezo. Variation in some of these transporters has already been reported in thalassaemic or other anemic subjects.37 Our study (Online Supplementary Figure S1A) showed for the first time that resistance to both spontaneous and induced hemolysis is a distinctive feature of bThal+ RBC storability. However, diversity in bThal+ mutations worldwide should be also taken into consideration. Indeed, certain bThal+ donors exhibited control baseline levels of RBC osmotic fragility (Table 1). The effects of such genetic variation on bThal+ RBC storability and performance deserve further evaluation by future studies in various ethnic groups. Apart from hemolysis, PS exposure (a potent signal for RBC clearance in vivo) was also lower in bThal+ RBC at late storage, when ROS accumulation and membrane expression of phospholipid scramblase38 were also low. In contrast, Ca2+ concentration was higher in bThal+ RBC at baseline (and slightly higher during storage) compared to controls, consistent with the distorted divalent cation homeostasis reported in bThal+.39 This finding deserves further attention in the light of bThal+-specific variation in Ca2+-dependent membrane proteins that are involved in critical signaling events in RBC. Finally, alterations in the hexosamine pathway in bThal+ RBC are suggestive of either up-regulation of UDP-N-acetyl-glucosamine usage for O-GlcNAcylation or decreased synthesis through blockade of the late steps in this biosynthetic pathway. To the best of our knowledge, this is the first report of this pathway being affected in bThal+. Interestingly, OGlcNAcylation is a common post-translational modification in Plasmodium falciparum proteins following malaria infection, to the extent that therapeutic blockade of this pathway has been proposed as an intervention to shorten the life cycle of the Plasmodium.40 As such, our observahaematologica | 2022; 107(1)

The favorable storability of bThal+ RBC

tions may contribute to explain one of the molecular mechanisms through which bThal+ protects from malaria infection.22 Reports on the oxidative profile of bThal+ RBC and on the plasma antioxidant capacity in vivo are quite contradicting.12 Many studies, however, pointed to increased activity and/or upregulation of several antioxidant enzymes and proteins41 in bThal+ RBC as a possible adaptation to mild sustained oxidative stress. Metabolism exhibits a similar adaptive response according to the currently reported data showing lower baseline levels of oxidant stress-induced purine deamination, higher levels of sulfur-containing metabolites (methionine and cysteine), increase in glycolysis upstream of pyruvate kinase and an apparent increase in fluxes through the PPP in bThal+ RBC. Combined with increases in PPP-derived reducing equivalents (NADPH), in the absence of significant alterations of total glutathione pools and ratios of oxidized/reduced forms, these data are suggestive of an overall higher antioxidant armamentarium in bThal+ RBC. “Training” to mild oxidative stress in vivo may modulate bThal+ RBC storability,41 as evidenced by ROS accumulation, oxidative hemolysis, and oxidative defects to membrane components. In the same context, the membrane profile of stress markers and the low intracellular allantoin42 are suggestive of the protective action of molecular chaperones and other antioxidant molecules, such as urate, present in excess in bThal+ RBC. Despite the fact that urate can act as a pro- or anti-oxidant factor, its minimal oxidation to allantoin at the same period of low ROS generation in bThal+ RBC points towards its antioxidant effect during storage. Moreover, it mostly accounted for the elevated antioxidant capacity of bThal+ plasma/supernatant. Network paired analysis of fresh versus stored samples confirmed that the extracellular antioxidant capacity (like osmotic fragility) is an almost “heritable” feature of RBC units in both donor groups,43 and further “signs” bThal+ status in stored RBC (Online Supplementary Figure S1). Of note, the levels of TAC and UAdAC in bThal+ units exhibited inverse correlations with acyl-carnitine RBC metabolites, such as the redox related acetyl-carnitine. These metabolites have been suggested as markers of osmotic fragility in blood donors undergoing testosterone-replacement therapy,44 consistent with a role of this pathway in the Lands cycle-dependent remodeling of oxidized membrane lipids. Genetic variation among donors may have a positive or negative impact on transfusion performance. G6PD-deficient RBC, for instance, are susceptible to hemolysis and oxidative stress post-transfusion,6 and exhibit lower 24hour recovery in autologous transfusion recipients.7 Similarly, sickle cell trait RBC were shown to have reduced recovery in animal models.28 Quite to the contrary, the unique storability of bThal+ RBC is suggestive of good post-transfusion recovery. Apart from the low intracellular levels of hypoxanthine, a metabolic marker of storage lesion exhibiting negative correlation with recovery in vivo,45 the low extracellular Hb/potassium and RBC removal signaling are in favor of therapeutic outcomes mainly through preservation of nitric oxide bioavailability and alleviation of hemolysis/iron-associated toxicity, including inflammatory responses and bacterial infections.46 In addition, the advantageous redox phenotype of stored bThal+ is expected to promote their functionality in vivo and to ameliorate oxidative stresshaematologica | 2022; 107(1)

associated adverse clinical outcomes in susceptible patients with G6PD-deficiency and sickle cell disease.47 Finally, the modified arginine/nitric oxide metabolism (involved in the induction of HbF48) in bThal+ RBC during storage may be relevant to their transfusion in susceptible recipients, including hemoglobinopathies, cancer, lung and cardiovascular disease patients. In the light of those results, we have initiated a study on post-transfusion performance of bThal+ RBC. Heterozygotes for bThal+ constitute a non-negligible proportion of blood donors not only in the Mediterranean, but also in Africa, Asia and in under-represented minorities of US.49 Their typical profile of low Hb concentration, and thus, of lower dose of Hb per unit of RBC, would predispose to reduced Hb increment and lower efficacy of the transfusion therapy compared to control. However, the currently presented data show that bThal+ RBC are unique in terms of susceptibility to hemolysis, redox homeostasis and nitrogen/hexosaminerelated metabolism during storage. This favorable storability and thereof improved quality of stored bThal+ RBC may counterpoise the negative effects of lower Hb dose on the recipient. Despite the fact that further studies are needed to clarify the post-transfusion performance of bThal+ RBC in the context of intra-group genetic heterogeneity and recipient variation, the currently reported data may allow considering bThal+ individuals as a novel, high quality, “good storer” group. Indeed, bThal+ RBC may be safely stored for long time periods before transfusion, highlighting the notion of molecular versus storage age of blood.5 Moreover, bThal+ RBC are probably ideal candidates for alternative storage strategies, like cryopreservation, which involves a deglycerolization step during which RBC are highly susceptible to hypo-osmotic lysis.50 These promising data and potentials highlight the emergence of bThal+ donors for a distinct and potentially valuable role in blood transfusions. Disclosures Though unrelated to the contents of this manuscript, ADA declares that he is a founder of Omix Technologies Inc and Altis Biosciencens LLC and a consultant for Hemanext Inc. All other authors declare no conflicts of interest. Contributions VLT and ATA are co-first authors; VLT and MHA designed the study; VLT, ATA, MHA and ADA wrote the manuscript; VLT, ATA and AGK performed the physiological and biochemical experiments; DS, FC, LB and FG performed the metabolomics analyses; MD performed the proteomics experiments; MHA and PR performed the flow cytometry analyses; VLT, ATA, MHA and ADA analyzed the data; AV, VS, KS, and AGK provided project support and organized participant recruitment; MD, OET, ISP, KS, AGK and ADA provided expert knowledge. Acknowledgements The authors would like to thank all blood donors that voluntarily participated in this study; M.S. Jacovides Hellas S.A. for the kind offer of the LTRC blood bags; Dr Pantelis Constantoulakis and Dr Stavros Bournazos (Genotypos Science Labs) for their contribution in the mutation identification; the NKUA students Dimitrios G. Karadimas (MSc) and Christos Christogeorgos for their participation in a part of the physiological experiments performed; Dr Hara Georgatzakou, Mrs 123

V.L. Tzounakas et al.

Maria Sioumala and the UniWA students Efthimia Pavlou, Panagiotis Drosos, Maria Tsoumpeli, Eirini Tsolou and Kariofyllis Karamperis for their contribution in blood sampling and blood cell counting measurements performed in fresh samples or stored RBC units.

References 1. Tzounakas VL, Kriebardis AG, Papassideri IS, Antonelou MH. Donor-variation effect on red blood cell storage lesion: A close relationship emerges. Proteomics Clin Appl. 2016;10(8):791-804. 2. Kanias T, Lanteri MC, Page GP, et al. Ethnicity, sex, and age are determinants of red blood cell storage and stress hemolysis: results of the REDS-III RBC-Omics study. Blood Adv. 2017;1(15):1132-1141. 3. Stefanoni D, Fu X, Reisz JA, et al. Nicotine exposure increases markers of oxidant stress in stored red blood cells from healthy donor volunteers. Transfusion. 2020;60(6):11601174. 4. Roubinian NH, Plimier C, Woo JP, et al. Effect of donor, component, and recipient characteristics on hemoglobin increments following red blood cell transfusion. Blood. 2019;134(13):1003-1013. 5. D'Alessandro A, Zimring JC, Busch M. Chronological storage age and metabolic age of stored red blood cells: are they the same? Transfusion. 2019;59(5):1620-1623. 6. Tzounakas VL, Kriebardis AG, Georgatzakou HT, et al. Glucose 6-phosphate dehydrogenase deficient subjects may be better "storers" than donors of red blood cells. Free Radic Biol Med. 2016;96:152-165. 7. Francis RO, D'Alessandro A, Eisenberger A, et al. Donor glucose-6-phosphate dehydrogenase deficiency decreases blood quality for transfusion. J Clin Invest. 2020;130(5): 2270-2285. 8. Acquaye CT, Oldham JH, Konotey-Ahulu FI. Blood-donor homozygous for hereditary persistence of fetal haemoglobin. Lancet. 1977;1(8015):796-797. 9. Singha K, Taweenan W, Fucharoen G, Fucharoen S. Erythrocyte indices in a large cohort of beta-thalassemia carrier: Implication for population screening in an area with high prevalence and heterogeneity of thalassemia. Int J Lab Hematol. 2019;41 (4):513-518. 10. Hillier J, Hoffman JF, Parpart AK, Wolman IJ. Ultrastructure of erythrocyte membranes in thalassemia major and minor. Blood. 1956;11(10):946-956. 11. Ficarra S, Tellone E, Giardina B, et al. Derangement of erythrocytic AE1 in betathalassemia by caspase 3: pathogenic mechanisms and implications in red blood cell senescence. J Membr Biol. 2009;228(1):4349. 12. Manafikhi H, Drummen G, Palmery M, Peluso I. Total antioxidant capacity in betathalassemia: a systematic review and metaanalysis of case-control studies. Crit Rev Oncol Hematol. 2017;110:35-42. 13. Fortier N, Snyder LM, Garver F, Kiefer C, McKenney J, Mohandas N. The relationship between in vivo generated hemoglobin skeletal protein complex and increased red cell membrane rigidity. Blood. 1988;71(5): 1427-1431. 14. Pantaleo A, De Franceschi L, Ferru E, Vono R, Turrini F. Current knowledge about the functional roles of phosphorylative changes

124

Funding This project has received funding from the Hellenic Foundation for Research and Innovation (HFRI) and the General Secretariat for Research and Technology (GSRT), under grant agreement No 2032 (VLT).

of membrane proteins in normal and diseased red cells. J Proteomics. 2010;73(3):445455. 15. Mandal D, Baudin-Creuza V, Bhattacharyya A, et al. Caspase 3-mediated proteolysis of the N-terminal cytoplasmic domain of the human erythroid anion exchanger 1 (band 3). J Biol Chem. 2003;278(52):52551-52558. 16. Schrier SL, Rachmilewitz E, Mohandas N. Cellular and membrane properties of alpha and beta thalassemic erythrocytes are different: implication for differences in clinical manifestations. Blood. 1989;74(6):21942202. 17. Tzounakas VL, Anastasiadi AT, Karadimas DG, et al. Temperature-dependent haemolytic propensity of CPDA-1 stored red blood cells vs whole blood - Red cell fragility as donor signature on blood units. Blood Transfus. 2017;15(5):447-455. 18. Tzounakas VL, Georgatzakou HT, Kriebardis AG, et al. Donor variation effect on red blood cell storage lesion: a multivariable, yet consistent, story. Transfusion. 2016;56(6):1274-1286. 19. Falco C, Vaya A, Iborra J, Moreno I, Palanca S, Aznar J. Erythrocyte aggregability and disaggregability in thalassemia trait carriers analyzed by a laser backscattering technique. Clin Hemorheol Microcirc. 2003;28(4):245-249. 20. Triantafyllou AI, Vyssoulis GP, Karpanou EA, et al. Impact of beta-thalassemia trait carrier state on cardiovascular risk factors and metabolic profile in patients with newly diagnosed hypertension. J Hum Hypertens. 2014;28(5):328-332. 21. Maioli M, Pettinato S, Cherchi GM, et al. Plasma lipids in beta-thalassemia minor. Atherosclerosis. 1989;75(2-3):245-248. 22. Kuesap J, Chaijaroenkul W, Rungsihirunrat K, Pongjantharasatien K, Na-Bangchang K. Coexistence of malaria and thalassemia in malaria endemic areas of Thailand. Korean J Parasitol. 2015;53(3):265-270. 23. Harboe M. A method for determination of hemoglobin in plasma by near-ultraviolet spectrophotometry. Scand J Clin Lab Invest. 1959;11:66-70. 24. Reisz JA, Chessler KM, Dzieciatkowska M, D'Alessandro A, Hansen KC. Blood and plasma proteomics: targeted quantitation and posttranslational redox modifications. Methods Mol Biol. 2017;1619:353-371. 25. Nemkov T, Reisz JA, Gehrke S, Hansen KC, D'Alessandro A. High-throughput metabolomics: isocratic and gradient mass spectrometry-based methods. Methods Mol Biol. 2019;1978:13-26. 26. Nemkov T, Hansen KC, D'Alessandro A. A three-minute method for high-throughput quantitative metabolomics and quantitative tracing experiments of central carbon and nitrogen pathways. Rapid Commun Mass Spectrom. 2017;31(8):663-673. 27. Sun K, Zhang Y, D'Alessandro A, et al. Sphingosine-1-phosphate promotes erythrocyte glycolysis and oxygen release for adaptation to high-altitude hypoxia. Nat Commun. 2016;7:12086.

28. Osei-Hwedieh DO, Kanias T, Croix CS, et al. Sickle cell trait increases red blood cell storage hemolysis and post-transfusion clearance in mice. EBioMedicine. 2016;11:239-248. 29. Kanias T, Stone M, Page GP, et al. Frequent blood donations alter susceptibility of red blood cells to storage- and stress-induced hemolysis. Transfusion. 2019;59(1):67-78. 30. Howie HL, Hay AM, de Wolski K, et al. Differences in Steap3 expression are a mechanism of genetic variation of RBC storage and oxidative damage in mice. Blood Adv. 2019;3(15):2272-2285. 31. Gunn RB, Silvers DN, Rosse WF. Potassium permeability in -thalassemia minor red blood cells. J Clin Invest. 1972;51(5):10431050. 32. Olivieri O, De Franceschi L, Capellini MD, Girelli D, Corrocher R, Brugnara C. Oxidative damage and erythrocyte membrane transport abnormalities in thalassemias. Blood. 1994;84(1):315-320. 33. Tzounakas V, Drossos P, Karadimas D, et al. Hematological and physiological characteristics of regular blood donors with beta-thalassemia traits. Vox Sang. 2019;114(s1): abstract P-059. 34. Belfiore F, Meldolesi J, Calcara G. Erythrocyte enzymes in thalassaemia and thalassodrepanocytosis. Acta Haematol. 1965;34(6):329-337. 35. Reisz JA, Tzounakas VL, Nemkov T, et al. Metabolic linkage and correlations to storage capacity in erythrocytes from glucose 6phosphate dehydrogenase-deficient donors. Front Med (Lausanne). 2017;4:248. 36. D'Alessandro A, Fu X, Kanias T, et al. Donor sex, age and ethnicity impact stored red blood cell antioxidant metabolism through mechanisms in part explained by glucose 6-phosphate dehydrogenase levels and activity. Haematologica. 2021;106(5): 1290-1302. 37. Omar AK, Ahmed KA, Helmi NM, et al. The sensitivity of Na(+), K(+) ATPase as an indicator of blood diseases. Afr Health Sci. 2017;17(1):262-269. 38. Wesseling MC, Wagner-Britz L, Nguyen DB, et al. Novel insights in the regulation of phosphatidylserine exposure in human red blood cells. Cell Physiol Biochem. 2016;39 (5):1941-1954. 39. Shalev O, Mogilner S, Shinar E, Rachmilewitz EA, Schrier SL. Impaired erythrocyte calcium homeostasis in beta-thalassemia. Blood. 1984;64(2):564-566. 40. Kupferschmid M, Aquino-Gil MO, ShamsEldin H, et al. Identification of OGlcNAcylated proteins in Plasmodium falciparum. Malar J. 2017;16(1):485. 41. Teran MM, Monaco ME, Lazarte SS, et al. Genetic regulation of redox balance in betathalassemia Trait. Hemoglobin. 2020;44(2): 122-127. 42. Kand'ar R, Zakova P. Allantoin as a marker of oxidative stress in human erythrocytes. Clin Chem Lab Med. 2008;46(9):12701274. 43. Tzounakas VL, Karadimas DG, Anastasiadi AT, et al. Donor-specific individuality of red

haematologica | 2022; 107(1)

The favorable storability of bThal+ RBC

blood cell performance during storage is partly a function of serum uric acid levels. Transfusion. 2018;58(1):34-40. 44. Alexander K, Hazegh K, Fang F, et al. Testosterone replacement therapy in blood donors modulates erythrocyte metabolism and susceptibility to hemolysis in cold storage. Transfusion. 2021;61(1):108-123. 45. Nemkov T, Sun K, Reisz JA, et al. Hypoxia modulates the purine salvage pathway and decreases red blood cell and supernatant levels of hypoxanthine during refrigerated stor-

haematologica | 2022; 107(1)

age. Haematologica. 2018;103(2):361-372. 46. Hod EA, Zhang N, Sokol SA, et al. Transfusion of red blood cells after prolonged storage produces harmful effects that are mediated by iron and inflammation. Blood. 2010;115(21):4284-4292. 47. Kanias T, Acker JP. Biopreservation of red blood cells--the struggle with hemoglobin oxidation. FEBS J. 2010;277(2):343-356. 48. Cokic VP, Smith RD, Beleslin-Cokic BB, et al. Hydroxyurea induces fetal hemoglobin by the nitric oxide-dependent activation of

soluble guanylyl cyclase. J Clin Invest. 2003;111(2):231-239. 49. Meena LP, Kumar K, Singh VK, Bharti A, Rahman SK, Tripathi K. Study of mutations in beta-thalassemia trait among blood donors in eastern Uttar Pradesh. J Clin Diagn Res. 2013;7(7):1394-1396. 50. Anastasiadi AT, Tzounakas VL, Kriebardis AG, Stamoulis KE, Seghatchian J, Antonelou MH. When I need you most: frozen red blood cells for transfusion. Transfus Apher Sci. 2020;59(3):102786.

125

ARTICLE Ferrata Storti Foundation

Bone Marrow Failure

Predicting response of severe aplastic anemia to immunosuppression combined with eltrombopag Yoshitaka Zaimoku,1 Bhavisha A. Patel,1 Ruba Shalhoub,2 Emma M. Groarke,1 Xingmin Feng,1 Colin O. Wu2 and Neal S. Young1 Hematology Branch, National Heart, Lung, and Blood Institute, National Institutes of Health and 2Office of Biostatistics Research, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD, USA

Haematologica 2022 Volume 107(1):126-133

ABSTRACT

Correspondence: YOSHITAKA ZAIMOKU zaimokuyoshitaka@gmail.com Received: January 24, 2021. Accepted: April 2, 2021. Pre-published: April 29, 2021. https://doi.org/10.3324/haematol.2021.278413

retreatment blood counts, particularly an absolute reticulocyte count ≥25×109/L, correlate with response to immunosuppressive therapy in severe aplastic anemia. In recent trials, eltrombopag combined with standard immunosuppressive therapy yielded superior responses than those to immunosuppressive therapy alone. Our single institution retrospective study aimed to elucidate whether historical predictors of response to immunosuppressive therapy alone were also associated with response to immunosuppressive therapy plus eltrombopag. We sought correlations of blood counts, thrombopoietin levels and the presence of paroxysmal nocturnal hemoglobinuria clones with both overall and complete responses in 416 patients with severe aplastic anemia, aged 2-82 years (median, 30 years), initially treated with immunosuppressive therapy plus eltrombopag between 2012 and 2019 (n=176) or with immunosuppressive therapy alone between 1999 and 2010 (n=240). Compared to non-responders, patients in the group of overall responders to immunosuppressive therapy plus eltrombopag had significantly higher pretreatment absolute reticulocyte counts, higher neutrophil counts and reduced thrombopoietin levels, as also observed for the group treated with immunosuppressive therapy alone. Addition of eltrombopag markedly improved the overall response in subjects with an absolute reticulocyte count between 10-30×109/L from 60% (54 of 90) to 91% (62 of 68). Absolute lymphocyte count correlated with complete response in the groups treated with immunosuppressive therapy with or without eltrombopag, especially in adolescents aged ≥10 years and adults, but the correlation was reversed in younger children. Platelet count and the presence of a paroxysmal nocturnal hemoglobinuria clone did not correlate with responses to immunosuppressive therapy. Blood counts remain the best predictors of response to nontransplant therapies in severe aplastic anemia. Addition of eltrombopag to immunosuppressive therapy shifted patients with a lower absolute reticulocyte count into a better prognostic category.

Introduction Immune aplastic anemia (AA), characterized by T-cell-mediated destruction of hematopoietic stem and progenitor cells, is successfully treated with hematopoietic stem cell transplantation (HSCT) or immunosuppressive therapy (IST).1 HSCT is preferred for severe AA (SAA) in children and young adults aged ≤40 years, if an HLA-matched sibling donor is available, as it is usually a curative option and has acceptable toxicities. Most adult patients undergo IST since transplantation-related mortality increases with age.2 IST may result in incomplete hematologic recovery, relapse, or clonal evolution to a myeloid malignancy. Standard IST has been established as horse antithymocyte globulin (hATG) plus cyclosporine, to which 60-70% of patients respond with clinically meaningful hematologic improvement. Experimental clinical protocols conducted at our institution using rabbit ATG, alemtuzumab, cyclophosphamide, or with the addition to standard treatment of

126

haematologica | 2022; 107(1)

Predicting response to IST plus EPAG in SAA

mycophenolate mofetil and sirolimus, and by others with the addition of androgens, granulocyte – colony-stimulating factor and erythropoietin to standard IST have failed to demonstrate superiority to standard hATG and cyclosporine.3 However, eltrombopag, a synthetic thrombopoietin receptor agonist, when added to standard IST resulted in better outcomes compared to the standard regimen, with an overall response rate of >80% and 2-year survival rate of 97%,4 and this combination is now approved for first-line use in SAA in the USA. A large, prospective, randomized trial comparing standard IST to IST plus eltrombopag has broadly confirmed these results,5 and smaller case series in a variety of populations have also suggested improved outcomes with the addition of thrombopoietin mimetics.6-11 Response to conventional ATG-based IST without eltrombopag correlates with blood counts before therapy, including absolute reticulocyte count (ARC),12-18 absolute neutrophil count (ANC)12,16,17,19,20 and absolute lymphocyte count (ALC).12,21,22 The presence of somatic mutations,23 having a paroxysmal nocturnal hemoglobinuria (PNH) clone,13,15,20,24,25 elevated thrombopoietin levels,16 normal telomere length of lymphocytes,15,26 a subpopulation of regulatory T cells27 and B cells28 have all been reported as prognostic markers of response. Studies typically have emphasized overall response rather than complete response, but a positive feature of IST combined with eltrombopag is the high complete response rate.4 Previously, the robustness of response has been correlated with better overall survival.29 The aim of this study was to elucidate whether IST response predictors before the introduction of eltrombopag remain associated with predictors of treatment response with IST plus eltrombopag and to identify subgroups of patients with SAA that might particularly benefit from IST plus eltrombopag. We sought correlations of age, initial blood counts, thrombopoietin levels, and the presence of PNH clones with both the overall and complete responses in a large dataset including groups of patients with SAA treated on consecutive clinical protocols with IST alone and IST plus eltrombopag at the National Heart, Lung, and Blood Institute (NHLBI), National Institutes of Health (NIH).

hATG and cyclosporine; the IST alone group included other hATG-based IST.

Definitions All definitions have been consistent across all our protocols. SAA was diagnosed when bone marrow cellularity was <30% and at least two of the following three criteria were met: ANC <0.5×109/L, ARC <60×109/L, and platelet count <20×109/L. Hematologic responses were evaluated 6 months after administration of ATG. Overall response was defined as no longer meeting the criteria for SAA in the absence of recent transfusions and granulocyte colony-stimulating factor administration; complete response required all ANC ≥1.0×109/L, hemoglobin level ≥10 g/dL and platelet count ≥100×109/L; and partial response was defined as an overall response not meeting the criteria for complete response. For the purpose of statistical analysis, patients who did not complete 6 months of initial IST due to death, HSCT or who underwent a second course of IST were counted as having had no response, consistent with our previous study.12 Blood values assessed as potential response predictors were ARC, ANC, ALC, platelet count, presence of a PNH clone and plasma thrombopoietin levels. The lowest values during the 4 weeks preceding IST were used as pretreatment blood counts, as previously described.12 Plasma thrombopoietin levels were measured by magnetic Luminex assay with a standard protocol.33 PNH clones were defined as the presence of 1% or more clonal populations deficient in glycosylphosphatidylinositol anchored proteins on the surface of neutrophils or red blood cells; information on PNH clones was available only for neutrophils in the IST plus eltrombopag study.

Statistics Statistical analyses were performed using an EZR software package (version 1.41), graphical user interphase of R (version 3.6.1).34 The Fisher exact test and Mann-Whitney U test were used to compare categorical and numerical variables, respectively. Correlations between baseline numerical variables were tested with the Spearman rank correlation test. In the logistic regression model, continuous variables were transformed to the natural log (X+1), where X represents the variable, consistent with the previous study.12 All variables with P<0.2 in the univariate analysis were included in the multivariate analysis and a backward stepwise selection procedure was performed to create the final model.

Methods Results Patients and treatment Patients with SAA, aged ≥2 years, treated with hATG-based regimens at the NIH Clinical Center, Bethesda, between May 1999 and August 2019, were included in the analysis. Studies before 1999 were excluded as information on complete response was not available. The patients were treated on any of four clinical studies approved by the Institutional Review Board of the NHLBI (NCT00001964, NCT00061360, NCT00260689, NCT01623167) and were done under the principles of the Declaration of Helsinki. Informed consent was obtained on the respective protocols. Treatment regimens have been published:4,30-32 hATG (ATGAM, Pfizer) and cyclosporine were administered to all the patients; an investigational agent, including mycophenolate mofetil, sirolimus or eltrombopag, was added to IST in some protocols (Online Supplementary Methods and Online Supplementary Figure S1). The IST plus eltrombopag group referred to four cohorts administered eltrombopag in addition to

haematologica | 2022; 107(1)

Patients A total of 417 patients with SAA, aged 2 to 82 years, were initially treated with standard hATG and cyclosporine, with or without an investigational agent, between May 1999 and August 2019, on any of four prospective studies (Online Supplementary Figure S1): (i) a single arm study of standard IST plus mycophenolate mofetil performed from May 1999 to June 2003 (n=103);30 (ii) a direct comparison of standard IST (n=42) and IST plus sirolimus (n=35), from June 2003 to November 2005;31 (iii) a direct comparison of standard IST (n=60) and rabbit ATG-based IST (n=60; not included), from December 2005 to July 2010;32 and (iv) a single-arm study of IST plus eltrombopag (n=177), from July 2012 to August 2019, including cohort 1 (n=30), cohort 2 (n=31), cohort 3 (n=31) and an extension cohort after the publication (n=85).4 One patient who received IST plus eltrombopag in the exten127

Y. Zaimoku et al.

sion cohort did not reach the 3-month evaluation after a hematologic response and was excluded from the statistical analysis. Data on PNH clones were not available for 23 subjects: 19 in the IST plus eltrombopag group were not analyzable as their ANC were too low, at a median (IQR) of 0 (00.01)×109/L; for the other four treated with IST plus mycophenolate mofetil, PNH clone information was missing. Thrombopoietin levels were quantified in 140 of the 176 patients treated with IST plus eltrombopag; we used serum samples instead of plasma in nine patients, as levels of thrombopoietin in serum and plasma obtained at the same time were equivalent (in 15 patients with AA; correlation coefficient of 0.95; Wilcoxon signed rank test, P=0.26) (Online Supplementary Figure S2). The clinical characteristics of the patients are summarized in Table 1. The 176 patients treated with IST plus eltrombopag showed significantly higher ARC and lower platelet counts compared to the 240 patients treated with IST alone; age, sex, ANC, ALC and the prevalence of PNH clones were well balanced between the two groups. The IST plus eltrombopag group had significantly higher overall and complete response rates than did the group treated with IST alone: overall response of 82% (144 of 176) versus 62% (149 of 240, P<0.0001) and complete response of 39% (69 of 176) versus 13% (32 of 240, P<0.0001), as previously published.4 There were no significant differences in the response rates across the cohorts in the IST plus eltrombopag group and across the non-eltrombopag regimens, except for the absence of complete response in 35 patients treated with standard IST plus sirolimus (Online Supplementary Figure S1).

Correlates of overall response Patients who responded to IST plus eltrombopag, as compared to those who did not, showed significantly higher ARC (median [IQR], 21 [10-36]×109/L vs. 9 [722]×109/L, P=0.00090), higher ANC (0.33 [0.140.53]×109/L vs. 0.06 [0.01-0.25]×109/L, P=0.00027) and a tendency to higher ALC (1.29 [0.99-1.65]×109/L vs. 1.00

[0.59-1.46]×109/L, P=0.079) before therapy, similar to the group treated with IST alone (Figure 1; Table 2). Plasma thrombopoietin levels were significantly elevated in nonresponders to IST plus eltrombopag (3,000 [2,610-3,430] ng/mL vs. 2,540 [2,200-2,950] ng/mL, P=0.0037), as reported by others for IST alone.16 Age, sex, platelet counts or the presence of PNH clones did not correlate significantly with overall response to IST with or without eltrombopag. Multivariate logistic regression analysis revealed that ARC and ALC were independent predictors of overall response among all the 416 patients, consistent with the findings of a previous study;12 ARC and thrombopoietin were retained in the final model of 176 patients in the IST Table 1. Patients’ characteristics at baseline.

IST+EPAG (n=176) IST alone (n=240)

Age, years Median (IQR) 32 (19-55) Range 3-82 Sex Male 87 (49%) Female 89 (51%) Blood count, ×109/L ARC 18.8 (8.7-34.4) ANC 0.28 (0.078-0.51) ALC 1.26 (0.93-1.61) Platelets 8.0 (4.8-11.3) TPO, pg/mL* 2610 (2220-3080) PNH clone† ≥1% 60 (38%) <1% 97 (62%) Response at 6 months Overall response 144 (82%) Complete response 69 (39%)

30 (18-53) 2-82

0.47

141 (59%) 99 (41%)

0.073

14.3 (4.9-30.7) 0.30 (0.11-0.50) 1.28 (0.90-1.66) 9.0 (6.0-13) NA

0.0086 0.36 0.89 0.031

90 (38%) 146 (62%)

149 (62%) 32 (13%)

<0.0001 <0.0001

Data are n (%) or median (interquartile range) unless otherwise noted. IST: immunosuppressive therapy; EPAG: eltrombopag; IQR: interquartile range; ARC: absolute reticulocyte count; ANC: absolute neutrophil count; ALC: absolute lymphocyte count; TPO: thrombopoietin; PNH: paroxysmal nocturnal hemoglobinuria; NA: not assessed. *TPO was studied in 140 patients treated with IST+EPAG. †The PNH clone was not tested in 19 and four patients treated with IST plus EPAG and IST alone, respectively.

Figure 1. Pretreatment blood counts and overall response. Absolute reticulocyte count, absolute neutrophil count, absolute lymphocyte count and plasma thrombopoietin level at baseline were compared between patients who achieved a response and those who did not at 6 months according to whether they were treated with immunosuppressive therapy plus eltrombopag or immunosuppressive therapy alone. ARC: absolute reticulocyte count; ANC: absolute neutrophil count; ALC: absolute lymphocyte count; TPO: thrombopoietin; NR: no response; OR: overall response; IST: immunosuppressive therapy; EPAG: eltrombopag.

128

haematologica | 2022; 107(1)

Predicting response to IST plus EPAG in SAA

plus eltrombopag group (Online Supplementary Table S1). Although not statistically significant, the patients with PNH clones showed slightly higher overall response in both IST groups (odds ratio 1.32, P=0.24) (Online Supplementary Table S1), consistent with previous studies, but the favorable tendency disappeared in the multivariate model (odds ratio 1.06, P=0.87), as the presence of PNH clones correlated with higher ALC (P=0.018) and higher ARC (P=0.073). We also noted that the 19 patients who did not have enough ANC to assess PNH clones showed a significantly reduced overall response to IST plus eltrombopag at 53% (10 of 19; P=0.0019), which was analogous to the response rate in 19 patients with ANC <0.02×109/L at baseline. The response curves revealed that the reduction of

overall response to IST plus eltrombopag was observed only in patients with pre-treatment ARC lower than 10×109/L while overall response to IST alone decreased below an ARC of 30×109/L (Figure 2A). The overall response in subjects with ARC between 10-30×109/L was significantly improved in the group treated with IST plus eltrombopag, to the levels of those of patients with ARC >30×109/L (from 60% to 91%, P<0.0001) (Figure 2B). Addition of eltrombopag did not appear to significantly improve the overall response in subjects with ARC <10×109/L (63% in the IST plus eltrombopag group and 46% in the IST alone group, P=0.062) nor in those with ARC >30×109/L (90% vs. 89%, P=1), but was beneficial in terms of increasing complete response (27% vs. 8% in the groups with ARC <10×109/L, P=0.0036; 42% vs. 21% in

Table 2. Pretreatment variables and responses to immunosuppressive therapy.

Therapy Age, years

IST+EPAG IST alone Male sex IST+EPAG IST alone 9 ARC, ×10 /L IST+EPAG IST alone ANC, ×109/L IST+EPAG IST alone ALC, ×109/L IST+EPAG IST alone Plt, ×109/L IST+EPAG IST alone PNH clone ≥1% IST+EPAG IST alone TPO, ng/mL IST+EPAG

N 32 91 32 91 32 91 32 91 32 91 32 91 23 90 26

Value

32 (15-58) 32 (19-55) 17 (53%) 55 (60%) 9.2 (6.6-22.0) 9.6 (3.2-19.1) 0.06 (0.01-0.25) 0.20 (0.06-0.45) 1.00 (0.59-1.5) 1.07 (0.77-1.5) 5.5 (2.0-11.3) 8.0 (5.5-11.0) 8 (35%) 30 (33%) 3000 (2610-3430)

75 117 75 117 75 117 75 117 75 117 75 117 71 115 61

Value

31 (20-56) 28 (18-50) 35 (47%) 71 (61%) 24.5 (9.6-39.9) 22.0 (7.6-36.0) 0.36 (0.17-0.64) 0.38 (0.19-0.53) 1.22 ( 0.98-1.51) 1.32 (1.01-1.69) 8.0 (5.0-11.0) 9.0 (6.0-14.0) 28 (39%) 49 (43%) 2560 (2130-2880)

69 32 69 32 69 32 69 32 69 32 69 32 63 31 53

Value

OR vs. NR

32 (19-53) 24 (14-48) 35 (51%) 15 (47%) 18.3 (10.5-35.6) 17.8 (10.9-36.6) 0.28 (0.09-0.43) 0.32 (0.16-0.51) 1.33 (1.13-1.77) 1.51 (1.17-1.72) 8.0 (5.0-13.0) 9.5 (5.8-12.3) 24 (38%) 11 (35%) 2530 (2250-3000)

0.91 0.17 0.70 0.69 0.0009 <0.0001 0.00027 0.0019 0.079* 0.0015 0.12 0.30 0.82 0.27 0.0037

CR vs. no CR 0.88 0.15 0.88 0.18 0.49 0.086 0.60 0.63 0.065* 0.095* 0.28 0.57 1 0.84 0.60

Data are n (%) or median (interquartile range) unless otherwise noted. NR: no response; PR: partial response; CR: complete response; IST: immunosuppressive therapy; EPAG, eltrombopag; ARC: absolute reticulocyte count; ANC: absolute neutrophil count; ALC: absolute lymphocyte count; Plt: platelet count; PNH: paroxysmal nocturnal hemoglobinuria; TPO: thrombopoietin. *P<0.05 in patients aged ≥10 years.

Figure 2. Pretreatment absolute reticulocyte count and response rates to immunosuppressive therapy. (A) The overall response rates and their 90% confidence interval of individuals with an initial absolute reticulocyte count between X and (X+10) ×109/L in the group treated with immunosuppressive therapy (IST) plus eltrombopag (EPAG) and in the group treated with IST alone. (B) The overall response rates and the complete response rates were compared between the IST plus EPAG group and IST alone group according to initial absolute reticulocyte count. Overall response includes complete responses and partial responses. OR: overall response; ARC: absolute reticulocyte count; PR: partial response; CR: complete response.

haematologica | 2022; 107(1)

129

Y. Zaimoku et al.

the group with ARC >30×109/L, P=0.024) (Figure 2B). There was no such clear difference of response curve between the IST groups according to ANC or ALC (Online Supplementary Figure S3).

Prediction of complete response Pretreatment ALC was especially elevated in patients who achieved a complete response in both treatment groups, IST plus eltrombopag and IST alone (P-values, 0.065 and 0.095, compared to no complete response) (Figure 3; Table 2), which was statistically significant in all patients combined (P=0.020). Other variables, including the strong predictors of overall response, were unexpectedly not discriminatory for complete response. A large retrospective study of 312 pediatric AA patients found significantly reduced pretreatment ALC in IST responders compared to non-responders (median of 1.6×109/L vs. 2.0×109/L, respectively; P=0.006);21 and other studies in children also showed similar trends, while the correlation of higher ALC with better outcomes was mostly described in adult patients (Online Supplementary Table S2). We therefore hypothesized that the impact of ALC on IST responsiveness differed by age. We did observe a significant inverse correlation of ALC with complete response in 34 younger children aged <10 years for the two IST groups; six complete responders compared to the other 28 patients had significantly reduced ALC at a median (IQR) of 0.41 (0.31-1.25)×109/L versus 1.50 (1.072.13)×109/L (P=0.037) (Online Supplementary Figure S4A). ALC retained a significant positive correlation with complete response in adolescents aged 10-19 years (P=0.012) and in adults aged 20 years or older (P=0.029). Patient age of 10 years was the best cutoff to discriminate the inverse relationship of ALC with complete response in younger and older individuals (Online Supplementary Figure S4B-D). Therefore, exclusion of the young children aged <10 years from the statistical analyses led to a significant correlation of ALC with complete response (P=0.010 in the IST plus eltrombopag group and P=0.012 in the group treated with

IST alone) as well as with overall response (P=0.016 in the IST plus eltrombopag group and P=0.00011 in the IST alone group). Complete response tended to associate with higher ALC in all the age groups over 10 years old, but the association was especially obvious in adolescents and young adults, aged 10-40 years (Online Supplementary Figure S4A). The complete response rate to IST plus eltrombopag in the adolescents and young adults reached 60% (28 of 47) if the ALC was more than 1.3×109/L before therapy, which was significantly higher than the 32% (15 of 47, P=0.013) in the other half of patients with lower ALC; this cut-off of ALC could also stratify complete response probability in the group treated with IST alone (21% [14 of 68] vs. 3% [2 of 66]; P=0.0024). We did not observe such an age-dependent inverse correlation in other predictors of overall response (Online Supplementary Figure S5).

Rabbit antithymocyte globulin-based immunosuppressive therapy Because the correlation of ALC with complete response has not been reported, we additionally assessed this relationship in 54 patients, aged 10 years or older, treated with rabbit ATG and cyclosporine (NCT00260689).32 Four of the 54 patients achieved a complete response; they had significantly higher ALC compared to the remaining 50 patients at a median (IQR) of 1.97 (1.74-2.41)×109/L versus 1.14 (0.81-1.36)×109/L (P=0.0045) (Online Supplementary Figure S6). None of six children aged <10 years treated with rabbit ATG-based IST achieved a complete response. ARC and ANC did not correlate with complete response, but as expected they did with overall response, similar to hATG-based IST.

Discussion Eltrombopag combined with standard IST is now the best nontransplant therapy for SAA, but little has been

Figure 3. Blood counts and complete response. Pretreatment absolute reticulocyte count, absolute neutrophil count, absolute lymphocyte count and plasma thrombopoietin level of patients in the groups treated with immunosuppressive therapy (IST) plus eltrombopag (EPAG) and IST alone are shown according to responses (complete, partial and none) at 6 months. Initial absolute lymphocyte count tended to be elevated in patients who achieved complete response after IST both with and without EPAG, but absolute reticulocyte count, absolute neutrophil count and thrombopoietin, could not discriminate who would go on to have a complete response. ARC: absolute reticulocyte count; ANC: absolute neutrophil count; ALC: absolute lymphocyte count; TPO: plasma thrombopoietin; NR: no response; PR: partial response; CR: complete response.

130

haematologica | 2022; 107(1)

Predicting response to IST plus EPAG in SAA

reported concerning response predictors, partly because of the very limited number of non-responders after such highly effective therapy. Our large retrospective study in 177 SAA patients treated with IST plus eltrombopag, incorporating 85 new subjects after the initial publication,4 has now revealed that pretreatment blood counts remain strong predictors of overall response to IST plus eltrombopag, similar to their predictive value in 240 patients treated with IST alone. We also noted the relationship of ALC with complete response for the first time in both groups of patients, i.e., those treated with IST plus eltrombopag and those treated with IST alone. The overall response rate to IST plus eltrombopag exceeded 90% in patients who had an ARC of at least 10×109/L, and the complete response rate reached 60% in adolescents and young adults who had an ALC of more than 1.3×109/L prior to treatment. The addition of eltrombopag to IST markedly improved responses in patients who had an ARC in the 10-30×109/L range. In contrast, the addition of eltrombopag to IST did not significantly increase overall response in patients who had an ARC lower than 10×109/L or higher than 30×109/L, but was still beneficial as it induced significantly more complete responses compared to IST only. Multiple studies have attempted to predict response to IST in patients with SAA because HSCT is available as a competing treatment option. In practice, only the patient’s age, in addition to the availability of an appropriate donor, is generally considered to decide initial therapy.2 This algorithm is based on outcomes of patients treated with IST alone,35,36 and may need to be revised in the eltrombopag era. In the short-term, patients with residual hematopoiesis might benefit more from IST plus eltrombopag than from HSCT, but long-term follow-up especially of relapse and clonal evolution with this new combination is necessary and is currently underway. The impact of pre-transplant blood counts on the outcomes after HSCT needs to be studied as well in AA. The correlations of initial ARC and ANC with overall resposne were not surprising, as residual hematopoietic stem and progenitor cells should have an impact on hematopoietic recovery after immunosuppression, and they themselves are criteria for response. Similarly, the impact of thrombopoietin levels would be attributable to their inverse correlation with platelet production,37,38 another overall response criterion, although pretreatment platelet counts, affected by platelet transfusion, were not predictive of responses to IST. Complete response was unexpectedly not discriminated by the strong overall response predictors, including ARC, ANC and thrombopoietin levels; these may not represent the quantity or quality of hematopoietic stem cells needed for complete hematologic recovery, as blood counts of patients with AA are often supported by aberrant hematopoiesis from PNH clones,24,39 HLA class I allele lacking clones,40-42 or clones associated with agerelated somatic mutations.23 These are usually detected in myeloid and erythroid cells but are less frequent or absent in lymphocytes.43-45 We observed an association of ALC with complete response and further validated the correlation in rabbit ATG-based IST. ALC may be more representative of levels of residual hematopoietic stem cells than other blood counts. On the other hand, recent studies in HSCT indi-

haematologica | 2022; 107(1)

cated that ATG dose adjustment based on ALC, rather than body weight, is beneficial to achieve an optimal level of immunosuppression because circulating lymphocytes bind and clear ATG;46-49 a low pretreatment ALC leads to reduced clearance of ATG and prolonged immunesuppression, which is likely unfavorable for hematologic recovery, as observed in patients treated with rabbit ATG-based IST.32 The impact of ALC was age-dependent, as complete response was associated with lower ALC in young children aged <10 years, but with higher ALC in adolescents and adult patients. Our literature review supported the differential impact of pretreatment ALC in children and in adults. The difference by age is probably associated with higher normal ALC and higher lymphoproliferative capacity in young children than in adolescents and adults;50,51 our body weight-based ATG dosage might be insufficient to suppress proliferative lymphocytes of young children unless ALC is markedly reduced before treatment. A tendency to a slightly higher overall response in patients with PNH clones was attributed to the correlations of the presence of PNH clones with higher ALC and ARC, which may represent escape hematopoiesis by PNH-phenotype hematopoietic stem cells and underestimation of PNH clones in very severe disease in which neutrophils are absent and red blood cells are replaced by transfusion. A lower threshold allows detection of patients having small PNH clones but also creates more difficulties in very severe disease as many more cells are needed to identify the small populations. Care must be taken to interpret biomarker studies on prediction of response to IST, as they are also potentially influenced by the blood counts. For example, clonal hematopoiesis that preferentially involves myeloid progenitors will be underestimated in patients with low ANC; thrombopoietin levels show an inverse correlation with platelet production;37,38 and lymphocyte subsets also corelate with myelopoiesis.28 In summary, residual hematopoiesis, as indicated by baseline ARC, ANC and thrombopoietin, is predictive of initial overall response to eltrombopag plus IST, but does not correlate with complete response. We did, however, observe an age-dependent association between complete response and ALC. These findings should be helpful for future protocols and in decision-making regarding treatments for individual patients. Disclosures No conflicts of interest to disclose. Contributions YZ, PAB and NSY designed the study and wrote the first draft of the manuscript. PAB, EMG and NSY participated in the patients’ care. YZ, RS and COW analyzed the data. YZ and XF quantified thrombopoietin levels. All authors critically reviewed the draft and approved the final version for publication. Funding This investigation was supported by the Intramural Research Program of the NIH and the NHLBI. NCT01623167 is funded by Novartis and previously by GSK through a cooperative research and development agreement (CRADA). Data-sharing statement Individual patients’ information will not be disclosed.

131

Y. Zaimoku et al.

References 1. Young NS. Aplastic anemia. N Engl J Med. 2018;379(17):1643-1656. 2. Bacigalupo A. How I treat acquired aplastic anemia. Blood. 2017;129(11):1428-1436. 3. Scheinberg P, Young NS. How I treat acquired aplastic anemia. Blood. 2012;120 (6):1185-1196. 4. Townsley DM, Scheinberg P, Winkler T, et al. Eltrombopag added to standard immunosuppression for aplastic anemia. N Engl J Med. 2017;376(16):1540-1550. 5. Peffault de Latour R, Marsh J, Iacobelli S, et al. Results of the EBMT SAAWP phase III prospective randomized multicenter RACE study of horse ATG and ciclosporin with or without eltrombopag in naïve SAA patients [abstract]. In: 46th Annual Meeting of the EBMT. 2020 Aug 31-Sep 1; Virtual: 2020. Abstract O018. 6. Olnes MJ, Scheinberg P, Calvo KR, et al. Eltrombopag and improved hematopoiesis in refractory aplastic anemia. N Engl J Med. 2012;367(1):11-19. 7. Desmond R, Townsley DM, Dumitriu B, et al. Eltrombopag restores trilineage hematopoiesis in refractory severe aplastic anemia that can be sustained on discontinuation of drug. Blood. 2014;123(12):18181825. 8. Assi R, Garcia-Manero G, Ravandi F, et al. Addition of eltrombopag to immunosuppressive therapy in patients with newly diagnosed aplastic anemia. Cancer. 2018;124(21):4192-4201. 9. Lengline E, Drenou B, Peterlin P, et al. Nationwide survey on the use of eltrombopag in patients with severe aplastic anemia: a report on behalf of the French Reference Center for Aplastic Anemia. Haematologica. 2018;103(2):212-220. 10. Lee JW, Lee SE, Jung CW, et al. Romiplostim in patients with refractory aplastic anaemia previously treated with immunosuppressive therapy: a dose-finding and long-term treatment phase 2 trial. Lancet Haematol. 2019;6(11):e562-e572. 11. Ecsedi M, Lengline É, Knol-Bout C, et al. Use of eltrombopag in aplastic anemia in Europe. Ann Hematol. 2019;98(6):1341-1350. 12. Scheinberg P, Wu CO, Nunez O, Young NS. Predicting response to immunosuppressive therapy and survival in severe aplastic anaemia. Br J Haematol. 2009;144(2):206216. 13. Afable MG 2nd, Shaik M, Sugimoto Y, et al. Efficacy of rabbit anti-thymocyte globulin in severe aplastic anemia. Haematologica. 2011;96(9):1269-1275. 14. Shin SH, Yoon JH, Yahng SA, et al. The efficacy of rabbit antithymocyte globulin with cyclosporine in comparison to horse antithymocyte globulin as a first-line treatment in adult patients with severe aplastic anemia: a single-center retrospective study. Ann Hematol. 2013;92(6):817-824. 15. Narita A, Muramatsu H, Sekiya Y, et al. Paroxysmal nocturnal hemoglobinuria and telomere length predicts response to immunosuppressive therapy in pediatric aplastic anemia. Haematologica. 2015;100 (12):1546-1552. 16. Elmahdi S, Muramatsu H, Narita A, et al. Markedly high plasma thrombopoietin (TPO) level is a predictor of poor response to immunosuppressive therapy in children with acquired severe aplastic anemia. Pediatr Blood Cancer. 2016;63(4):659-664. 17. Vaht K, Goransson M, Carlson K, et al. Low response rate to ATG-based immuno-

132

suppressive therapy in very severe aplastic anaemia - a Swedish nationwide cohort study. Eur J Haematol. 2018;100(6):613620. 18. Cabannes-Hamy A, Boissel N, Peffault De Latour R, et al. The effect of age in patients with acquired aplastic anaemia treated with immunosuppressive therapy: comparison of Adolescents and Young Adults with children and older adults. Br J Haematol. 2018;183(5):766-774. 19. Chang MH, Kim KH, Kim HS, et al. Predictors of response to immunosuppressive therapy with antithymocyte globulin and cyclosporine and prognostic factors for survival in patients with severe aplastic anemia. Eur J Haematol. 2010;84(2):154-159. 20. Zhao X, Zhang L, Jing L, et al. The role of paroxysmal nocturnal hemoglobinuria clones in response to immunosuppressive therapy of patients with severe aplastic anemia. Ann Hematol. 2015;94(7):1105-1110. 21. Yoshida N, Yagasaki H, Hama A, et al. Predicting response to immunosuppressive therapy in childhood aplastic anemia. Haematologica. 2011;96(5):771-774. 22. Boddu P, Garcia-Manero G, Ravandi F, et al. Clinical outcomes in adult patients with aplastic anemia: a single institution experience. Am J Hematol. 2017;92(12):1295-1302. 23. Yoshizato T, Dumitriu B, Hosokawa K, et al. Somatic mutations and clonal hematopoiesis in aplastic anemia. N Engl J Med. 2015;373(1):35-47. 24. Sugimori C, Chuhjo T, Feng X, et al. Minor population of CD55-CD59- blood cells predicts response to immunosuppressive therapy and prognosis in patients with aplastic anemia. Blood. 2006;107(4):1308-1314. 25. Kulagin A, Lisukov I, Ivanova M, et al. Prognostic value of paroxysmal nocturnal haemoglobinuria clone presence in aplastic anaemia patients treated with combined immunosuppression: results of two-centre prospective study. Br J Haematol. 2014;164 (4):546-554. 26. Sakaguchi H, Nishio N, Hama A, et al. Peripheral blood lymphocyte telomere length as a predictor of response to immunosuppressive therapy in childhood aplastic anemia. Haematologica. 2014;99(8): 1312-1316. 27. Kordasti S, Costantini B, Seidl T, et al. Deep phenotyping of Tregs identifies an immune signature for idiopathic aplastic anemia and predicts response to treatment. Blood. 2016;128(9):1193-1205. 28. Zaimoku Y, Patel BA, Kajigaya S, et al. Deficit of CD19+ CD24hi CD38hi regulatory B cells in severe aplastic anaemia. Br J Haematol. 2020;190(4):610-617. 29. Rosenfeld S, Follmann D, Nunez O, Young NS. Antithymocyte globulin and cyclosporine for severe aplastic anemia: association between hematologic response and long-term outcome. JAMA. 2003;289 (9):1130-1135. 30. Scheinberg P, Nunez O, Wu C, Young NS. Treatment of severe aplastic anaemia with combined immunosuppression: anti-thymocyte globulin, ciclosporin and mycophenolate mofetil. Br J Haematol. 2006;133(6):606611. 31. Scheinberg P, Wu CO, Nunez O, Boss C, Sloand EM, Young NS. Treatment of severe aplastic anemia with a combination of horse antithymocyte globulin and cyclosporine, with or without sirolimus: a prospective randomized study. Haematologica. 2009;94(3):348-354. 32. Scheinberg P, Nunez O, Weinstein B,

Biancotto A, Wu CO, Young NS. Horse versus rabbit antithymocyte globulin in acquired aplastic anemia. N Engl J Med. 2011;365(5):430-438. 33. Feng X, Scheinberg P, Wu CO, et al. Cytokine signature profiles in acquired aplastic anemia and myelodysplastic syndromes. Haematologica. 2011;96(4):602606. 34. Kanda Y. Investigation of the freely available easy-to-use software 'EZR' for medical statistics. Bone Marrow Transplant. 2013;48(3): 452-458. 35. Bacigalupo A, Brand R, Oneto R, et al. Treatment of acquired severe aplastic anemia: bone marrow transplantation compared with immunosuppressive therapy-the European Group for Blood and Marrow Transplantation experience. Semin Hematol. 2000;37(1):69-80. 36. Yoshida N, Kobayashi R, Yabe H, et al. Firstline treatment for severe aplastic anemia in children: bone marrow transplantation from a matched family donor versus immunosuppressive therapy. Haematologica. 2014;99 (12):1784-1791. 37. Schrezenmeier H, Griesshammer M, Hornkohl A, et al. Thrombopoietin serum levels in patients with aplastic anaemia: correlation with platelet count and persistent elevation in remission. Br J Haematol. 1998;100(3):571-576. 38. Zhao X, Feng X, Wu Z, et al. Persistent elevation of plasma thrombopoietin levels after treatment in severe aplastic anemia. Exp Hematol. 2018;58:39-43. 39. Dunn DE, Tanawattanacharoen P, Boccuni P, et al. Paroxysmal nocturnal hemoglobinuria cells in patients with bone marrow failure syndromes. Ann Intern Med. 1999;131(6): 401-408. 40. Katagiri T, Sato-Otsubo A, Kashiwase K, et al. Frequent loss of HLA alleles associated with copy number-neutral 6pLOH in acquired aplastic anemia. Blood. 2011;118 (25):6601-6609. 41. Zaimoku Y, Takamatsu H, Hosomichi K, et al. Identification of an HLA class I allele closely involved in the autoantigen presentation in acquired aplastic anemia. Blood. 2017;129(21):2908-2916. 42. Babushok DV, Duke JL, Xie HM, et al. Somatic HLA mutations expose the role of class I-mediated autoimmunity in aplastic anemia and its clonal complications. Blood Adv. 2017;1(22):1900-1910. 43. Katagiri T, Kawamoto H, Nakakuki T, et al. Individual hematopoietic stem cells in human bone marrow of patients with aplastic anemia or myelodysplastic syndrome stably give rise to limited cell lineages. Stem Cells. 2013;31(3):536-546. 44. Maruyama H, Katagiri T, Kashiwase K, et al. Clinical significance and origin of leukocytes that lack HLA-A allele expression in patients with acquired aplastic anemia. Exp Hematol. 2016;44(10):931-939. 45. Lee-Six H, Obro NF, Shepherd MS, et al. Population dynamics of normal human blood inferred from somatic mutations. Nature. 2018;561(7724):473-478. 46. Admiraal R, van Kesteren C, Jol-van der Zijde CM, et al. Association between antithymocyte globulin exposure and CD4+ immune reconstitution in paediatric haemopoietic cell transplantation: a multicentre, retrospective pharmacodynamic cohort analysis. Lancet Haematol. 2015;2(5): e194-203. 47. Admiraal R, Nierkens S, de Witte MA, et al. Association between anti-thymocyte globu-

haematologica | 2022; 107(1)

Predicting response to IST plus EPAG in SAA

lin exposure and survival outcomes in adult unrelated haemopoietic cell transplantation: a multicentre, retrospective, pharmacodynamic cohort analysis. Lancet Haematol. 2017;4(4):e183-e191. 48. Soiffer RJ, Kim HT, McGuirk J, et al. Prospective, randomized, double-blind, phase III clinical trial of anti-T-lymphocyte globulin to assess impact on chronic graftversus-host disease-free survival in

haematologica | 2022; 107(1)

patients undergoing HLA-matched unrelated myeloablative hematopoietic cell transplantation. J Clin Oncol. 2017;35(36):40034011. 49. Kennedy VE, Chen H, Savani BN, et al. Optimizing antithymocyte globulin dosing for unrelated donor allogeneic hematopoietic cell transplantation based on recipient absolute lymphocyte count. Biol Blood Marrow Transplant. 2018;24(1):150-155.

50. Shearer WT, Rosenblatt HM, Gelman RS, et al. Lymphocyte subsets in healthy children from birth through 18 years of age: the Pediatric AIDS Clinical Trials Group P1009 study. J Allergy Clin Immunol. 2003;112(5): 973-980. 51. McCarron M, Osborne Y, Story CJ, Dempsey JL, Turner DR, Morley AA. Effect of age on lymphocyte proliferation. Mech Ageing Dev. 1987;41(3):211-218.

133

ARTICLE Ferrata Storti Foundation

Haematologica 2022 Volume 107(1):134-142

Chronic Lymphocytic Leukemia

The impact of early discontinuation/dose modification of venetoclax on outcomes in patients with relapsed/refractory chronic lymphocytic leukemia: post-hoc analyses from the phase III MURANO study Anthony R. Mato,1 Jeff P. Sharman,2 Juliana M.L. Biondo,3 Mei Wu,3 Yong Mun,3 Su Y. Kim,4 Kathryn Humphrey,5 Michelle Boyer,5 Qian Zhu3 and John F. Seymour6 Memorial Sloan Kettering Cancer Center, New York, NY, USA; 2Willamette Valley Cancer Institute and Research Center, Eugene, OR, USA/US Oncology Research, US Oncology Network, The Woodlands, TX, USA; 3Genentech, Inc., South San Francisco, CA, USA; 4 AbbVie, North Chicago, IL, USA; 5Roche Products Limited, Welwyn Garden City, UK and 6 Peter MacCallum Cancer Centre, Royal Melbourne Hospital, The University of Melbourne, Melbourne, Victoria, Australia 1

ABSTRACT

Correspondence: ANTHONY R. MATO matoa@mskcc.org Received: July 10, 2020. Accepted: November 27, 2020. Pre-published: December 17, 2020. https://doi.org/10.3324/haematol.2020.266486

134

ixed-duration venetoclax plus rituximab (VenR) has a manageable safety profile and improves survival in patients with relapsed/refractory (R/R) chronic lymphocytic leukemia (CLL). We present data from the phase III MURANO study on the impact of venetoclax modification or premature discontinuation on outcomes in patients with R/R CLL. Timedependent Cox proportional hazards regression models, stratified by 17p deletion and risk status, evaluated the impact of venetoclax discontinuation/modification on investigator-assessed progression-free survival (PFS) and overall survival (OS). Analyses were performed retrospectively (without type-1 error control) in intention-to-treat patients from the VenR arm of MURANO. Overall, 140 of 194 (72%) patients in the VenR arm completed 2 years of therapy; 54 of 194 (28%) patients prematurely discontinued treatment. Inferior PFS was observed in patients prematurely discontinuing venetoclax for any reason (disease progression excluded; P<0.0001) and specifically in patients discontinuing due to adverse event (AE) (P<0.0001), versus those who did not discontinue early. Risk of a PFS/OS event was significantly reduced by each extra month (exposure cycle) of venetoclax therapy (P=0.0263 for PFS; P<0.0001 for OS). Treatment interruption for AE occurred in 134 of 194 (69%) patients, most commonly due to neutropenia (84 of 194; 43%), per protocol requirements. Treatment interruption had no impact on PFS or OS, regardless of duration. Dose reductions were required by 45 of 194 (23%) patients, but had no significant impact on outcomes. In MURANO, premature discontinuation was associated with suboptimal outcomes; venetoclax treatment modification was not. These data highlight the importance of effective toxicity control to realize the full benefit of venetoclax treatment (clinicaltrials gov. Identifier: NCT02005471).

Introduction Chronic lymphocytic leukemia (CLL), the most common adult leukemia in Western countries, accounts for approximately one-third of new adult leukemia diagnoses in the United States.1,2 Although no curative treatment for CLL exists (apart from allogeneic stem cell transplantation, which is unsuitable for most patients),3 chemoimmunotherapy achieves disease control and prolongs survival, and has historically been the standard treatment. More recently, novel targeted therapies have improved outcomes over chemoimmunotherapy in first-line (1L) and relapsed/refractory (R/R) CLL.4 These comprise selective kinase inhibitors targeting the B-cell receptor and its downstream proteins (including the Bruton tyro-

haematologica | 2022; 107(1)

Impact of discontinuation/modifications on outcome

sine kinase [BTK] inhibitors ibrutinib and acalabrutinib, and the phosphatidylinositol 3-kinase [PI3K] inhibitors idelalisib and duvelisib) and the potent B-cell lymphoma2 (BCL-2) inhibitor venetoclax.5 Mechanisms of action differ between targeted therapies. BTK inhibitors reduce CLL cell chemotaxis toward the chemokines CXCL12 and CXCL13 and block B-cell receptor signaling.6 This drives apoptosis and/or disrupts cell migration and adherence to protective tumor microenvironments. PI3K inhibitors impede one or more of the PI3K enzymes, part of the PI3K/AKT/mTOR pathway that regulates many cellular functions (e.g., growth control, metabolism and translation initiation), resulting in tumor suppression.7 Venetoclax is a BH3-mimetic, which blocks the anti-apoptotic BCL-2 protein, leading to programmed cell death of CLL cells.8 Ibrutinib, idelalisib, duvelisib and acalabrutinib are used until progression or toxicity, while venetoclax provides a chemotherapy-free treatment option that allows for a finite duration of therapy.9 Venetoclax has a fixed treatment duration of 1 year when given concomitantly with obinutuzumab for 1L CLL, 2 years when given in combination with rituximab for R/R CLL, and until disease progression (PD) or toxicity as a monotherapy for R/R CLL.10-13 Although targeted therapies have manageable safety profiles and in some instances improve life expectancy for patients with CLL, both premature discontinuations (before PD) and modifications to therapy are reported in a proportion of cases. To date, information on whether premature discontinuations and/or treatment modifications have an impact on survival outcomes in patients with CLL (in both real-world studies and clinical trials) has not been summarized collectively. However, numerous publications address this question, with significant literature available on the impact of disruption of ibrutinib on survival (Online Supplemental Table S1).14-28 In brief, the literature suggests that ibrutinib discontinuation is associated with poor survival outcomes,16,19,21 while the impact of interruption on outcomes seems contradictory (with some indicating an effect on survival and others suggesting a lack of difference in outcomes).14,15,18-20,22,23 Limited data are available for venetoclax; and given the different mechanism of action, more rapid kinetics of response and deeper remissions with venetoclax (especially when used with anti-CD20 antibodies), extrapolation of ibrutinib data is unlikely to be valid for venetoclax. The MURANO phase III, international, randomized, open-label study evaluated the efficacy of six cycles of venetoclax in combination with rituximab (VenR) then venetoclax once daily for 2 years in total compared with six cycles of bendamustine plus rituximab (BR) in patients with R/R CLL.29 VenR was superior to BR; after a median follow-up period of 23.8 months, investigator-assessed progression-free survival (PFS) was significantly longer in the VenR group (32 events in 194 patients) than in the BR group (114 events in 195 patients).29 Additionally, PFS benefit of VenR over BR was sustained (hazard ratio [HR], 0.19 [95% Confidence Interval [CI]: 0.14–0.25]; P<0.0001), with 4-year PFS estimates of 57.3% (95% CI: 49.4–65.3) versus 4.6% (95% CI: 0.1–9.2), respectively.30 While the minimum duration of venetoclax necessary to achieve therapeutic benefit remains unclear, the available data suggest that no substantial further deepening of responses occurs beyond 24 months of therapy.31 haematologica | 2022; 107(1)

This manuscript reports novel venetoclax data from post-hoc analyses of MURANO to assess the impact of discontinuation/modification on outcomes and provides the first review of the current published literature (for all oral targeted agents approved for use in CLL as of August 2019: ibrutinib, idelalisib and duvelisib; see the Online Supplemental Appendix), to contextualize these data.

Methods MURANO study methodology Analyses were performed on data from the VenR cohort of patients in the MURANO study (NCT02005471), which has been described previously.29 MURANO enrolled patients aged ≥18 years, who had received 1–3 prior lines of therapy (including at least one chemotherapycontaining regimen), had an Eastern Cooperative Oncology Group performance status of 0 or 1, and had adequate bone marrow, renal, and hepatic function. Patients received venetoclax over a 5week dose ramp-up period (to reach the 400 mg daily target dose) followed by six cycles of rituximab (single infusion on day 1 of each 28-day cycle: 375 mg/m2 on day 1, cycle 1 and 500 mg/m2 for day 1, cycles 2–6) with ongoing daily dosing of venetoclax. Patients continued daily venetoclax therapy for a maximum of 2 years (from day 1, cycle 1). All relevant ethical committees approved the MURANO study.

Data and statistical analyses Efficacy analyses were based on the intention-to-treat population (all patients who underwent randomization). After trial drug initiation, all adverse events (AE), including those resulting in discontinuation or treatment modification (including treatment interruption and dose reduction), were reported up to 28 days after the last dose of trial drug (maximum of 2 years) or up to 90 days after the last dose of rituximab, whichever was longer. Discontinuation and modification events were recorded (see the Online Supplementary Appendix). Primary endpoints of the post-hoc analyses included investigator-assessed PFS and overall survival (OS). PFS was defined as time from randomization to first occurrence of PD, relapse or death. In order to determine the effect of treatment discontinuations and modifications on clinical outcomes, eight time-dependent Cox proportional hazards regression models stratified by chromosome 17p deletion (del(17p)) and disease risk status (high or low risk; see the Online Supplementary Appendix) were used to evaluate investigator-assessed PFS and OS (Table 1). Time-fixed baseline covariates and time-dependent serious AE (SAE) indicators were considered in each model. Time-dependent analyses were performed to account for varying time when treatment discontinuation/modification occurred, helping to assure that longer PFS and OS were not falsely attributed to patients who simply had treatment discontinuation/modification occurring at the later time (i.e., immortal-time bias). The covariate selection process for the final multivariate model consisted of a bivariate or trivariate time-dependent Cox proportional-hazards regression model stratified by del(17p) and risk status with independent time-dependent variable (variables) and the given time-fixed covariate. Covariates were included in the final multivariate model if the P-value of the given time-fixed covariate was <0.1. The final multivariate time-dependent Cox proportional-hazards regression model was then stratified by del(17p) and risk status and included time-dependent covariates in addition to the significant time-fixed covariates (P<0.1; Table 1). The complete-case method was used in the analyses, which

135

A.R. Mato et al.

included patients without missing values for all covariates used in the final multivariate model. All analyses performed were posthoc and are therefore considered hypothesis-generating, with no statistical adjustment for type-I error. A P-value of <0.05 was considered significant except for in the covariate selection process described above. Data cut-off for this analysis was May 8, 2019.

Results In total, 194 patients were included in the ITT population for the VenR arm of the MURANO study, of whom 140 completed 2 years of venetoclax therapy. Patient demographics and baseline characteristics have been reported previously.29 Half of the patients were aged <65 years (50.0%), most patients were male (70.1%), white (93.3%) and had Rai stage 0–II at enrollment (71.6%). In total, 46 of 173 patients (26.6%) who were assessed for del(17p) status had this deletion, 48 of 192 patients (25.0%) who were tested for TP53 mutation status had TP53 mutations, and 123 of 180 patients (68.3%) who were tested

for immunoglobulin heavy chain variable region (IGVH) mutational status had unmutated IGVH.

Treatment discontinuation Early discontinuation of venetoclax (<2 years) was reported in 27.8% of patients (54 of 194). The two most common reasons for discontinuation was AE (29 of 54, 53.7%), followed by PD (12 of 54, 22.2%). Two of the remaining 13 patients died, and the other 11 discontinued due to withdrawal by patient (n=5), physician decision (n=3), other reasons (n=2) and non-compliance (n=1). In total, 22 of 28 (79%) deaths reported for the MURANO study were in patients who discontinued venetoclax prematurely. Of those, 13 of 22 (59.1%) deaths were in patients who discontinued venetoclax due to AE (two were related to venetoclax), while four of 22 (18.2%) deaths were in patients who discontinued due to PD (Table 2). Two of the remaining five of 22 (22.7%) deaths were in patients who discontinued due to death, and three were in patients who discontinued due to withdrawal by patient (n=1), physician decision (n=1; related to venetoclax) and other reasons (n=1; Table 2).

Table 1. MURANO: stratified time-dependent Cox proportional-hazards regression models used to assess clinical outcomes.

Time-dependent indicators

Clinical outcome

Model 1

Discontinuation due to any reason excluding patients who discontinued venetoclax due to PD

PFS

Model 2

Discontinuation due to AE Discontinuation due to PD/other

PFS

Discontinuation due to venetoclax-related AEs

PFS

Venetoclax interruption of ≥8 consecutive days

PFS

Model 3

OS Model 4

Venetoclax interruption of ≥14 consecutive days

PFS OS

Model 5

Venetoclax interruption of ≥21 consecutive days

PFS OS

Model 6

Venetoclax interruption of ≥1 consecutive days

PFS OS

Model 7

Model 8

Cumulative treatment (cycle) exposure excluding patients who discontinued venetoclax due to PD

PFS

Time-dependent venetoclax dose reduction indicator (Yes/No)

PFS

Covariates included in final multivariate model TP53 mutation status, number of prior regimens, refractoriness to most recent prior therapy, hepatic impairment status and SAE time-dependent indicator Age, TP53 mutation status, IGVH mutation status, number of prior regimens, refractoriness to most recent prior therapy, hepatic impairment status and SAE time-dependent indicator TP53 mutation status, IGVH mutation status, refractoriness to most recent prior therapy and SAE time-dependent indicator TP53 mutation status, IGVH mutation status, bulky disease, refractoriness to most recent prior therapy and SAE time-dependent indicator TP53 mutation status, number of prior regimens and refractoriness to most recent prior therapy TP53 mutation status, IGVH mutation status, bulky disease, refractoriness to most recent prior therapy and SAE time-dependent indicator TP53 mutation status, number of prior regimen and refractoriness to most recent prior therapy TP53 mutation status, IGVH mutation status, bulky disease, refractoriness to most recent prior therapy and SAE time-dependent indicator TP53 mutation status, number of prior regimen and refractoriness to most recent prior therapy TP53 mutation status, IGVH mutation status, bulky disease, refractoriness to most recent prior therapy and SAE time-dependent indicator TP53 mutation status, number of prior regimen and refractoriness to most recent prior therapy TP53 mutation status, IGVH mutation status and SAE time-dependent indicator Chromosome 11q deletion, TP53 mutation status, Rai stage, number of prior regimen and refractoriness to most recent prior therapy TP53 mutation status, IGVH mutation status, bulky disease, refractoriness to most recent prior therapy and SAE time-dependent indicator TP53 mutation status, number of prior regimens and refractoriness to most recent prior therapy

AE: adverse event; IGVH: immunoglobulin heavy chain gene; OS: overall survival; PD: progressive disease; PFS: progression-free survival; SAE: serious adverse event.

136

haematologica | 2022; 107(1)

Impact of discontinuation/modifications on outcome

Median PFS and median OS for early discontinuation due to any reason except PD was 24.3 months (95% CI: 20.8–31.9) and not reached (NR; 95% CI: 24.3–NR), respectively, compared with 52.3 (95% CI: 47.9–NR) and NR for all patients in the VenR arm and NR (95% CI: 52.3–NR) and NR in patients who completed venetoclax treatment. Discontinuing treatment early (for any reason except PD) was significantly associated with shorter PFS (n=181; HR 5.98, 95% CI: 3.31–10.82; P<0.0001). A similar adverse impact on PFS was observed for patients who discontinued treatment due to an AE (n=174; HR 5.82, 95% CI: 2.93–11.57; P<0.0001); this inferior survival was consistent with treatment discontinuation due to venetoclax-related AE (n=174; HR 2.34, 95% CI: 1.10–4.98; P=0.0272). For OS, modeling was insufficiently powered, with too few deaths (six of 140, 4.3%) in the group of patients completing 2 years of venetoclax therapy, hence it was not performed. Overall, the median duration of venetoclax therapy for patients who discontinued venetoclax early (n=54) was 11.3 months (range, 0.1–24.9; Table 3), compared with 24.4 months (range, 0.1–27.9) for all patients. In patients discontinuing due to AE, median treatment duration was 11.3 months (range, 0.5–23.3), compared with 16.4 months (range, 4.4–24.9) for patients discontinuing for PD (P=0.063, Table 3). The majority of discontinuations occurred after completion of the combination therapy (VenR) phase, with 15 patients discontinuing during the combination therapy period (cycles 1–6, 0 to <6 months; one patient was determined as discontinued during postcombination/follow-up but is included in the 0 to <6 months group in Table 3), nine of these due to AE.

Table 2. MURANO: cause of death for patients who discontinued treatment.

Cause of venetoclax Cause of death discontinuation Adverse event

Patients who discontinued venetoclax and died (n=22)

Acute respiratory failure Cardiac failure Colorectal cancer Respiratory failure Sepsis secondary to transformed acute myeloid leukemia Metastatic malignant melanoma Myelodysplastic syndrome (related to venetoclax) Myocardial infarction (related to venetoclax) Pancreatic carcinoma Pneumonia Status epilepticus Sudden death Progressive disease Progressive disease Pneumonia Death Pneumonia Sudden cardiac death Withdrawal by patient Progressive disease Physician decision Sepsis (related to venetoclax) Other Not reported (patient had gastric cancer and was transferred to palliative care) haematologica | 2022; 107(1)

1 (4.5%) 1 (4.5%) 2 (9.1%) 1 (4.5%) 1 (4.5%) 1 (4.5%) 1 (4.5%) 1 (4.5%) 1 (4.5%) 1 (4.5%) 1 (4.5%) 1 (4.5%) 3 (13.6) 1 (4.5%) 1 (4.5%) 1 (4.5%) 1 (4.5%) 1 (4.5%) 1 (4.5%)

Discontinuation due to AE was more common earlier (34.5% [ten of 29] between 0 and <6 months vs. 20.7% [six of 29] for ≥18 months), with a visible trend for fewer discontinuations due to AE as time progressed (Table 3). Overall, 41 AE led to treatment discontinuation in 32 patients (16.5%; Table 4); however, the primary reason for early venetoclax discontinuation was only determined as AE for 29 patients (primary reasons for the other three patients were physician decision [n=2] and death [n=1]). Notably, 22 of 29 (75.9%) patients had a dose reduction or interrupted treatment before prematurely discontinuing venetoclax due to an AE and 18 of 29 (62.1%) patients discontinued treatment early due to venetoclax related AE (Online Supplementary Table S2). The most common classes of AE (MedDRA preferred terms) leading to discontinuation were blood and lymphatic system disorders (7.2%); neoplasms benign, malignant and unspecified (2.6%); general disorders and administration site conditions (2.1%); and infections and infestations (2.1%). The most common individual AE resulting in treatment discontinuation were neutropenia (six of 194; 3.1%) and thrombocytopenia (five of 194; 2.6%). Objective response rate (ORR: complete response [CR], CR with incomplete marrow recovery, partial response [PR], and nodular PR) for patients who prematurely discontinued venetoclax due to AE was 79.3% (23 of 29; Table 4). Best minimal residual disease (MRD) response in peripheral blood (measured by allele-specific oligonucleotide polymerase chain reaction) was undetectable MRD (uMRD; where the threshold for MRD was defined as one tumor cell per 104 white cells) for 62.1% of these patients (18 of 29), while the uMRD rate at the end of combination treatment response visit was 48.3% (14 of 29; Table 5). Patients who had an uMRD status on or before venetoclax discontinuation and did not have PD before venetoclax discontinuation (n=24), had a median PFS calculated from cessation of 26.6 (95% CI: 4.7–32.4) months. There were no robustly significant baseline demographic predictors for discontinuation due to AE; baseline covariates were not significantly associated with discontinuation due to venetoclax related AE, but duration of response to most recent prior therapy was marginally significant (P=0.0467) for treatment discontinuation due to non-related AE (Online Supplementary Table S3).

Table 3. MURANO: discontinuation of venetoclax treatment for all patients, due to adverse events and due to progressive disease.

Patients who discontinued venetoclax For any reason Due to AE Due to PD (n=54) (n=29) (n=12) Duration of Ven treatment, months Mean (SD) Median (range) Ven discontinuation at 0 to <6 months*, n (%) 6 to <12 months, n (%) 12 to <18 months, n (%) 18 to <24 months, n (%) ≥24 months

11.7 (8.0) 11.3 (0.1-24.9)

10.6 (7.1) 11.3 (0.5-23.3)

15.4 (7.8) 16.4 (4.4-24.9)

16 (29.6) 13 (24.1) 9 (16.7) 10 (18.5) 6 (11.1)

10 (34.5) 7 (24.1) 6 (20.7) 5 (17.2) 1 (3.4)

1 (8.3) 4 (33.3) 1 (8.3) 3 (25.0) 3 (25.0)

*One patient was determined as discontinued during post-combination/follow-up, after discontinuing Ven 33 days after rituximab discontinuation (which occurred on cycle 2 day 1); total Ven treatment duration was 98 days (63 days after rituximab was dosed). AE,: adverse event; PD: progressive disease; SD: standard deviation; Ven: venetoclax.

137

A.R. Mato et al.

Treatment interruption Of the 194 patients receiving VenR, 137 (70.6%) required a total of 316 interruptions to venetoclax treatment. Most patients interrupted treatment multiple times, with only 55 of 137 (40.1%) patients interrupting treatment just once. Of the other 82 of 137 patients, 37 (27.0%) interrupted treatment twice, 20 (14.6%) three times, 12 (8.8%) four times and 13 five or more times (n=6 for five interruptions [4.4%], n=2 for six interruptions [1.5%], n=3 for seven Table 4. MURANO: adverse events leading to discontinuation, dose interruption or dose reduction of venetoclax treatment.

Venetoclax treatment (n=194) Discontinuation Interruption Dose reduction Number of patients with ≥1 AE, n (%) 32 (16.5) Total number of events, n 41 Blood and lymphatic system disorders, n (%) Total number of patients with ≥1 AE 14 (7.2) Total number of events 16 Neutropenia 6 (3.1) Thrombocytopenia 5 (2.6) Autoimmune hemolytic anemia* 2 (1.0) Anemia 1 (0.5) Febrile neutropenia 1 (0.5) Immune thrombocytopenic purpura 1 (0.5) Neoplasms benign, malignant and unspecified,† n (%) Total number of patients with ≥1 AE 5 (2.6) Total number of events 5 Colorectal cancer 2 (1.0) Metastasis 1 (0.5) Metastatic malignant melanoma 1 (0.5) Pancreatic carcinoma 1 (0.5) General disorders and administration site conditions, n (%) Total number of patients with ≥1 AE 4 (2.1) Total number of events 4 Asthenia 1 (0.5) Pyrexia 1 (0.5) Sudden cardiac death* 1 (0.5) Sudden death 1 (0.5) Infections and infestations, n (%) Total number of patients with ≥1 AE 4 (2.1) Total number of events 5 Pneumonia* 2 (1.0) Appendicitis 1 (0.5) Lung infection 1 (0.5) Peritoneal tuberculosis 1 (0.5) Upper respiratory tract infection 0 Gastrointestinal disorders, n (%) Total number of patients with ≥1 AE 3 (1.5) Total number of events 3 Ascites 1 (0.5) Crohn’s disease 1 (0.5) Diarrhea 1 (0.5) Abdominal discomfort 0 Small intestinal obstruction 0 Investigations, n (%) Total number of patients with ≥1 AE 2 (1.0) Total number of events 2 Alanine aminotransferase increased 1 (0.5) Neutrophil count decreased 1 (0.5) Aspartate aminotransferase increased 0 138

134 (69.1) 344

28 (14.4) 37

interruptions [2.2%] and n=2 for eight interruptions [1.5%]). Reason for treatment interruption was not captured specifically in the MURANO study, unless required for management of AE. Most of the interruptions (212 of 316) occurred during the combination treatment phase, with 117 of 194 patients (60.3%) interrupting treatment during combination therapy (212 interruptions) versus 62 of 194 patients (32.0%) with interruptions during the monotherapy treatment phase (104 interruptions). The median consecutive days duration of treatment interruption was 9 days (range, 1–93). Of the 137 patients with treatment interruptions, 61 (44.5%) had interruptions lasting 1–7 consecutive days, 30 (21.9%) had interruptions lasting 8–14 consecutive days, 13 (9.5%) had interruptions lasting 15–21 consecutive days, 21 (15.3%) had interruptions lasting 22–28 consecutive days and 12 (8.8%) had interruptions lasting >28 consecutive days (based on the continued from previous column

95 (49.0) 180 84 (43.3) 9 (4.6) 2 (1.0) 3 (1.5) 3 (1.5) 0

19 (9.8) 25 16 (8.2) 0 0 1 (0.5) 2 (1.0) 0

1 (0.5) 1 0 0 1 (0.5) 0

0 0 0 0 0 0

8 (4.1) 8 1 (0.5) 3 (1.5) 0 0

0 0 0 0 0 0

47 (24.2) 70 8 (4.1) 1 (0.5) 1 (0.5) 0 8 (4.1)

1 (0.5) 1 0 0 0 0 1 (0.5)

16 (8.2) 29 0 0 9 (4.6) 0 0

4 (2.1) 4 0 0 2 (1.0) 1 (0.5) 1 (0.5)

15 (7.7) 25 3 (1.5) 5 (2.6) 0

2 (1.0) 3 1 (0.5) 1 (0.5) 1 (0.5)

Respiratory, thoracic and mediastinal disorders, n (%) Total number of patients with ≥1 AE 2 (1.0) Total number of events 2 Acute respiratory failure 1 (0.5) Hydrothorax 1 (0.5) Ear and labyrinth disorders, n (%) Total number of patients with ≥1 AE 1 (0.5) Total number of events 1 Vertigo 1 (0.5) Nervous system disorders, n (%) Total number of patients with ≥1 AE 1 (0.5) Total number of events 1 Status epilepticus 1 (0.5) Metabolism and nutrition disorders, n (%) Total number of patients with ≥1 AE 0 Total number of events 0 Musculoskeletal and connective tissue disorders, n (%) Total number of patients with ≥1 AE 0 Total number of events 0 Skin and subcutaneous tissue disorders, n (%) Total number of patients with ≥1 AE 0 Total number of events 0 Cardiac disorders, n (%) Total number of patients with ≥1 AE 0 Total number of events 0 Hepatobiliary disorders, n (%) Total number of patients with ≥1 AE 0 Total number of events 0 Injury, poisoning and procedural complications, n (%) Total number of patients with ≥1 AE 0 Total number of events 0 Vascular disorders, n (%) Total number of patients with ≥1 AE 0 Total number of events 0 Reproductive system and breast disorders, n (%) Total number of patients with ≥1 AE 0 Total number of events 0

2 (1.0) 2 0 0

0 0 0 0

0 0 0

2 (1.0) 2 0

0 0 0

10 (5.2) 11

1 (0.5) 2

3 (1.5) 4

0 0

3 (1.5) 3

1 (0.5) 1

2 (1.0) 2

0 0

2 (1.0) 2

1 (0.5) 1

2 (1.0) 2

0 0

2 (1.0) 2

0 0

1 (0.5) 1

0 0

*Includes event for patient whose primary reason for discontinuation was not due to an AE (primary reason for discontinuation for the patient with autoimmune hemolytic anemia was physician decision, for the patient with sudden cardiac death was death and for the patient with pneumonia was physician decision); †Including cysts and polyps. AE: adverse event.

haematologica | 2022; 107(1)

Impact of discontinuation/modifications on outcome

longest consecutive number of days per person). Treatment interruption, regardless of duration (≥1, ≥8, ≥14 and ≥21 consecutive days of missed venetoclax treatment) had no statistically significant effect on clinical outcomes (Table 6) compared with no treatment interruption. Thirty-six (18.6%) patients who had treatment interruptions later discontinued venetoclax prior to completion of therapy. Previous treatment interruptions did not predict likelihood of prematurely discontinuing venetoclax; discontinuation rate was 26.3% in patients who previously interrupted treatment versus 31.6% in patients who did not interrupt treatment (statistically similar; P=0.4842). Overall, 134 patients (69.1%) required treatment interruption for AE (a total of 344 AE). The most common types of AE leading to treatment interruption were blood and lymphatic system disorders (49.0%), infections and infestations (24.2%), gastrointestinal disorders (8.2%) and investigations (7.7%; Table 4). The most common individual AE resulting in treatment interruption were neutropenia (84 of 194; 43.3%), thrombocytopenia (nine of 194; 4.6%) and diarrhea (nine of 194; 4.6%). Notably, 116 of 134 (86.6%) patients interrupted treatment due to venetoclax related AE (Online Supplementary Table S2). Number of prior regimens was the only variable found to be statistically significant (P=0.0048) from the logistic regression model for interruption due to venetoclax related AE (Online Supplementary Table S4). Baseline covariates were not significantly associated with treatment interruption due to AE that were not related to venetoclax (Online Supplementary Table S4). Median time to first onset of AE leading to venetoclax dose interruption was 1.6 months (range, 0.0–22.8).

Dose reduction Of the 194 patients receiving VenR, 45 (23.2%) required dose reductions for venetoclax. A total of 76 dose reductions were required, with some patients having multiple dose reductions (for different reasons); 64 due to AE (in 40 patients), seven due to unknown reasons (six patients), two due to medication error (two patients), two due to patient choice (one patient) and one due to a medical decision (one patient). Venetoclax dose was reduced to 300 mg for three of 76 (3.9%) dose reductions (n=3), 200 mg for 44 of 76 (57.9%) dose reductions (n=36), 100 mg for 23 of 76 (30.3%) dose reductions (n=20), 50 mg for two of 76 (2.6%) dose reductions (n=2) and 20 mg for four of 76 (5.3%) dose reductions (n=4). Forty-five dose reductions (in 30 patients) occurred during the combination treatment phase, compared with 31 dose reductions (in 19 patients) during the monotherapy treatment phase.

Dose reduction had no statistically significant effect on PFS (n=165; HR 1.19, 95% CI: 0.65–2.18; P=0.5765) or OS (n=192; HR 0.91, 95% CI: 0.35–2.38;P=0.8519) compared with patients with no dose reduction. Following the first dose reduction, 26 (57.8 %) patients returned to the original dose; median duration from the first dose reduction before returning to the original dose was 16.5 days (range, 1–293). Of the 19 patients (42.2%) who never returned to the original dose, one patient subsequently discontinued venetoclax, eight interrupted venetoclax treatment, three interrupted and then discontinued venetoclax treatment and seven remained on the reduced dose (200 mg for six patients [85.7%] and 100 mg for one patient [14.3%]) until treatment completion. Although the reason for dose reduction was AE for 40 patients, information on type of AE and onset of AE was only available for 28 patients (37 AE). Dose reductions in 25 of 28 (89.3%) patients were due to venetoclax related AE (Online Supplementary Table S2). The most common types of AE and the most common individual AE resulting in dose reduction (neutropenia [16 of 194; 8.2%], febrile neutropenia [two of 194; 1.0%] and diarrhea [two of 194; Table 5. MURANO: best overall response and minimal residual disease status (ASO-PCR – peripheral blood) for patients who discontinued venetoclax due to adverse events.

Patients who discontinued due to AE (n=29) Best overall response (investigator-assessed), n (%) Objective response rate Complete response Complete response with incomplete bone marrow recovery Partial response Nodular partial response Stable disease Progressive disease Missing Best MRD response, n (%) Undetectable Positive Missing MRD response at end of combination treatment response visit,* n (%) Undetectable Positive Missing

23 (79.3) 3 (10.3) 2 (6.9) 17 (58.6) 1 (3.4) 2 (6.9) 1 (3.4) 3 (10.3) 18 (62.1) 3 (10.3) 8 (27.6) 14 (48.3) 6 (20.7) 9 (31.0)

*At the 9-month time point. AE: adverse event; ASO-PCR: allele-specific oligonucleotide polymerase chain reaction; MRD: minimal residual disease.

Table 6. MURANO: impact of interruption of venetoclax treatment versus no interruption on outcomes for all patients.

≥1 days Patients, n Progression-free survival Events, n (%) HR (95% CI) P-value Overall survival Events, n (%) HR (95% CI) P-value

Duration of treatment interruption (n=194 patients) ≥8 days ≥14 days

≥21 days

137 (70.6%)

76 (39.2%)

50 (25.8%)

34 (17.5% )

49 (35.8) 0.67 (0.38-1.19) 0.1709

29 (38.2) 1.01 (0.59-1.71) 0.9741

20 (40.0) 0.92 (0.51-1.65) 0.7671

13 (38.2) 0.82 (0.41-1.65) 0.5753

17 (12.4) 0.97 (0.43-2.21) 0.9474

11 (14.5) 1.35 (0.60-3.02) 0.4646

8 (16.0) 1.47 (0.63-3.45) 0.3730

5 (14.7) 1.31 (0.46-3.73) 0.6193

CI: confidence interval; HR: hazard ratio.

haematologica | 2022; 107(1)

139

A.R. Mato et al.

1.0%]) are presented in Table 4. There were no significant baseline demographic predictors for dose reduction due to venetoclax related AE (Online Supplementary Table S4) and logistic regression modeling was not performed for dose reduction due to AE that were not related to venetoclax, due to small patient numbers (n=3). Median time to first onset of AE leading to venetoclax dose reduction was 3.99 months (range, 0.3–20.5).

Relative dose intensity Median relative dose intensity (total dose received by patients ÷ expected total target dose, starting from first day of VenR combination therapy until the last day of venetoclax treatment or clinical cut-off, whichever is earlier) was 97.4% (range, 26.4–100.0; n=189). Relative dose intensity was classified into four categories; distribution of PFS were similar among these categories (Table 7), indicating that there was no compounded effect of both dose reduction and interruption. This analyses was not performed for OS as modeling was insufficiently powered, with only six deaths among the patients who completed venetoclax treatment.

Cumulative exposure Cumulative treatment (number of 28-day cycles) received by patients (excluding those discontinuing for PD) was statistically significantly associated with reduced risk of both PFS and OS events (PFS: n=167, HR 0.93, 95% CI: 0.88–0.99, P=0.0263; OS: n=161, HR 0.85, 95% CI: 0.78– 0.92, P<0.0001).

Discussion The results of these analyses highlight the importance of appropriately managing treatment modifications to ensure optimal outcomes for patients receiving targeted treatment for CLL. Treatment discontinuation prior to completing the MURANO regimen (fixed duration of 2 years of venetoclax) was associated with impaired PFS, while improved PFS and OS were associated with greater cumulative venetoclax treatment exposure. PFS was also statistically significantly lower with discontinuation specifically due to AE. The lack of a significant effect of treatment interruption and dose reduction on outcome highlights the importance of management of AE to allow alternative options such as dose modification or re-initiation of therapy following a treatment break, to maximize the duration of therapy. Early discontinuation of venetoclax was reported in 28% of patients participating in the MURANO study,

similar to the 35% discontinuation rate (median followup of 10 months) reported in a real-world study of venetoclax treatment.32 It is not clearly indicated whether discontinuation in that study was premature discontinuation (as in the MURANO study) or whether some of these patients were receiving venetoclax monotherapy, where treatment was until progression. The most common reason for discontinuation in the real-world study was PD (17%), with 4% reporting discontinuation due to toxicity.32 In contrast, the MURANO data indicate that most patients who discontinue venetoclax prematurely do so due to AE (14.9%), with PD being the second most common cause of discontinuation (6.2%). This emphasizes the difficulty of assessing the impact of treatment discontinuation due to toxicity or PD based on clinical studies and real-world data, especially given that there are currently insufficient real-world data to assess how venetoclax discontinuation due to toxicity affects outcomes.32 Furthermore, differences in data reported for the MURANO study compared with real-world data reflects the limitations of cross-comparing data without accounting for differences in patient population and treatment schedule. In the MURANO study, treatment interruption occurred in 137 of 194 (70.6%) of patients. Venetoclax treatment interruption (for ≥1, ≥8, ≥14 and ≥21 consecutive days) showed no statistically significant effect on clinical outcomes (PFS and OS). The rate of interruptions occurring in MURANO is higher than the rate reported in two real-world studies of venetoclax (41% [including patients with dose reductions] and 32%, respectively).33,34 Two explanations potentially account for this difference: firstly, the more stringent reporting of interruptions within a clinical trial; and secondly, the stringent protocol requirements for mild asymptomatic neutropenia mandated for the MURANO study (see the Online Supplementary Appendix), which might contrast with realworld management, where treatment is typically continued with the addition of granulocyte colony-stimulating factor support. Real-world studies of venetoclax provide further evidence that temporary interruption of treatment (length of interruption not disclosed for one study; median 7 days [range, 1–132] for the other) does not have an impact on PFS.33,34 Dose reduction occurred in 45 of 194 (23.2%) of patients in the MURANO study and had no statistically significant effect on PFS or OS. The number of patients requiring dose reductions in MURANO is similar to the 29% reported in the Roeker et al. real-world study of venetoclax.30 Furthermore, the real-world studies of venetoclax support the findings from MURANO that dose

Table 7. MURANO: landmark analysis of progression-free surviva by venetoclax relative dose intensity quartiles.

Events, n (%) Kaplan–Meier median, months (95% CI) HR (95% CI) P-value

Min (26.4%) – <Q1 (93.6%) n=33

Q1 (93.6%) – <Median (98.1%) n=35

Median (98.1) – <Q3 (99.5%) n=34

Q3 (99.5%) – Max (100.0%) n=35

8 (24.2) NE (22.9–NE) 1.0 1.0

7 (20.0) NE (28.1–NE) 0.57 (0.13–2.49) 0.4575

9 (26.5) 27.3 (18.8–NE) 1.01 (0.20–5.01) 0.9952

11 (31.4) 27.7 (22.3–NE) 0.95 (0.28–3.26) 0.9331

The landmark analysis was performed to study the effect of relative dose intensity on progression-free survival (PFS). The patients who completed venetoclax treatment and had not progressed or were censored at the last dose of venetoclax, were included. The PFS was calculated from the last dose of venetoclax to the first occurrence of progression or death from any cause. CI: confidence interval; HR: hazard ratio; Max: maximum; Min: minimum; NE: not estimable; PFS: progression-free survival; Q: quartile.

140

haematologica | 2022; 107(1)

Impact of discontinuation/modifications on outcome

reduction does not result in inferior PFS (Online Supplementary Table S5).33,34 Our literature review revealed evidence that treatment discontinuation of ibrutinib may be associated with poor survival outcomes; the impact of ibrutinib discontinuation may be significantly modified by factors such as the reason for discontinuation and the number of prior therapies received (Online Supplementary Table S1). The impact of ibrutinib interruption is harder to quantify from the literature (Online Supplementary Table S1); the phase III RESONATE study reported an impact on PFS with treatment interruption ≥8 consecutive days,16 while a phase II trial indicated that outcomes were not compromised by treatment interruption.19 Some real-world datasets also suggest an impact of interruption on outcomes,21,23,28 while others do not.24,26 Similarly the impact of ibrutinib dose reduction is difficult to quantify from the literature (Online Supplementary Table S5); one interventional and one real-world study suggest that there is no survival detriment with dose reduction,19,23 whereas inferior PFS is indicated in two real-world studies.24,28 The current literature consists of a mixture of clinical trials and real-world data; therefore patient characteristics, management of AE and the timing of data collection/availability of data may vary significantly. This could explain the inconsistencies in the data, where numerous studies have reported conflicting results, making conclusions particularly difficult to draw. In addition to this, it is important to also note the variation in the definitions used for dose reduction. While some studies indicate that data are representative of early dose reductions (e.g., within the first 8 weeks of initiating therapy28), other studies, such as MURANO, look at dose reductions occurring throughout therapy. Furthermore the lack of information indicating the permanence of dose reductions, or whether all dose reductions experienced by a patient were taken into account or not, could provide some explanation for inconsistencies in the data reported. Literature assessing the impact of discontinuation and interruption on outcomes was limited for both idelalisib and duvelisib. Discontinuation data from clinical trials of idelalisib showed similar OS in one study and reduced PFS in another,35,36 while real-world data indicated an effect on survival dependent upon reason for treatment discontinuation.37 Phase III data suggest an association between idelalisib interruption and longer PFS and OS (with length of interruption potentially having an impact on outcome), but this is difficult to confirm with only one study available (Online Supplementary Table S1).38 Similarly, conclusions are hard to draw for duvelisib treatment interruptions with only one study reporting the impact on PFS (which is similar regardless of duration of interruption [>1 week vs. >2 weeks]; Online Supplementary Table S1).39 The analyses presented here are the first to be conducted from MURANO for the impact of venetoclax dose modification or premature discontinuation. Since analyses were performed retrospectively, without type-I error control, the results should be interpreted with caution. This manuscript also provides the first literature review on this topic to collate the numerous articles and provide context for the MURANO analyses. Limited published

haematologica | 2022; 107(1)

data for the impact of venetoclax discontinuation and interruption on survival hindered comparison of the MURANO results with other studies of venetoclax. Moreover, the review was further limited by the inability to perform a meta-analysis in order to compare the MURANO results with studies of other targeted therapies (ibrutinib, idelalisib and duvelisib). In conclusion, these analyses provide the first comprehensive account of the impact of discontinuation, treatment interruption and dose reduction of oral targeted treatment on PFS and/or OS in patients with 1L or R/R CLL. These data highlight the importance of prompt treatment modification upon development of AE with venetoclax, as effective management of AE with treatment interruption or dose reduction allows patients to continue venetoclax treatment and avoid the negative impact of treatment discontinuations on outcomes. Disclosure ARM consults for or advises for AbbVie, AstraZeneca, Celgene, Genentech, Janssen, Loxo, PCYC, Sunesis and TG therapeutics, and has received research funding from AbbVie, AstraZeneca, Celgene, DTRM, Genentech, Janssen, Loxo, PCYC, Sunesis and TG Therapeutics; JPS consults for or advises for AbbVie, Acerta, AstraZeneca, Bayer Health, Celgene, F. Hoffmann–La Roche, Pharmacyclics and TG Therapeutics, has received research funding from AbbVie, Acerta, AstraZeneca, Celgene, Pharmacyclics and TG Therapeutics, has received travel support from AbbVie, AstraZeneca and Celgene, and has role or activity within Pfizer; JMLB, MW and YM are employees and stockholders of Genentech; SYK is an employee and stockholder of AbbVie, and has received travel support from AbbVie; KH and MB are employees and stockholders of F. Hoffmann–La Roche; QZ is an employee and stockholder of Edwards Life Sciences, and consults or advises for Genentech; JFS has received honoraria from and consults/advises for AbbVie, BMS, Celgene, F. Hoffmann–La Roche, Gilead, Janssen, Mei Pharma, Sunesis and Takeda, has participated in speaker’s bureaus for AbbVie and F. Hoffmann–La Roche, has received research funding from AbbVie, Celgene, F. Hoffmann– La Roche and Janssen, has provided expert testimony for F. Hoffmann–La Roche, and has received travel support from AbbVie, BMS, F. Hoffmann–La Roche and Janssen. Contributions All authors were involved in the analysis and interpretation of the data, as well as review and/or revision of the manuscript. Acknowledgments The authors would like to thank the patients and their families, investigators, study coordinators and support staff, and MURANO study team members. Funding Venetoclax is being developed in collaboration between Genentech and AbbVie. Genentech and AbbVie provided financial support for the study and participated in the design, study conduct, analysis, and interpretation of data, as well as the writing, review and approval of the manuscript. Medical writing support was provided by Rachel Dobb of Ashfield MedComms, an Ashfield Health company and was funded by F. Hoffmann– La Roche Ltd.

141

A.R. Mato et al.

References 1. Eichhorst B, Robak T, Montserrat E, et al. Chronic lymphocytic leukaemia: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2015;26 (Suppl 5):v78-84. 2. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2017. CA Cancer J Clin. 2017;67(1):730. 3. Dreger P. Allotransplantation for chronic lymphocytic leukemia. Hematology Am Soc Hematol Educ Program. 2009:602-609. 4. National Comprehensive Cancer Network Clinical Practice Guidelines in Oncology (NCCN Guidelines): Chronic Lymphocytic Leukemia/Small Lymphocytic Lymphoma. Version 2.2019. NCCN website. www.nccn.org/professionals/physician_gls/ pdf/cll.pdf. Published October, 2018. Accessed August, 2019. 5. Souers AJ, Leverson JD, Boghaert ER, et al. ABT-199, a potent and selective BCL-2 inhibitor, achieves antitumor activity while sparing platelets. Nat Med. 2013;19(2):202208. 6. Brown JR. Ibrutinib (PCI-32765), the first BTK (Bruton's tyrosine kinase) inhibitor in clinical trials. Curr Hematol Malig Rep. 2013;8(1):1-6. 7. Liu P, Cheng H, Roberts TM, et al. Targeting the phosphoinositide 3-kinase pathway in cancer. Nat Rev Drug Discov. 2009;8(8):627644. 8. Montero J, Letai A. Why do BCL-2 inhibitors work and where should we use them in the clinic? Cell Death Differ. 2018;25(1):56-64. 9. Fischer K, Al-Sawaf O, Bahlo J, et al. Venetoclax and obinutuzumab in patients with CLL and coexisting conditions. N Engl J Med. 2019;380(23):2225-2236. 10. Stilgenbauer S, Eichhorst B, Schetelig J, et al. Venetoclax in relapsed or refractory chronic lymphocytic leukaemia with 17p deletion: a multicentre, open-label, phase 2 study. Lancet Oncol. 2016;17(6):768-778. 11. Stilgenbauer S, Eichhorst B, Schetelig J, et al. Venetoclax for patients with chronic lymphocytic leukemia with 17p deletion: results from the full population of a phase II pivotal trial. J Clin Oncol. 2018;36(19):1973-1980. 12. Roberts AW, Seymour JF, Eichhorst B, et al. Pooled multi-trial analysis of venetoclax efficacy in patients with relapsed or refractory chronic lymphocytic leukemia. Blood. 2016;128(22):3230. 13. Jones J, Choi MY, Mato AR, et al. Venetoclax (VEN) Monotherapy for patients with chronic lymphocytic leukemia (CLL) who relapsed after or were refractory to Ibrutinib or Idelalisib. Blood. 2016;128(22):637. 14. Jain P, Keating M, Wierda W, et al. Outcomes of patients with chronic lymphocytic leukemia after discontinuing ibrutinib. Blood. 2015;125(13):2062-2067. 15. Maddocks KJ, Ruppert AS, Lozanski G, et al.

142

Etiology of ibrutinib therapy discontinuation and outcomes in patients With chronic lymphocytic leukemia. JAMA Oncol. 2015;1(1):80-87. 16. Barr PM, Brown JR, Hillmen P, et al. Impact of ibrutinib dose adherence on therapeutic efficacy in patients with previously treated CLL/SLL. Blood. 2017;129(19):2612-2615. 17. Jain P, Thompson PA, Keating M, et al. Longterm outcomes for patients with chronic lymphocytic leukemia who discontinue ibrutinib. Cancer. 2017;123(12):2268-2273. 18. O'Brien SM, Byrd JC, Hillmen P, et al. Outcomes with ibrutinib by line of therapy and post-ibrutinib discontinuation in patients with chronic lymphocytic leukemia: phase 3 analysis. Am J Hematol. 2019;94(5):554-562. 19. Ahn IE, Basumallik N, Tian X, et al. Clinically-indicated ibrutinib dose interruptions and reductions do not compromise long-term outcomes in CLL. Blood. 2019;133(22):2452-2455. 20. Sandoval-Sus JD, Chavez JC, Dalia S, et al. Outcomes of patients with relapsed/refractory chronic lymphocytic leukemia after ibrutinib discontinuation outside clinical trials: a single institution experience. Blood. 2015;126(23):2945. 21. Thompson PA, Levy V, Tam CS, et al. The impact of atrial fibrillation on subsequent survival of patients receiving ibrutinib as treatment of chronic lymphocytic leukemia (CLL): An international study. Blood. 2016; 128(22):3242. 22. Akhtar OS, Torka P, Bhat SA, et al. Disease progression on ibrutinib therapy is associated with a poor clinical outcome in chronic lymphocytic leukemia (CLL) patients managed in standard clinical practice. Blood. 2017;130(Suppl 1):S5350. 23. Follows GA, CLL Forum UK. Ibrutinib for relapsed/refractory CLL: an update of the UK and Ireland analysis of outcomes in 315 patients. Haematologica. 2017;35(S2):238239. 24. Rhodes J, Barr PM, Ujjani CS, et al. The impact of front-line ibrutinib dose reduction and interruption on outcomes in chronic lymphocytic leukemia (CLL) patients. Blood. 2017;130(Suppl 1):S4313. 25. Sharman JP, Black-Shinn JL, Clark J, et al. Understanding ibrutinib treatment discontinuation patterns for chronic lymphocytic leukemia. Blood. 2017;130(Suppl 1):S4060. 26. Winqvist M, Andersson P-O, Asklid A, et al. Real-world results on ibrutinib in relapsed/refractory CLL: 30-month followup of 95 Swedish patients treated in a compassionate use program. HemaSphere. 2018;2(Suppl 1):S128-129. 27. Hampel PJ, Ding W, Call TG, et al. Rapid disease progression following discontinuation of ibrutinib in patients with chronic lymphocytic leukemia treated in routine clinical practice. Leuk Lymphoma. 2019;60: (11):2712-2719.

28. Williams AM, Baran AM, Casulo C, et al. Ibrutinib dose adherence and therapeutic efficacy in non-Hodgkin lymphoma: a single-center experience. Clin Lymphoma Myeloma Leuk. 2019;19(1):41-47. 29. Seymour JF, Kipps TJ, Eichhorst B, et al. Venetoclax–rituximab in relapsed or refractory chronic lymphocytic leukemia. N Engl J Med. 2018;378(12):1107-1120. 30. Seymour JF, Kipps TJ, Eichhorst BF, et al. Four-year analysis of Murano study confirms sustained benefit of time-limited venetoclax-rituximab (VenR) in relapsed/refractory (R/R) chronic lymphocytic leukemia (CLL). Blood. 2019;134(Suppl 1):S355. 31. Lew TE, Anderson MA, Lin VS, et al. Undetectable peripheral blood MRD should be the goal of venetoclax in CLL, but attainment plateaus after 24 months. Blood Adv. 2020;4(1):165-173. 32. Mato AR, Tam CS, Allan JN, et al. Disease and patient characteristics, patterns of care, toxicities, and outcomes of chronic lymphocytic leukemia (CLL) patients treated with venetoclax: A multicenter study of 204 patients. Blood Adv. 2017;130(Suppl 1):S4315. 33. Eyre TA, Kirkwood AA, Gohill S, et al. Efficacy of venetoclax monotherapy in patients with relapsed chronic lymphocytic leukaemia in the post-BCR inhibitor setting: a UK wide analysis. Br J Haematol. 2019;185(4):656-669. 34. Roeker LE, Fox CP, Eyre TA, et al. Tumor lysis, adverse events, and dose adjustments in 297 venetoclax-treated CLL patients in routine clinical practice. Clin Cancer Res. 2019;25(14):4264-4270. 35. Brown JR, Ghia P, Jones JA, et al. Outcomes of patients with relapsed and refractory chronic lymphocytic leukemia (CLL) who discontinue idelalisib treatment. J Clin Oncol. 2016;34(Suppl):S7531. 36. Thompson PA, Stingo F, Keating MJ, et al. Outcomes of patients with chronic lymphocytic leukemia treated with first-line idelalisib plus rituximab after cessation of treatment for toxicity. Cancer. 2016;122(16): 2505-2511. 37. Bange E, Nabhan C, Brander DM, et al. Realworld evidence for durable treatment responses after toxicity related discontinuation of idelalisib. Blood. 2017;130(Suppl 1):S4325. 38. Ma S, Chan RJ, Ye W, et al. Survival outcomes following idelalisib interruption in the treatment of relapsed or refractory indolent non-Hodgkin's lymphoma and chronic lymphocytic leukemia. Blood. 2018;132 (Suppl 1):S3149. 39. Flinn I, Montillo M, Illés Á, et al. Effect of dose modifications on response to duvelisib in patients with relapsed/refractory (R/R) CLL/SLL in the DUO trial. J Clin Oncol. 2019;37(Suppl 15):S7523.

haematologica | 2022; 107(1)

ARTICLE

Chronic Lymphocytic Leukemia

The miR-200c/141-ZEB2-TGFb axis is aberrant in human T-cell prolymphocytic leukemia

Ferrata Storti Foundation

Stefan J. Erkeland,1 Christiaan J. Stavast,1 Joyce Schilperoord-Vermeulen,1 Giada Dal Collo,1 Harmen J.G. van de Werken,1,2 Leticia G. Leon,1 Antoinette van Hoven-Beijen,1 Iris van Zuijen,1 Yvonne M. Mueller,1 Eric M. Bindels,3 Dick de Ridder,4 Mies C. Kappers-Klunne,3 Kirsten van Lom,3 Vincent H.J. van der Velden1 and Anton W. Langerak1 1

Department of Immunology, Erasmus MC, University Medical Center, Rotterdam; Cancer Computational Biology Center, Cancer Institute, Erasmus MC, University Medical Center, Rotterdam; 3Department of Hematology, Erasmus MC, University Medical Center, Rotterdam and 4Bioinformatics Group, Wageningen University, Wageningen, the Netherlands 2

Haematologica 2022 Volume 107(1):143-153

ABSTRACT

-cell prolymphocytic leukemia (T-PLL) is mostly characterized by aberrant expansion of small- to medium-sized prolymphocytes with a mature post-thymic phenotype, high aggressiveness of the disease and poor prognosis. However, T-PLL is more heterogeneous with a wide range of clinical, morphological, and molecular features, which occasionally impedes the diagnosis. We hypothesized that T-PLL consists of phenotypic and/or genotypic subgroups that may explain the heterogeneity of the disease. Multi-dimensional immuno-phenotyping and gene expression profiling did not reveal clear T-PLL subgroups, and no clear T-cell receptor a or b CDR3 skewing was observed between different T-PLL cases. We revealed that the expression of microRNA (miRNA) is aberrant and often heterogeneous in T-PLL. We identified 35 miRNA that were aberrantly expressed in T-PLL with miR-200c/141 as the most differentially expressed cluster. High miR200c/141 and miR-181a/181b expression was significantly correlated with increased white blood cell counts and poor survival. Furthermore, we found that overexpression of miR-200c/141 correlated with downregulation of their targets ZEB2 and TGFbR3 and aberrant TGFb1induced phosphorylated SMAD2 (p-SMAD2) and p-SMAD3, indicating that the TGFb pathway is affected in T-PLL. Our results thus highlight the potential role for aberrantly expressed oncogenic miRNA in T-PLL and pave the way for new therapeutic targets in this disease.

Introduction T-cell prolymphocytic leukemia (T-PLL) is a relatively rare disease accounting for approximately 2% of mature lymphocytic leukemia patients older than 30 years.1 T-PLL is characterized by an uncontrolled expansion of malignant, mature, postthymic T lymphocytes in the bone marrow (BM), lymph nodes, liver and spleen, resulting in hepato-splenomegaly, lymphadenopathy, skin lesions and a high leukocyte count. Whole-transcriptome analysis of T-PLL cells revealed that the most significantly deregulated genes are involved in T-cell receptor signaling, cytokine signaling and p53-controlled apoptosis.2 In addition, frequent recurrent mutations in DNA damage repair/tumor suppressor genes (e.g., ATM, TP53, MSH3, MSH6, SAMHD1, PARP10, HERC1, HERC2), and oncogenes (JAK1 and JAK3) have been described in T-PLL.3,4 The majority of T-PLL cases exhibit a complex karyotype.3 Many distinct chromosomal abnormalities, including common chromosomal translocations involving chromosome 8, 11 and 14, deletions in 11q, 8p and 7q34q36 and gains of 8q and 8p11p12 have been reported.2,5 However, most T-PLL cases have abnormalities of chromosome 14, of which inv(14)(q11q32) and t(14:14)(q11;q32) are found most frequently.5 Chromosome 14 abnormalities bring the T-cell leukemia/lymphoma 1A (TCL1A, TCL1) oncogene on chromo-

haematologica | 2022; 107(1)

Correspondence: STEFAN J. ERKELAND s.erkeland@erasmusmc.nl ANTON W. LANGERAK a.langerak@erasmusmc.nl Received: June 18, 2020. Accepted: January 22, 2021. Pre-published: February 18, 2021. https://doi.org/10.3324/haematol.2020.263756

143

S.J. Erkeland et al.

some 14q32 in close proximity to the T-cell receptor-a/d (TRA/TRD) regulatory region on chromosome 14q11, or more rarely to T-cell receptor-b (TRB) regulatory elements located on 7q35, resulting in overexpression of TCL1.6 TCL1 is an adapter protein of 14 kDa that functions in kinase complexes and enhances T-cell receptor-mediated pro-survival signaling.7 The gene encoding mature T-cell proliferation-1 (MTCP1), is highly homologous to TCL1 and is involved in the less frequent translocations t(X;14)(q28;q11) and t(X;7)(q28;q35).5 The total incidence of translocations involving TCL1 and MTCP1 is around 90%.5 The high incidence of TCL1 and MTCP1 translocations strongly suggest that this family of proteins plays a key role in the pathogenesis of T-PLL. In full agreement, different TCL1 and MTCP1 transgenic mouse models demonstrate a role for TCL1 and MTCP1 in the initiation of malignant transformation to overt leukemia similar to T-PLL,8-10 albeit with a long latency of 12-20 months. These data indicate that secondary oncogenic events are required for full oncogenic transformation of the mature T cells. Identification of co-operating leukemia genes in the Em-TCL1 transgenic mouse with a Sleeping Beauty transposon-mediated mutagenesis screen revealed increased nuclear factor (NF)-kB signaling as a collaborating event in TCL1-mediated leukemogenesis.11 The role for small noncoding RNA in the pathogenesis of T-PLL has not been investigated yet. MicroRNA (miRNA) are an abundant class of small noncoding RNA of 19-24 nucleotides, which guide the RNAinduced silencing complex (RISC) to reverse and partially complementary sequences in the 3’-untranslated regions (3’-UTR) of mRNA and control gene expression through translational inhibition and mRNA destabilization.12 Aberrant expression, biogenesis and activities of miRNA are hallmarks of human cancer, including malignant hematological disorders.13-17 Numerous miRNA have been identified as diagnostic biomarkers for human leukemia.16 Some miRNA are potent oncogenic drivers of leukemogenesis, while others have critical tumor-suppressing activities.18-22 Here, we characterized a cohort of 31 T-PLL cases by cytomorphology, immuno-phenotyping and immunogenetic analysis, as well as mRNA and miRNA profiling to identify novel mechanisms that contribute to leukemic transformation. We revealed that the expression of miRNA is aberrant, though often heterogeneous, in T-PLL. Additionally, we present the first evidence that aberrant expression of miR-200c/141 affects TGFb-controlled mechanisms in T-PLL that may contribute to its pathogenesis.

Methods Patient cohort The patient cohort consisted of 31 well-defined T-PLL cases diagnosed between 1985 and 2011. Sampling was done at the moment of diagnosis. Diagnosis of T-PLL is based on a combination of clinical features, morphology, cytogenetics and immunophenotype. The cohort consisted of males (62%) and females (38%) with a median age at diagnosis of 66 years (range, 41-89 years) (Table 1). In our cohort, 52% of the T-PLL patients were CD4+ (n=14), 33% were CD4+ CD8+ (n=9), 11% were CD8+ (n=3), 4% were CD4–CD8– (n=1) and one was not determined. All patients were characterized for known aberrations in genes such

144

as ATM, TP53, TCL1 and IGH. Use of T-PLL patient samples was approved by the Erasmus MC Medical Ethics Committee (MEC2015-617). A cohort of 31 T-PLL cases was characterized morphologically, cytogenetically and immunophenotypically. We could isolate high quality RNA from 23 T-PLL patients for gene expression profiling. The average RIN value of the T-PLL samples was 8.7 with a standard deviation of 0.9. Of these, the RNA of 21 T-PLL cases was sufficient for miRNA expression profiling. All RNA samples resulted in high-quality cDNA libraries. Four additional T-PLL samples (T-PLL44, 46, 48 and 49) were used for validation experiments and Western Blotting.

Normal T-cell subsets Peripheral blood from healthy donors were obtained from buffy coats (Sanquin) with approval by the Erasmus MC Medical Ethics Committee (MEC-2016-202). For gene expression profiling, peripheral blood mononuclear cells (PBMC) were first isolated by Ficoll-Paque (density 1.077 g/mL, Pharmacia). CD4+ T cells were labeled with anti-CD3-PerCP-Cy5.5 (cat.#340948) and anti-CD4PE-Cy7 (cat.#557852) and CD8+ T cells were labeled with antiCD3-PerCP-Cy5.5 (cat.#340948) and anti-CD8-APC-H7 (cat.#560273)(BD Biosciences). Naïve T cells (CD27+ and CD45RA+), memory (CD27+ and CD45RO+ and effector (CD27– and CD45RA+) T-cell subsets were labeled with, anti-CD45RA-PE (cat.# 555489), anti-CD27-APC (cat.#558664), anti-CD45RO-FITC (cat.#555492) and anti-CD45RA-PE cat.#555489) antibodies (BD Biosciences) (Online Supplementary Table S1). In order to prevent antibody-mediated activation of T cells, all cell populations were kept on ice during antibody staining. Cells were sorted at 4°C and directly in TRIzol (Thermofisher) to prevent cell activation, cell death and/or RNA degradation. Cells fractions were sorted on a FACS Aria II cell sorter (BD Biosciences). Data availability: i) GSE147930: microarray gene expression data; ii) GSE147931: mRNA sequencing data; iii) GSE147932: small RNA sequencing data. For a detailed description of the immunophenotyping and cytmorphological analysis, fluorescence in situ hybridization analysis and cytogenetics, T-cell receptor gene rearrangement analysis, Lentiviral miRNA expression vectors and transduction of Jurkat and HeLa cells, gene expression profiling as well as RNA sequencing and data analysis see the Online Supplementary Appendix.

Results Cytomorphologic, immunophenotypic and immunogenetic analysis of T-cell prolymphocytic leukemia As T-PLL is a heterogeneous disease,23 we first characterized our T-PLL cohort with respect to clinical, morphological, phenotypical and molecular features (Table 1). Cytomorphologic analysis resulted in three different TPLL subgroups: i) cells with small lymphocytic cell morphology (64%); ii) cells with large blast-like morphology (10%); iii) cells with equivocal morphology (26%) (Online Supplementary Figure S1A). We determined deletions in TP53 (24% of cases tested) and ATM (64% of cases tested) as well as translocations involving TCL1 located on 14q32 (67% of cases tested). However, we did not observe any correlation between cell morphology and different parameters including known aberrations in genes, e.g., ATM, TP53, and TCL1. In order to determine the cell of origin of T-PLL, we performed flowcytometric immunophenotyping of T-PLL samples using the EuroFlow T-CLPD panel24 (www.euroflow.org) taking flow haematologica | 2022; 107(1)

Oncogenic microRNA in T prolymphocytic leukemia

Table 1. Characteristics of the cohort of 31 T-cell prolymphocytic leukemia patients.

Sex

Age

WBC

Physical examination L H S

Other

1 2 3 4 5 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 28 29 44 46 48 49

M M M F F M M F M M M F M M M F M F M M F F M M F M M M F F M

65 58 41 63 50 74 60 77 78 79 49 67 62 67 62 77 61 67 69 69 72 89 67 77 48 44 52 81 76 77 75

740 670 79 393 14 42 69 11 9 55 34 115 245 59 643 72 61 493 28 24 231 550 25 243 31 27 50 120 232 293 48

+ + + + + + + + + n.a. + + + + + n.a. + n.a. -

+ + + + + + + + + + + + + + n.a. + + + n.a. + n.a. -

Sk CNS Sk Sk, L Sk P n.a. n.a. n.a. -

+ + + + + + + + + + + n.a. + + + n.a. n.a. +

Survival

Karyotype Major abnormalities

0.2 inv(14)(q32) 9 inv(14)(q32)/t(8;8) 9 idic(8)(p11) 0.5 inv(14)(q32) 14 n.d. 28 n.d. 7 n.d. 3.5 none 7 n.d. 4 n.d. 87 n.d. 6 n.d. 5 none 26 n.d. 8 n.d. 10 n.d. 47+ t(6;12) 16+ n.d. n.a. n.d. n.a. n.d. 8 n.d. 9 n.d. 130+ n.d. 4.5 inv(14)(q32) 7.7 n.d. 2+ none 3.5+ n.d. dead inv(14)(q32) dead n.d. n.a. inv(14)(q32) dead n.d.

FISH Clonality Immunophenotype 17p/ 11q / 14q/ TRB/TRG CD4/CD8 cyTCL1 TP53 ATM TCL1 n.d. N n.d. n.d. n.d. N + N N n.d. N N + N N + + N N N N + N N n.d. N N n.d. n.d. n.d. n.d.

n.d. + n.d. n.d. n.d. N + N N + + + + N + N N + + N + + N + n.d. + + n.d. n.d. n.d. n.d.

n.d. + N n.d. n.d. n.d. + n.d. N + n.d. + + + + + n.d. + n.d. n.d. N + n.d. + n.d. n.d. + n.d. n.d. n.d. n.d.

clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal clonal

CD4+CD8– CD4+CD8– CD4-CD8+ CD4+CD8– CD4+CD8+ CD4+CD8– CD4+CD8+ CD4+CD8– CD4-CD8CD4+CD8+ CD4+CD8+ CD4+CD8+ CD4+CD8– CD4+CD8– CD4+CD8+ CD4+CD8– CD4+CD8– CD4+CD8– CD4+CD8– CD4+CD8– CD4+CD8+ CD4+CD8– CD4+CD8– CD4+CD8+ CD4–CD8+ CD4–CD8+ CD4+CD8+ CD4+CD8– CD4+CD8– CD4+CD8– CD4+CD8–

+ + + + + + + + + + + + + + + + n.d. + + +

WBC: white blood cell count 109/L as determined at diagnosis; L: lymphadenopathy; H: hepatomegaly; S: splenomegaly; O: other; Sk: skin; L: lungs; P: pleural, A: ascites, CNS: central nervous system; TP53: TP53 abnormalities as detected by fluorescence in situ hybridization (FISH); ATM: ATM abnormalities as detected by FISH; TCL1: 14q32 translocations involving TCL1 as detected by FISH; cyTCL1: cytoplasmic TCL1 protein expression; n.a.: not available; n.d.: not determined; N: normal.

sorted CD4+ naïve (CD4N), CD4+ effector (CD4E), CD4+ memory (CD4M), CD8+ naïve (CD8N), CD8+ effector (CD8E) and CD8+ memory (CD8M) T-cell fractions from healthy individuals as reference. When using only T-PLL cases in a multidimensional APS plot analysis, the majority of T-PLL appeared to cluster according to similarities in their phenotype (Online Supplementary Figure S1B). However, two outliers were noted, T-PLL-10 and T-PLL24. Judging by the markers that were found to contribute to PC1 (mainly CD45RA) and PC2 (mainly CD45RA, CD45RO, CD27), the larger T-PLL cluster presumably would have a memory phenotype, while the outliers would resemble the naïve and/or effector T-cell subsets more (Online Supplementary Figure S1B). APS-based analysis of T-PLL cases that were added to fixed APS plots of normal T-cell subsets indeed confirmed that the majority of T-PLL cases did cluster close to or within the CD4+ and CD8+ memory T-cell subsets, whereas some T-PLL were more similar to non-memory subsets (Online haematologica | 2022; 107(1)

Supplementary Figure S1C). As clonal TRA and TRB gene rearrangements with stereotyped CDR3 have proven to be relevant for some T-cell leukemias, e.g., T-large granular lymphocyte leukemia (T-LGL),25,26 we wondered whether this could be relevant for T-PLL as well. First, we asked whether V and J gene usage would define any subgroups of the T-PLL samples. To this end, we performed Sanger sequencing of the clonal TRA and TRB gene rearrangements. In agreement with previously published data,27 sequence analysis of T-PLL samples revealed no clear skewing of V and J gene usage of TRA and TRB genes (Online Supplementary Figure S1D and E). In addition, we did not find stereotyped CDR3 motifs in the productive TRA and TRB gene rearrangements in T-PLL either (data not shown). In summary, most T-PLL displayed cell morphology and immunophenotypical characteristics reflecting normal memory T cells, with no clear evidence for antigen-driven leukemogenesis. 145

S.J. Erkeland et al.

Figure 1. Analysis of gene expression profiles of 23 T-cell prolymphocytic leukemia cases and 18 normal T-cell subsets from healthy controls. (A) A correlation plot showing that T-cell prolymphocytic leukemia (T-PLL) are separated from normal T-cell subsets based on their gene expression profiles. A subset of probesets that showed signal, i.e., for which the median absolute deviation (MAD) from the median exceeds threshold T=0.6 on a log2 scale were selected (total of 3,261 probesets). Color bar represents correlation score. (B) Heatmap of T-PLL cases and normal T-cell subsets after unsupervised clustering (T=0.7; total of 1,904 probesets). Most T-PLL cases tightly cluster, without evident subgroups being detectable. Except for cases 21 and 24, most T-PLL cases cluster, without obvious correlation with immunological subgroups. Color reflects expression level: blue=low, red=high. (C) Expression of the TCL1A oncogene in T-PLL subgroups (all T-PLL, CD4+CD8– T-PLL and CD4+CD8+ T-PLL) compared to indicated normal T-cell fractions. E: effector subset; M: memory subset; N: naïve subset. Significance between the groups was calculated with the Mann-Whitney-U test **P<0.01, ****P<0.0001.

146

haematologica | 2022; 107(1)

Oncogenic microRNA in T prolymphocytic leukemia

Figure 2. MicroRNA are deregulated in T-cell prolymphocytic leukemia. (A) Principal component analysis (PCA) based on microRNA (miRNA) expression profiles of indicated 21 T-cell prolymphocytic leukemia (T-PLL) cases and T cell fractions. (B) Heatmap of 35 most differential expressed miRNA between CD4+ effector T cells (CD4E) and T-PLL. Heatmap shows the robust normalized z-scores of each miRNA indicated by the color legend and set to maximum 2 (dark red color). The columns show 25 samples, which are hierarchically clustered using their Euclidean distances with complete linkage and Ward’s method. The miRNA are ordered top-down by increased false discovery rate values. (C) Expression of miR-141, miR-200c, miR-181a and miR-181b in T-PLL subgroups (all T-PLL, CD4+CD8– T-PLL and CD4+CD8+ T-PLL) compared to indicated normal T-cell fractions. E: effector subset; M: memory subset; N: naïve subset. Data was tested for normal distribution. Significance between the groups was calculated with the Mann-Whitney-U test (black) or with the Student’s t-test (red) *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. (D) TPLL cases were grouped based on high miR-200c/141 and or miR-181a/b expression (fold-change [FC] >2 compared to normal CD4E, n=16) and low miR-200c/141 and miR-181a/b expression (FC <2, n=5). White blood cell counts (WBC) are displayed. Statistical significance was determined with a Student’s t-test with Welch’s correction. (E) Kaplan Meier plot showing cumulative survival of patients with high vs. low miR-200c/141 expression. Statistical significance was calculated with the Mantel Cox test (log-rank test).

haematologica | 2022; 107(1)

147

S.J. Erkeland et al.

T-cell prolymphocytic leukemia show a relatively homogeneous gene expression profile with TCL1A amongst the highest upregulated genes We then performed gene expression array analysis of T-PLL samples and normal T-cell subsets. A correlation plot and unsupervised clustering analysis of gene expression showed that most T-PLL samples, with the exception of T-PLL21 and T-PLL24, and to a lesser extent T-PLL3, TPLL22, T-PLL-25, clustered together; at the other end, normal T-cell subsets showed a high correlation to each other (Figure 1A and B). Next, we focused on CD4+ T-PLL, which is the major group within T-PLL, and compared their gene expression to that of normal CD4+ T-cell subsets. We found 2,282 probe sets differentially expressed (at least 2-fold up- or downregulated) in CD4+ T-PLL compared to normal CD4 subsets. Of the top 100 differentially-expressed probe sets in CD4+ T-PLL, the average expression of most probe sets was upregulated, including the T-PLL characteristic T-cell leukemia driving oncogene TCL1A (Figure 1C; Online Supplementary Table S5). Collectively these data showed that the protein encoding gene expression profile of T-PLL is clearly different from normal T-cell subsets.

A subset of microRNA is differentially expressed in T-cell prolymphocytic leukemia of which miR-141 and miR-200c show prognostic impact In order to evaluate the role of small non-coding RNA in the T-PLL cohort, we then performed genome-wide miRNA expression profiling by small RNA sequencing of 21 T-PLL cases. Based on miRNA expression, all T-PLL cases clustered together in a principal component analysis (PCA) plot and were most similar to healthy effector CD4 cells (Figure 2A). Therefore, we then compared the expression of individual miRNA in T-PLL with healthy effector CD4 cells. Strikingly, we did not find any miRNA significantly downregulated in T-PLL compared to normal effector CD4 cells (false discovery rate [FDR]<0.05). However, we did identify a set of 35 miRNA that were abundantly expressed and significantly upregulated (FDR<0.05; fold change [log2FC] >1.5) in the T-PLL cohort compared to healthy effector CD4 cells (Figure 2B; Online Supplementary Table S6). These included miR-17-5p, miR-19b-3p, miR20a, miR-106b, miR-92a of the oncogenic miR-17~92 clusters and other well-known oncogenic miRNA, including miR-125a, that were also expressed at higher levels in TPLL cases (Online Supplementary Table S6). miRNA of the miR-200c/141 cluster (FDR=0.011; average overexpression FC >55; range, 0.95-1,211) and the miR-181a/b cluster (FDR=0.025; average overexpression FC >20; range, 0.2-451), which were the highest overexpressed miRNA in T-PLL (Figure 2B; Online Supplementary Table S6). Notably, miR-141, miR-200c, miR-181a and miR-181b were significantly overexpressed compared to other normal T-cell fractions (Figure 2C). Subgroups of T-PLL with pronounced overexpression of miR-200c/141 and/or miR-181a/b (T-PLL-1-5,10, 11, 16, 17, 21-25, 28) (range, 10-1,200-fold) significantly correlated with higher white blood cell counts (Figure 2D). Furthermore, after splitting the patients in two groups (>5-fold overexpression compared to effector CD4 cells and <5-fold overexpression compared to effector CD4 cells) and performing a survival analysis, we found that the ten patients with the higher miR-141 and miR-200c levels (T-PLL-1-4, 11, 21, 22, 23, 25 148

and 28, had a shorter survival compared with the eleven patients with lower levels of miR-141 and miR-200c (T-PLL-5, 9, 10, 16-20, 24, 26, 29) (P=0.042, Mantel-Cox log rank test) (Figure 2E). In contrast, no difference in survival was found based on miR-181 expression (data not shown). A multiple regression analysis with risk factors including age, sex, white blood cell count and immunophenotype did not determine a confounding effect on the survival. Together, these data present the first evidence that miRNA are deregulated but heterogeneously expressed in T-PLL and that miR-141 and miR-200c expression levels may have prognostic value to define a subgroup of T-PLL with relatively poor survival.

The miR-200c/141-ZEB2-TGFb axis is aberrant in T-cell prolymphocytic leukemia Next, we asked whether aberrant expression of miR200c/141 and miR-181a/b results in downregulation of their TargetScan-predicted well-conserved mRNA targets (Online Supplementary Tables S2 to S4) in T-PLL. In order to directly correlate miRNA levels to mRNA expression profiles, we generated transcriptomics data of CD4+ T-PLL samples with high expression levels (>10-fold overexpressed compared to effector CD4 cells) of these miRNA (high miR-200c/141 samples: T-PLL-1, 3, 4, 22, 23, 25; high miR-181a/b samples: T-PLL 3, 4, 10, 16, 17, 23 and 25) (Figure 2B). Effector CD4 T cells from healthy individuals were used as controls, as the miRNA profiles of T-PLL cells appeared most similar to this T-cell subset. We first analyzed several genes that may be correlated with the upregulation of miR-200c/141 and miR-181a/b. For instance, involvement of enhanced miRNA expression was postulated based on elevated Argonaute-2 (AGO2) expression as a result of chromosome 8 gains in a subset of T-PLL patients.2 However, AGO2 mRNA expression was not significantly increased in our T-PLL cohort (data not shown). The miR-200c/141 cluster is frequently coexpressed with protein tyrosine phosphatase, non-receptor type 6 (PTPN6) through transcriptional read-through and complex 3D chromatin interactions between PTPN6 and miR-200c/141 promoters.28 Accordingly, we found a significant upregulation of PTPN6 in T-PLL samples with enhanced miR-200c/141 expression (FC=3.7; FDR=0.0003; Online Supplementary Figure S2A). Furthermore, MIR200C/141 is known to be transcriptionally upregulated by MYC,41,42 an oncogenic transcription factor that is frequently overexpressed in T-PLL.2,43 In agreement, we noted a significant overexpression of MYC in T-PLL (FC=4.9; FDR=0.0019; Online Supplementary Figure S2B). In normal T-cell development miR-181 controls expression of various phosphatases, such as SHP2, PTPN22, DUSP5, DUSP6 and NRARP, thereby enhancing NOTCH and T-cell receptor signaling.29 In agreement, we found a significant downregulation of DUSP5 mRNA expression in T-PLL cells (FC=5.7; FDR=1.43x10-6; Online Supplementary Figure S2C). Gene set enrichment analysis (GSEA) showed a significant enrichment of miR-200c/141 and miR-181a/b targets in the downregulated fraction of genes in T-PLL compared to normal effector CD4 T cells (P<0.05) (Figure 3A and B). Ingenuity pathway analysis (IPA) of significantly deregulated genes in T-PLL revealed that the TGFb network is affected in T-PLL, including transforming growth factor b1 (TGFb1) (FC=0.276; P<7.68x10-10), TGFb2 (FC=0.339, FDR=1.85x10-5 and TGFb receptor 3 (TGFbR3) haematologica | 2022; 107(1)

Oncogenic microRNA in T prolymphocytic leukemia

Figure 3. MicroRNA target analysis. (A and B) Gene set enrichment analysis (GSEA) of miR-200c/141 (A) and miR-181a/b (B) predicted targets (www.targetscan.org). In red are shown upregulated genes and in blue downregulated genes. The enrichment is indicated by the green line. NES is the normalized enrichment score. The significance is given by the normalized (NOM) P-value, false discovery rate (FDR) q-val and family-wise error rate (FWER) P-val. (C) Significantly downregulated target genes for the indicated microRNA (miRNA) are shown. (D) Box plots showing the expression of ZEB2 and TGFbR3 in T-cell-prolymphocytic leukemia (T-PLL) compared to effector CD4 T cells. The statistical significance is indicated by the FDR value. (E) Average expression of ZEB2 and TGFbR3 relative to GAPDH and CD4 effector cells is shown. Samples were measured in triplicate. Statistical significance was determined with a Mann-Whitney-U test. E: effector subset; M: memory subset; N: naïve subset. (F) Expression of miR-200c/141 targets in Jurkat cells that are transduced with retroviruses expressing miR-200c/141, relative to the expression in Jurkat cells transduced with empty vector (EV)-control virus (red). Statistical significance was determined with a Student’s t-test.

haematologica | 2022; 107(1)

149

S.J. Erkeland et al.

Figure 4. The TGFbR pathway is deregulated in T-cell prolymphocytic leukemia. (A) Overexpression of miR-200c and miR-141 relative to U6 in HeLa-Mir-200c/141 compared to HeLa-empty vector (EV) is plotted. (B) The mRNA expression of ZEB2 and TGFbR3 relative to GAPDH in HeLa-Mir-200c/141 compared to HeLa-EV is plotted. (C) HeLa-Mir-200c/141 and HeLa-EV cells were treated with hu-TGFb1 (10 ng/mL, +) or not stimulated (0 ng/mL, - hu-TGFb1) for 1 hour. Samples of cell lysates were taken and analyzed by western blotting using phospho-specific antibodies against SMAD2 and SMAD3, total SMAD2/3, b-actin and TGFβR1. Data are representative of three independent experiments. (D) Cells were stimulated with hu-TGFb1 (+ hu-TGFb1) or not stimulated (0 ng/mL, - hu-TGFβ1) for 1 hour. Samples of indicated cell lysates were analyzed by western blotting with phospho-specific antibodies against SMAD2 and SMAD3, total SMAD (SMAD2 = upper band, SMAD3 = lower band) and β-actin. (E) Quantification of the data shown in (D). Induction of hu-TGFb1-mediated phosphorylation of indicated phospo-proteins relative to the total SMAD expression level and peripheral blood mononuclear cells (PBMC) values of the different samples are depicted.

(FC=0.046; FDR=9.66x10-19), which are all significantly downregulated in T-PLL (Online Supplementary Figure S3A), whereas the mRNA expression of TGFbR1 and TGFbR2 was not changed. Our array data showed reduced expression of TGFbR3 compared to all normal T-cell fractions tested, whereas TGFβR1 and TGFbR2 was not changed (Online Supplementary Figure S3B). Furthermore, we noted a reduced TGFbR3 membrane levels in T-PLL with high miR-200c/141 expression compared to normal CD4 T cells (Online Supplementary Figure S3C and D). Notably, despite unchanged TGFbR1 mRNA and protein levels in T-PLL, we did observe reduced TGFbR1 membrane 150

expression on T-PLL compared to normal CD4 T cells (Online Supplementary Figure S3C and E). A global test on the gene expression profiles of the predicted miR-200c/141 targets in high miR-200c/141 samples and miR-181a/b targets in high miR-181a/b samples revealed that 220 miR-200c/141 targets and 115 miR181a/b targets were significantly downregulated in T-PLL compared to effector CD4 cells, including Zinc finger E box-binding homeobox 2 (ZEB2, also known as Smadinteracting protein 1 [SIP1]) and TGFbR3 (Figure 3C and D). We noted that ZEB2 expression is highly variable in normal T-cell fractions with the highest expression in effector CD8 haematologica | 2022; 107(1)

Oncogenic microRNA in T prolymphocytic leukemia

cells and the lowest expression in naïve CD4 T cells (Figure 3E). Furthermore, we confirmed that the expression of ZEB2 in T-PLL is downregulated compared to effector CD4 T cells (Figure 3E). However, TGFbR3 expression is more homogeneously expressed in normal T-cell fractions and significantly downregulated in T-PLL compared to average of all normal T-cell fractions (Figure 3E). In addition, downregulation of ZEB2 and TGFbR3 (>80%) was found in a set of three additional CD4-positive T-PLL samples (T-PLL49, 44, 46) with high miR-200c and miR-141 levels compared to normal PBMC (Online Supplementary Figure S3E). ZEB2 mRNA contains four predicted well-conserved 8-mer sites, one 7-mer-m8 and one 7mer-A1 site for miR-200c-3p, three well-conserved 8-mer sites for miR-141-3p and is a validated target of miR-200c and miR-141.30 ZEB2, has been identified as a SMAD-interacting protein and regulates the transcriptional activities of the TGFb network.31 According to TargetScan, TGFbR3 contains one less conserved miR-200c site in the 3’-UTR. In order to further investigate the functions of miR-200c/141 in regulation of the TGFb network, we transduced Jurkat cells with pLX301-miR-200c/141 (co-expressing EGFP, miR-141 and miR-200c) or empty vector (EV) pLX301 expressing EGFP only. We noted that Jurkat-miR-200c/141 cells showed a 10-fold overexpression of miR-141 and a 5-fold overexpression of miR-200c compared to EV-transduced control Jurkat cells. We also observed a downregulation of both ZEB2 (P=0.013) and TGFbR3 mRNA (P<0.1) in Jurkat-miR-200c/141 cells compared to EV-transduced control Jurkat cells (Figure 3F). Ligand stimulated TGFb type II receptor (TGFbRII) interacts with and phosphorylates TGFbRI, which results in activation of its kinase domain. This allows the TGFbR1 to phosphorylate SMAD2 and SMAD3 transcription factors. TGFbRIII act as a regulator of TGFb signaling and SMAD pathway activation, which is highly contextdependent.32-34 In order to further investigate the role of miR-200c/141 in the regulation of TGFb-signaling, we transduced HeLa cells with either miR-200c/141 or EGFP only (EV) lentiviruses. HeLa-miR-200c/141 cells had a 397fold miR-200c and a 624-fold miR-141 overexpression compared to HeLa-EV control cells (Figure 4A). Downregulation of ZEB2 and TGFbR3 mRNA levels in HeLa-miR-200c/141 cells was confirmed by quantitative polymerase chain reaction (qPCR) (Figure 4B). Interestingly, hu-TGFb1 stimulation of HeLa-miR200c/141 cells, resulted in increased p-SMAD3 levels, whereas p-SMAD2 was reduced and SMAD2, SMAD3 and TGFbR1 levels remained similar to their expression in control HeLa cells (Figure 4C). Next, we investigated huTGFb1-induced SMAD2 and SMAD3 phosphorylation in T-PLL. Notably, the expression of SMAD3 relative to SMAD2 was higher in T-PLL compared to PBMC and HeLa (Figure 4C and D). In addition, we observed a decreased p-SMAD2 and p-SMAD3 induction after huTGFb1 stimulation compared to the unstimulated condition in T-PLL with high miR-200c/141 compared to T-PLL with normal levels of miR-200c/141 and healthy PBMC (Figure 4D and E). Together, our results indicate that the miR-200c/141-ZEB2-TGFb- axis is aberrant in T-PLL.

Discussion In this study, we characterized a cohort of T-PLL patients both phenotypically and molecularly. In full haematologica | 2022; 107(1)

agreement with recently published data,2 we found multiple oncogenic abnormalities in our T-PLL cohort, such as aberrations in ATM, TP53 and high TCL1 overexpression. We did not observe the clinical and morphological heterogeneity of T-PLL due to differences in protein-encoding gene expression profiles as reported in previous publications.2,35 This discrepancy can be largely explained by the relatively small cohort of T-PLL patients used for this study. Furthermore, we found that T-PLL generally displayed immunological characteristics mostly comparable to memory T cells, without evidence for identical TCR usage pointing towards a common antigen-driven leukemogenesis. Based on miRNA expression, T-PLL cells were most similar to effector T cells. Our study reports an abnormal expression of miRNA in T-PLL. In total, we identified a set of 35 aberrantly expressed miRNA, that were upregulated compared to normal effector CD4 T cells. Although most T-PLL cases showed a relatively homogenous gene expression profile, T-PLL samples exhibited differential miRNA expression that correlated with blood cell counts and survival. The observation that differentially expressed miRNA are mostly upregulated in T-PLL is striking, since global miRNA depletion is more commonly found in human cancer.14 However, we cannot rule out that some miRNA are actually downregulated but with high degree of variation, such that they are not picked up in this small cohort of T-PLL cases. On the other hand, involvement of enhanced miRNA expression was already postulated based on elevated AGO2 expression as a result of chromosome 8 gains in a subset of T-PLL patients.2 Increased AGO2 expression has been detected in some human cancer types and it is well-established that this results in enhanced miRNA stability and expression levels.36 As we did not find a general AGO2 upregulation in our cohort of T-PLL cases, it remains to be determined why miRNA are upregulated in T-PLL. Some upregulated miRNA in T-PLL concern wellknown oncogenic miRNA, including the oncogenic driver miR-19b, which represses expression of the tumor suppressor gene PTEN37 and miR-125a, which strongly represses the expression of genes involved in apoptosis, such as TP53, PUMA and BAK1.38 Expression of miR-200c/141 is strongly upregulated in a subset of T-PLL. The miR-200c/141 cluster is frequently co-expressed with PTPN6 through transcriptional readthrough and complex 3D chromatin interactions between PTPN6 and miR-200c/141 promoters.28 Accordingly, we found a significant upregulation of PTPN6 in T-PLL samples with enhanced miR-200c/141 expression. It is wellestablished that the expression of miR-200c/141 is strongly induced by oxidative stress.39 As reactive oxygen species (ROS) levels are dramatically increased in T-PLL cells,2 this may therefore largely explain the elevated expression levels of miR-200c/141. Furthermore, MIR200C/141 is known to be transcriptionally upregulated by MYC,40,41 an oncogenic transcription factor that is frequently overexpressed in T-PLL, including our cohort.2,42 In our search for the biological implications of miR200c/141 upregulation, we identified a potential role for the miR-200c/141-ZEB2-TGFbR axis in the pathogenesis of T-PLL. The TGFb-signaling pathway regulates critical cellular processes in hematopoietic cells such as proliferation, differentiation, apoptosis and cell migration. Leukemia cells are commonly resistant to TGFb-controlled mechanisms due to mutations and deletion of mol151

S.J. Erkeland et al.

ecules that are active in the TGFb pathway or due to disruption of the network by oncoproteins.43 We determined miR-200c and miR-141-mediated downregulation of TGFbR3 transcripts in HeLa cells. TGFbR3 is a contextdependent regulator of TGFb signaling and a potent tumor suppressor that is frequently inactivated in various types of cancer.32,34,44 In agreement with the gene expression data, we observed reduced TGFbR3 membrane expression on T-PLL cells. Although the levels of TGFbR1 mRNA and protein were unaffected, we did observe reduced TGFbR1 membrane expression on T-PLL cells. Reduced TGFbR3 expression in HeLa did not affect SMAD2/3 levels, though we noted an aberrant p-SMAD2 and pSMAD3 level. Furthermore, we found that T-PLL samples with high miR-200c/141 levels have decreased p-SMAD2 and p-SMAD3 levels. Aberrant p-SMAD3 plays critical roles in stemness of tumor cells and progression of malignancies.45,46 Although this would suggest an important role of p-SMAD proteins in T-PLL leukemogenesis, the oncogenic downstream functions of aberrant TGFbR3 expression and p-SMAD in T-PLL remain to be further investigated. ZEB1 and ZEB2 are highly related SMAD-interacting transcription factors and important players in the induction of the epithelial–mesenchymal transition, which promotes invasion of tumor cells into the surrounding tissues, drug resistance, cell survival and metastasis.47 ZEB2 binds to SMAD2/3 and to SMAD1/5/8 in ligand-stimulated cells.31,48 ZEB proteins are involved in transcriptional repression by interacting with co-repressor protein CtBP and the NuRD chromatin-remodeling corepressor complex and are transcriptional activators by binding to P300/PCAF.49 ZEB1 and ZEB2 contain multiple miR-200c/141 target sequences in their 3’-UTR.30 Endogenous miRNA levels strongly repress ZEB1 and/or ZEB2 expression in a cell type-dependent fashion. For instance, during EMT miR-200 family members repress translation of both ZEB1 and ZEB2, whereas in CD8+ T cells only ZEB2 levels are downregulated, a process that is presumably controlled by RNA-binding proteins (RBP).30 Overexpression of miR-200 in CD8+ T cells selectively repressed ZEB2 expression, but not ZEB1, and promoted memory T-cell development.30 Interestingly, we found strong ZEB2 downregulation in T-PLL, whereas ZEB1 expression was not affected. The mechanism behind the selective downregulation of ZEB2 in T-PLL cells most-likely involves RBP.30 However, the RBP involved in selective ZEB expression regulation remain to be identified. How downregulation of ZEB2 and the above discussed aberrant p-SMAD3 levels affect target gene expression in T-PLL and how this contributes to oncogenic transformation remains to be unraveled. We also found a strong upregulation of miR-181a and miR-181b in a subset of T-PLL samples. Upregulation of miR-181 is reported in human cancer, e.g., breast cancer and chronic lymphocytic leukemia (CLL).50-52 Although miR-181 is frequently downregulated in CLL as a whole, enhanced expression of miR-181a is associated with disease progression in trisomy 12 CLL cases.52 However, as this chromosomal aberration is not found in T-PLL,2 miR181 is presumably upregulated by a different mechanism. For instance, there is strong evidence that stress-induced

152

STAT3 enhances miR-181 expression in tumor cells,53 a signaling cascade that is commonly activated in T-PLL.2 Enhanced miR-181 expression may have oncogenic transforming functions in T-PLL by affecting multiple pathways. Ectopic expression of miR-181a in normal mouse CD4+ T cells augments the strength and sensitivity of TCR signaling to agonists.54 In normal T-cell development miR-181 controls expression of various phosphatases, such as SHP2, PTPN22, DUSP5, DUSP6 and NRARP, thereby enhancing NOTCH and TCR signaling.29 In agreement, we found a significant downregulation of DUSP5 mRNA expression in T-PLL cells, which may be caused by miR-181-mediated repression. The importance of miR-181 for T-cell leukemogenesis is further demonstrated by an experiment in which genomic deletion of MIR181A1/B1 specifically inhibits NOTCH-induced leukemogenesis in mice through repression of downstream feedback mechanisms of NOTCH and pre-TCR signaling.29 Thus, overexpression of miR-181a1/b1 may contribute to enhanced NOTCH and TCR signaling, thereby enhancing oncogenesis in T-PLL. In conclusion, we have shown that based on miRNA expression T-PLL is most similar to effector T-cell subsets albeit with some degree of heterogeneity between T-PLL cases. A set of 35 miRNA is aberrantly expressed in T-PLL, with miR-200c/141 being the most upregulated cluster. Aberrant expression of miR-200c/141 affects TGFb-controlled mechanisms that may contribute to the pathogenesis of T-PLL. Additionally, the extent of overexpression of miR-200c and miR-141 appeared to significantly correlate with increased white blood cell counts and poor survival and may thus have prognostic value. Our data highlight a potential role for aberrantly expressed oncogenic miRNA in the pathogenesis of T-PLL, which may pave the way for new therapeutic approaches in this disease. Disclosures No conflicts of interest to disclose. Contributions SJE and AWL are corresponding authors; SJE, AWL, KL and VHJV designed the study; SJE, AWL and CJS wrote the manuscript; MCKK collected clinical data; KL performed cytogenetics and morphology analyses; JSV performed gene expression profiling, T-cell receptor gene rearrangement analysis, immuno-phenotyping and FACsorting; GDC performed western blotting; CJS and YMM performed flowcytometry and miRNA experiments; IVZ, CJS, AHB, performed molecular cloning, cDNA library preparations and miRNA expression experiments; EMB performed next-generation sequencing; HVW, LGL and DR performed bioinformatic analyses. Acknowledgments We thank Steven Koetzier, Fabiënne van Opstal, Bernard Stikker and Stijn van den Broek for technical assistance during their bachelor and master projects. Funding This work was supported an unrestricted grant from Bayer, Mijdrecht, the Netherlands (to MCKK, KvL, and AWL). CJS and IVZ are supported by Dutch Cancer Society (KWF), grant number: 10948.

haematologica | 2022; 107(1)

Oncogenic microRNA in T prolymphocytic leukemia

References 1. Matutes E, Brito-Babapulle V, Swansbury J, et al. Clinical and laboratory features of 78 cases of T-prolymphocytic leukemia. Blood. 1991;78(12):3269-3274. 2. Schrader A, Crispatzu G, Oberbeck S, et al. Actionable perturbations of damage responses by TCL1/ATM and epigenetic lesions form the basis of T-PLL. Nat Commun. 2018;9(1):697. 3. Stengel A, Kern W, Zenger M, et al. Genetic characterization of T-PLL reveals two major biologic subgroups and JAK3 mutations as prognostic marker. Genes Chromosomes Cancer. 2016;55(1):82-94. 4. Johansson P, Klein-Hitpass L, Choidas A, et al. SAMHD1 is recurrently mutated in T-cell prolymphocytic leukemia. Blood Cancer J. 2018;8(1):11. 5. Maljaei SH, Brito-Babapulle V, Hiorns LR, Catovsky D. Abnormalities of chromosomes 8, 11, 14, and X in T-prolymphocytic leukemia studied by fluorescence in situ hybridization. Cancer Genet Cytogenet. 1998;103(2):110-116. 6. Virgilio L, Narducci MG, Isobe M, et al. Identification of the TCL1 gene involved in T-cell malignancies. Proc Natl Acad Sci U S A. 1994;91(26):12530-12534. 7. Herling M, Patel KA, Teitell MA, et al. High TCL1 expression and intact T-cell receptor signaling define a hyperproliferative subset of T-cell prolymphocytic leukemia. Blood. 2008; ì111(1):328-337. 8. Gritti C, Dastot H, Soulier J, et al. Transgenic mice for MTCP1 develop T-cell prolymphocytic leukemia. Blood. 1998;92(2):368-373. 9. Pekarsky Y, Drusco A, Kumchala P, Croce CM, Zanesi N. The long journey of TCL1 transgenic mice: lessons learned in the last 15 years. Gene Expr. 2015;16(3):129-135. 10. Virgilio L, Lazzeri C, Bichi R, et al. Deregulated expression of TCL1 causes T cell leukemia in mice. Proc Natl Acad Sci U S A. 1998;95(7):3885-3889. 11. Zanesi N, Balatti V, Riordan J, et al. A Sleeping Beauty screen reveals NF-kB activation in CLL mouse model. Blood. 2013;121(21):4355-4358. 12. Bartel DP. Metazoan MicroRNAs. Cell. 2018;173(1):20-51. 13. Jongen-Lavrencic M, Sun SM, Dijkstra MK, Valk PJ, Lowenberg B. MicroRNA expression profiling in relation to the genetic heterogeneity of acute myeloid leukemia. Blood. 2008;111(10):5078-5085. 14. Lu J, Getz G, Miska EA, et al. MicroRNA expression profiles classify human cancers. Nature. 2005;435(7043):834-838. 15. Fernandes Q. MicroRNA: defining a new niche in Leukemia. Blood Rev. 2017; 31(3):129-138. 16. Yeh CH, Moles R, Nicot C. Clinical significance of microRNAs in chronic and acute human leukemia. Mol Cancer. 2016;15(1):37. 17. Stavast CJ, Erkeland SJ. The non-canonical aspects of microRNAs: many roads to gene regulation. Cells. 2019;8(11):1465. 18. Krowiorz K, Ruschmann J, Lai C, et al. MiR139-5p is a potent tumor suppressor in adult acute myeloid leukemia. Blood Cancer J. 2016;6(12):e508. 19. Alemdehy MF, Haanstra JR, de Looper HW, et al. ICL-induced miR139-3p and miR199a3p have opposite roles in hematopoietic cell expansion and leukemic transformation. Blood. 2015;125(25):3937-3948. 20. Wojtowicz EE, Lechman ER, Hermans KG, et al. Ectopic miR-125a expression induces

haematologica | 2022; 107(1)

long-term repopulating stem cell capacity in mouse and human hematopoietic progenitors. Cell Stem Cell. 2016;19(3):383-396. 21. Cimmino A, Calin GA, Fabbri M, et al. miR15 and miR-16 induce apoptosis by targeting BCL2. Proc Natl Acad Sci U S A. 2005;102(39):13944-13949. 22. Stavast CJ, Leenen PJM, Erkeland SJ. The interplay between critical transcription factors and microRNAs in the control of normal and malignant myelopoiesis. Cancer Lett. 2018;427:28-37. 23. Laribi K, Lemaire P, Sandrini J, Baugier de Materre A. Advances in the understanding and management of T-cell prolymphocytic leukemia. Oncotarget. 2017;8(61):104664104686. 24. van Dongen JJ, Lhermitte L, Bottcher S, et al. EuroFlow antibody panels for standardized n-dimensional flow cytometric immunophenotyping of normal, reactive and malignant leukocytes. Leukemia. 2012;26(9):1908-1975. 25. Garrido P, Ruiz-Cabello F, Barcena P, et al. Monoclonal TCR-Vbeta13.1+/CD4+/NKa+/ CD8-/+dim T-LGL lymphocytosis: evidence for an antigen-driven chronic T-cell stimulation origin. Blood. 2007;109(11):4890-4898. 26. Sandberg Y, Kallemeijn MJ, Dik WA, et al. Lack of common TCRA and TCRB clonotypes in CD8(+)/TCRalphabeta(+) T-cell large granular lymphocyte leukemia: a review on the role of antigenic selection in the immunopathogenesis of CD8(+) T-LGL. Blood Cancer J. 2014;4(1):e172. 27. Kotrova M, Novakova M, Oberbeck S, et al. Next-generation amplicon TRB locus sequencing can overcome limitations of flow-cytometric Vbeta expression analysis and confirms clonality in all T-cell prolymphocytic leukemia cases. Cytometry A. 2018; 93(11):1118-1124. 28. Batista L, Bourachot B, Mateescu B, Reyal F, Mechta-Grigoriou F. Regulation of miR200c/141 expression by intergenic DNAlooping and transcriptional read-through. Nat Commun. 2016;7:8959. 29. Fragoso R, Mao T, Wang S, et al. Modulating the strength and threshold of NOTCH oncogenic signals by mir-181a-1/b-1. PLoS Genet. 2012;8(8):e1002855. 30. Guan T, Dominguez CX, Amezquita RA, et al. ZEB1, ZEB2, and the miR-200 family form a counterregulatory network to regulate CD8(+) T cell fates. J Exp Med. 2018; 215(4):1153-1168. 31. Verschueren K, Remacle JE, Collart C, et al. SIP1, a novel zinc finger/homeodomain repressor, interacts with Smad proteins and binds to 5'-CACCT sequences in candidate target genes. J Biol Chem. 1999; 274(29):20489-20498. 32. Vander Ark A, Cao J, Li X. TGF-beta receptors: In and beyond TGF-beta signaling. Cell Signal. 2018;52:112-120. 33. Bilandzic M, Stenvers KL. Betaglycan: a multifunctional accessory. Mol Cell Endocrinol. 2011;339(1-2):180-189. 34. Gatza CE, Oh SY, Blobe GC. Roles for the type III TGF-beta receptor in human cancer. Cell Signal. 2010;22(8):1163-1174. 35. Durig J, Bug S, Klein-Hitpass L, et al. Combined single nucleotide polymorphismbased genomic mapping and global gene expression profiling identifies novel chromosomal imbalances, mechanisms and candidate genes important in the pathogenesis of T-cell prolymphocytic leukemia with inv(14)(q11q32). Leukemia. 2007; 21(10): 2153-2163. 36. Diederichs S, Haber DA. Dual role for arg-

onautes in microRNA processing and posttranscriptional regulation of microRNA expression. Cell. 2007;131(6):1097-1108. 37. Han YC, Vidigal JA, Mu P, et al. An allelic series of miR-17 approximately 92-mutant mice uncovers functional specialization and cooperation among members of a microRNA polycistron. Nat Genet. 2015;47(7):766-775. 38. Shaham L, Binder V, Gefen N, Borkhardt A, Izraeli S. MiR-125 in normal and malignant hematopoiesis. Leukemia. 2012;26(9):20112018. 39. Mateescu B, Batista L, Cardon M, et al. miR141 and miR-200a act on ovarian tumorigenesis by controlling oxidative stress response. Nat Med. 2011;17(12):1627-1635. 40. Chen P, Guo X, Zhang L, et al. MiR-200c is a cMyc-activated miRNA that promotes nasopharyngeal carcinoma by downregulating PTEN. Oncotarget. 2017;8(3):5206-5218. 41. Lin CH, Jackson AL, Guo J, Linsley PS, Eisenman RN. Myc-regulated microRNAs attenuate embryonic stem cell differentiation. EMBO J. 2009;28(20):3157-3170. 42. Hsi AC, Robirds DH, Luo J, et al. T-cell prolymphocytic leukemia frequently shows cutaneous involvement and is associated with gains of MYC, loss of ATM, and TCL1A rearrangement. Am J Surg Pathol. 2014; 38(11):1468-1483. 43. Dong M, Blobe GC. Role of transforming growth factor-beta in hematologic malignancies. Blood. 2006;107(12):4589-4596. 44. Bernabeu C, Lopez-Novoa JM, Quintanilla M. The emerging role of TGF-beta superfamily coreceptors in cancer. Biochim Biophys Acta. 2009;1792(10):954-973. 45. Tarasewicz E, Jeruss JS. Phospho-specific Smad3 signaling: impact on breast oncogenesis. Cell Cycle. 2012;11(13):2443-2451. 46. Naka K, Jomen Y, Ishihara K, et al. Dipeptide species regulate p38MAPK-Smad3 signalling to maintain chronic myelogenous leukaemia stem cells. Nat Commun. 2015;6:8039. 47. Fardi M, Alivand M, Baradaran B, Farshdousti Hagh M, Solali S. The crucial role of ZEB2: From development to epithelial-tomesenchymal transition and cancer complexity. J Cell Physiol. 2019;234(9):14783-14799. 48. Conidi A, Cazzola S, Beets K, et al. Few Smad proteins and many Smad-interacting proteins yield multiple functions and action modes in TGFbeta/BMP signaling in vivo. Cytokine Growth Factor Rev. 2011;22(56):287-300. 49. Scott CL, Omilusik KD. ZEBs: Novel players in immune cell development and function. Trends Immunol. 2019;40(5):431-446. 50. Mansueto G, Forzati F, Ferraro A, et al. Identification of a new pathway for tumor progression: microRNA-181b up-regulation and CBX7 down-regulation by HMGA1 protein. Genes Cancer. 2010;1(3):210-224. 51. Strotbek M, Schmid S, Sanchez-Gonzalez I, et al. miR-181 elevates Akt signaling by cotargeting PHLPP2 and INPP4B phosphatases in luminal breast cancer. Int J Cancer. 2017; 140(10):2310-2320. 52. Visone R, Rassenti LZ, Veronese A, et al. Karyotype-specific microRNA signature in chronic lymphocytic leukemia. Blood. 2009;114(18):3872-3879. 53. Iliopoulos D, Jaeger SA, Hirsch HA, Bulyk ML, Struhl K. STAT3 activation of miR-21 and miR-181b-1 via PTEN and CYLD are part of the epigenetic switch linking inflammation to cancer. Mol Cell. 2010;39(4):493-506. 54. Li QJ, Chau J, Ebert PJ, et al. miR-181a is an intrinsic modulator of T cell sensitivity and selection. Cell. 2007;129(1):147-161.

153

ARTICLE Ferrata Storti Foundation

Hematopoiesis

Loss of Nupr1 promotes engraftment by tuning the quiescence threshold of hematopoietic stem cells via regulation of the p53-checkpoint pathway Tongjie Wang,1,2,3,* Chengxiang Xia,1,3,* Qitong Weng,1 Kaitao Wang,1 Yong Dong,1 Sha Hao,4 Fang Dong,4 Xiaofei Liu,1 Lijuan Liu,1 Yang Geng,1 Yuxian Guan,1 Juan Du,1 Tao Cheng,4 Hui Cheng4 and Jinyong Wang1,2,3

Haematologica 2022 Volume 107(1):154-166

State Key Laboratory of Experimental Hematology, CAS Key Laboratory of Regenerative Biology, Guangzhou Institutes of Biomedicine and Health, Chinese Academy of Sciences, Guangzhou; 2Bioland Laboratory (Guangzhou Regenerative Medicine and Health Guangdong Laboratory), Guangzhou; 3Guangdong Provincial Key Laboratory of Stem Cell and Regenerative Medicine, Guangzhou Institutes of Biomedicine and Health, Chinese Academy of Sciences, Guangzhou and 4State Key Laboratory of Experimental Hematology & National Clinical Research Center for Blood Diseases, Institute of Hematology & Blood Diseases Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Tianjin, China 1

*TJW and CXX contributed equally as co-first authors.

ABSTRACT

Correspondence: JINYONG WANG wang_jinyong@gibh.ac.cn HUI CHENG chenghui@ihcams.ac.cn Received: September 24, 2019. Accepted: November 25, 2020. Pre-published: December 10, 2020. https://doi.org/10.3324/haematol.2019.239186

154

ematopoietic stem cells (HSC) are dominantly quiescent under homeostasis, which is a key mechanism of maintaining the HSC pool for life-long hematopoiesis. Dormant HSC are poised to be immediately activated in certain conditions and can return to quiescence after homeostasis has been regained. At present, the molecular networks of regulating the threshold of HSC dormancy, if existing, remain largely unknown. Here, we show that deletion of Nupr1, a gene preferentially expressed in HSC, activated quiescent HSC under homeostasis, which conferred a competitive engraftment advantage for these HSC without compromising their stemness or multi-lineage differentiation capacity in serial transplantation settings. Following an expansion protocol, the Nupr1-/- HSC proliferated more robustly than their wild-type counterparts in vitro. Nupr1 inhibits the expression of p53 and rescue of this inhibition offsets the engraftment advantage. Our data reveal a new role for Nupr1 as a regulator of HSC quiescence, which provides insights for accelerating the engraftment efficacy of HSC transplantation by targeting the HSC quiescence-controlling network.

Introduction Hematopoietic stem cells (HSC), the seeds of the adult blood system, generate all the blood lineages via hierarchical hematopoiesis. Under steady-state, the majority of HSC are maintained in quiescence, providing a pool of HSC for life-long hematopoiesis.1 However, the dormant HSC can be rapidly activated for stress hematopoiesis in emergency conditions, such as excessive blood loss, radiation injury, and chemotherapy damage.2 Mounting evidence points to the existence of an intrinsic molecular machinery of regulating HSC dormancy. In haploinsufficient Gata2+/- mice, there is a slight increase of quiescent HSC in conditions of homeostasis.3 Dnmt3a-knockout HSC have increased self-renewal ability and their number in the bone marrow is increased.4,5 JunB inactivation deregulates the cell-cycle machinery and reduces quiescent HSC.6 Hif-1a-deficient mice also show a decrease in dormant HSC.7 Conditional knockout of cylindromatosis (CYLD) induces dormant HSC to exit quiescence and abrogates their repopulating and self-renewal potential.8 CDK6, a protein not expressed in long-term HSC but present in short-term HSC, regulates exit from quiescence in human HSC, and overexpression of this protein promotes engraftment.9 Nevertheless, the underlying signaling regulatory network of HSC quiescence remains largely unknown.

haematologica | 2022; 107(1)

Nupr1 regulates the quiescence threshold of HSC

Nuclear protein transcription regulator 1 (NUPR1) is a member of the high-mobility group of proteins, which was first discovered in the rat pancreas during the acute phase of pancreatitis and was initially called p8.10 The same gene was discovered in breast cancer and was named Com1.11 NUPR1 has various roles, being involved in apoptosis, stress response, and cancer progression, depending on distinct cellular contexts. In certain cancers, such as breast cancer, NUPR1 inhibits tumor cell apoptosis and induces tumor establishment and progression.12-15 In stark contrast, in prostate cancer and pancreatic cancer, NUPR1 has an inhibitory effect on tumor growth.16,17 There is accumulating evidence that NUPR1 is a stress-induced protein: interference of NUPR1 can upregulate the sensitivity of astrocytes to oxidative stress;18 loss of it can promote resistance of fibroblasts to adriamycin-induced apoptosis;19 NUPR1 mediates cannabinoid-induced apoptosis of tumor cells;20 and overexpression of NUPR1 can negatively regulate MSL1-dependent histone acetyltransferase activity in Hela cells, which induces chromatin remodeling and relaxation allowing access of the repair machinery to DNA.21 Nonetheless, the potential roles of Nupr1, which is preferentially expressed in HSC among the hematopoietic stem and progenitor cells, in hematopoiesis remain elusive. NUPR1 interacts with p53 to regulate cell cycle and apoptosis responding to stress in breast epithelial cells.19,22 p53 plays several roles in homeostasis, proliferation, stress, apoptosis, and aging of hematopoietic cells.23-27 Deletion of p53 upregulates HSC self-renewal but impairs the repopulating ability of these cells and leads to tumors.28 Hyperactive expression of p53 in HSC decreases the size of the HSC pool, and reduces engraftment and deep quiescence.29-31 These findings support the essential check-point role of p53 in regulating HSC fate. Nonetheless, it is unknown whether NUPR1 and p53 coordinately regulate the quiescence of HSC. Here, we used a Nupr1 conditional knockout model to investigate the consequences of loss of function of Nupr1 in the context of HSC. Nupr1 deletion in HSC led to the cells exiting from quiescence under homeostasis. In a competitive repopulation setting, Nupr1-deleted HSC proliferated robustly and showed dominant engraftment over their wild-type counterparts. Nupr1-deleted HSC also expanded abundantly and preserved their stemness in vitro. The rescued expression of p53 by Mdm2+/- offset the effects introduced by loss of Nupr1 in HSC. Our studies reveal a new role and signaling mechanism of Nupr1 in regulating the quiescence of HSC.

Methods Mice Animals were housed in the animal facility of the Guangzhou Institutes of Biomedicine and Health (GIBH). Nupr1fl/fl mice were provided by Beijing Biocytogen Co., Ltd. CD45.1, Vav-cre, Mx1cre, and Mdm2+/- mice were purchased from the Jackson Laboratory. All the mouse lines were maintained on a pure C57BL/6 genetic background. All experiments were conducted in accordance with experimental protocols approved by the Animal Ethics Committee of GIBH.

Hematopoietic stem cell cycle analysis We first labeled the HSC with (CD2, CD3, CD4, CD8, Ter119, B220, Gr1, CD48)-Alexa Fluor700, Sca1-Percp-cy5.5, c-

haematologica | 2022; 107(1)

kit-APC-cy7, CD150-PE-cy7, CD34-FITC and CD135-PE. The cells were then fixed using 4% paraformaldehyde. After washing, the fixed cells were permeabilized with 0.1% saponin in phosphate-buffered saline together with Ki-67-APC staining for 45 min. Finally, the cells were resuspended in DAPI solution for staining for 1 h. The data were analyzed using Flowjo software.

Bromodeoxyuridine incorporation assay Nupr1-/- mice and WT littermates were injected with 1 mg bromodeoxyuridine (BrdU) on day 0. They were then allowed to drink water containing BrdU (0.8 mg/mL) ad libitum. On days 3, 4, and 5 after the injection of BrdU, four mice of each group were sacrificed. The rate of BrdU incorporation was analyzed by flow cytometry according to the BD Pharmingen™ APC BrdU Flow Kit instructions.

Hematopoietic stem cell culture The HSC culture protocol has been described elsewhere.32 Briefly, 50 HSC were sorted into fibronectin (Sigma)-coated 96well U-bottomed plates directly and were cultured in F12 medium (Life Technologies), 1% insulin-transferrin- seleniumethanolamine (ITSX; Life Technologies), 10 mM HEPES (Life Technologies), 1% penicillin/streptomycin/glutamine (P/S/G; Life Technologies), 100 ng/mL mouse thrombopoietin, 10 ng/mL mouse stem cell factor and 0.1% polyvinyl alcohol (P8136, Sigma-Aldrich). Half the medium was changed every 2-3 days, by manually removing medium by pipetting and replacing it with fresh medium, as indicated.

Limiting dilution assays For limiting dilution assays,33 cells cultured for 10 days were transplanted into lethally irradiated C57BL/6-CD45.1 recipient mice, together with 2×105 CD45.1 bone-marrow competitor cells. Recipients were analyzed every 4 weeks. Limiting dilution analysis was performed using ELDA software.34 based on 1% peripheral blood multilineage chimerism as the threshold for positive engraftment.

Bone marrow competitive repopulation assay One day before bone marrow transplantation, adult C57BL/6 recipient mice (CD45.1, 8-10 weeks old) were irradiated with two doses of 4.5 Gy (RS 2000, Rad Source) at a 4hour interval. Bone marrow nucleated cells (BMNC; 2.5×105) from Nupr1-/- mice (CD45.2) and their WT (CD45.1) counterparts were mixed and injected into irradiated CD45.1 recipients by retro-orbital injection. Control BMNC (CD45.2), Mdm2+/-Nupr1-/- BMNC (CD45.2) or Mdm2+/- BMNC (CD45.2) were also mixed with the same number of competitors (CD45.1) and transplanted into recipients. For Nupr1fl/flMx1-cre transplantation, cre expression was induced through intraperitoneal injection of polyinosinic-polycytidylic acid (pI-pC, 250 mg/mouse) every other day 1 week before transplantation. The same number (2.5×105) of Nupr1fl/flMx1-cre and WT (CD45.1) BMNC were mixed and transplanted into the lethally irradiated CD45.1 recipients. Mx1-cre+ mice were taken as the experiment control. Mx1-cre and WT (CD45.1) BMNC (2.5×105) were used for the transplant control. The transplanted mice were maintained on trimethoprim-sulfamethoxazole-treated water for 2 weeks. For secondary transplantation, BMNC (1×106) were obtained from primary competitive transplanted recipients and injected into irradiated CD45.1 recipients (2 doses of 4.5 Gy, 1 day before transplantation). Donor-derived cells and hematopoietic lineages in peripheral blood were assessed monthly by flow cytometry.

155

T. Wang et al.

Results Loss of Nupr1 accelerates the turn-over rates of hematopoietic stem cells under homeostasis The majority of long-term HSC are quiescent under homeostasis, which is a key mechanism for maintaining the HSC pool for life-long steady hematopoiesis. We hypothesized that genes preferentially expressed in HSC but immediately downregulated in multipotent progenitors (MPP) might form an intrinsic regulatory network for maintaining HSC quiescence. To test our hypothesis, we explored candidate factors by RNA-sequencing analysis of sorted HSC (Lin– CD48– Sca1+ c-kit+ CD150+) and MPP (Lin– Sca1+ c-kit+ CD150–). Analysis of differentially expressed genes showed a pattern of transcription factors preferentially present in HSC, including Rorc, Hoxb5, Rarb, Gfi1b, Mllt3, and Nupr1. By literature search, we found that most of the candidate genes except Nupr1 were reportedly not involved in regulating HSC homeostasis. Thus, we focused on the Nupr1 gene, the role of which in hematopoiesis has not been reported. The expression of Nupr1 in HSC was significantly higher (>25-fold, P=0.002) than that in MPP (Figure 1A, left). Real-time polymerase chain reaction (PCR) analysis further confirmed the same expression pattern (P<0.001), indicating an unknown role for Nupr1 in HSC (Figure 1A, right). To study whether Nupr1 has any potential impact on the hematopoiesis of HSC, we created Nupr1 conditional knockout mice by introducing two loxp elements flanking exons 1 and 2 of the Nupr1 locus using a C57BL/6 background mESC line (Figure 1B). The resultant Nupr1fl/fl mice were further crossed to Vav-Cre mice to generate Nupr1fl/fl; Vav-Cre compound mice (Nupr1-/- mice). The deletion of Nupr1 was confirmed by PCR in HSC (Online Supplementary Figure S1A-C). Adult Nupr1-/- mice (8-10 weeks old) had a normal percentage of blood lineage cells in peripheral blood, including CD11b+ myeloid, CD19+ B, and CD90.2+ T lineage cells (Online Supplementary Figure S2). We further investigated the potential alterations of HSC homeostasis in the absence of Nupr1. Flow cytometry analysis demonstrated that the Nupr1-/- HSC pool was comparable to the wild-type counterpart in terms of ratios and absolute numbers (Online Supplementary Figure S3). Subsequently, we examined the cell cycle status of Nupr1-/HSC using the proliferation marker Ki-67 and DAPI staining and found that the ratio of Nupr1-/- HSC in G0-status was reduced significantly (P<0.001). Compared with WT HSC (median value: Nupr1-/- HSC =73.67%, WT HSC = 87.15%), more Nupr1-/- HSC entered G1-S-S2 and M phases (Figure 1C, D). To further confirm this novel phenotype, we performed a BrdU incorporation assay, which is conventionally used to assess the turn-over rates of blood cells in vivo.35 The 8-week-old Nupr1-/- mice and littermates were injected intraperitoneally with 1 mg BrdU on day 0, followed by continuous administration of BrdU via water (0.8 mg/mL) for up to 5 days (Figure 1E). After 3 days of BrdU labeling, ~50% of Nupr1-/- HSC became BrdU+ compared with ~35% of WT HSC. The BrdU incorporation rates in HSC differed between the two mouse models (WT and Nupr1-/-, P<0.001), and the dynamics changed along with time elapsed (P=0.012, two-way analysis of variance [ANOVA]). Kinetic analysis of BrdU incorporation from day 3 to day 5 revealed that Nupr1-/- HSC contained a 1.5fold larger BrdU+ population over WT HSC (Figure 1F, G). Collectively, these data indicate that the Nupr1-deletion 156

drives HSC to enter the cell cycle and accelerates their turnover rates in homeostasis.

Nupr1-/- hematopoietic stem cells show repopulating advantage without their multilineage differentiation potential being compromised To confirm whether Nupr1-/- HSC have a repopulating advantage or disadvantage in vivo, we performed a typical HSC competitive repopulation assay. BMNC (2.5x105) from Nupr1-/- mice (CD45.2) were transplanted into lethally irradiated recipients (CD45.1) along with the same number of WT (CD45.1) competitor cells. Bone marrow cells from littermates (Vav-Cre+, CD45.2+) were mixed with WT (CD45.1) competitors and transplanted into the recipients as the experiment control (Figure 2A). Sixteen weeks later, 1x106 BMNC from the primary recipients were transplanted into lethally irradiated recipients to assess long-term engraftment. We observed that donor Nupr1-/- cells accounted for ~70% of cells in the primary recipients, while the control cells accounted for 50%-60% in the recipients from the transplantation control assay. Nupr1-/- cells gradually dominated in peripheral blood of recipients over time after transplantation (Figure 2B). In the chimeras, ~70% of myeloid cells and B lymphocytes were Nupr1-/- donorderived cells, while ~60% of T lymphocytes were from CD45.1 competitive cells (Figure 2C). To further explore whether Nupr1-/- HSC dominated in the HSC pool, we sacrificed the recipients and analyzed HSC 16 weeks after transplantation. The proportion and absolute number of Nupr1-/- HSC were significantly higher (~1.5-fold) than those of the control HSC in primary recipients (Figure 2D, E). Previous research documented that HSC proliferated rapidly at the expense of their long-term repopulating ability.36-40 Interestingly, consistent with the dominating trend in the primary transplants, Nupr1-/- cells continuously dominated in the peripheral blood of secondary recipients (Figure 3A). Nupr1-/- HSC further occupied up to 90% of the total HSC in the bone marrow of secondary recipients. However, the control HSC accounted for less than 10% in the secondary recipients (Figure 3B, C). Collectively, these results indicate that the deletion of Nupr1 promotes the repopulating ability of HSC without impairing their longterm engraftment ability.

Nupr1-/- hematopoietic stem cells are highly sensitive to irradiation-stress but re-cover fast HSC in the cell cycle were reported to be more sensitive to irradiation damage.41,42 To explore whether Nupr1-/- HSC with a fast turn-over rate are more sensitive to irradiation, WT mice and Nupr1-/- mice were exposed to a single dose of total body irradiation (4 Gy dose, 1 Gy/min). Apoptosis and cell cycle status were analyzed 6 h (early stage) and 24 h (later stage) later. As expected, Nupr1-/- HSC showed a significantly enhanced sensitivity to irradiation: only ~40% of Nupr1-/- HSC lived 6 h after irradiation, whereas ~70% of WT HSC were still alive (Online Supplementary Figure S4A). The proportion of radiation-induced apoptosis (annexin V+) of Nupr1-/- HSC was significantly higher (~2-fold, P=0.02) than that of WT HSC (Online Supplementary Figure S4A). Furthermore, ~60% of the residual Nupr1-/- HSC were in G1-S-G2-M proliferative phases compared with ~50% of the residual WT HSC (P=0.01) (Online Supplementary Figure S4B), indicating an accelerated replenishment rate in response to irradiation damage. At a later stage (24 h) after irradiation, we observed more living Nupr1-/- HSC (WT vs. haematologica | 2022; 107(1)

Nupr1 regulates the quiescence threshold of HSC

Figure 1. Loss of Nupr1 activates hematopoietic stem cells that are dormant during homeostasis. (A) Expression pattern of Nupr1 in hematopoietic stem cells (HSC) and multipotent progenitors (MPP) examined by RNA-sequencing and real-time polymerase chain reaction (qPCR). One thousand HSC or MPP from bone marrow of wild-type (WT) mice were sorted as individual samples for RNA-sequencing (n=4). HSC were defined as Lin– (i.e., CD2–, CD3–, CD4–, CD8–, Mac1–, Gr1–, Ter119–, B220–), CD48–, Sca1+, c-kit+, and CD150+. MPP were defined as Lin–, Sca1+, c-kit+, and CD150–. Data were analyzed using an unpaired Student t-test (twotailed) and are represented as mean ± standard deviation (SD) (qPCR, n=3 mice for each group). **P<0.01, ***P<0.001. FPKM: fragments per kilobase of exon per million mapped reads. (B) Targeting strategy for the knockout of the Nupr1 gene in mice. WT Nupr1 exons 1 and 2 are shown as green boxes. Two loxp elements flanking exon 1 and exon 2 were inserted. (C) Cell cycle analysis of Nupr1-/- HSC under homeostasis. Representative plots of cell cycle from representative WT and Nupr1-/- mice (8 weeks old). WT littermates (8 weeks old) were used as controls. HSC (Lin– CD48– Sca1+ c-kit+ CD150+ CD34– CD135–) were analyzed by DNA content (DAPI) versus Ki-67: G0 (Ki-67lowDAPI2N), G1 (Ki-67highDAPI2N), G2-S-M (Ki-67highDAPI>2N-4N). (D) Statistical analysis of the HSC cycle. The percentages (%) of HSC in G0 and in G1-G2-S-M stages were analyzed. Data were analyzed using an unpaired Student t-test (two-tailed) and are represented as mean ± SD (n=6 mice for each group). **P<0.01. (E) The strategy of the BrdU incorporation assay. The 8-week-old Nupr1-/- mice and littermates were injected intraperitoneally with 1 mg BrdU on day 0. The mice were then allowed to drink BrdU (0.8 mg/mL) water ad libitum until analyzed on days 3, 4, and 5. (F) Dynamic analysis of BrdU+ HSC after BrdU administration, as determined by flow cytometry on days 3, 4, and 5. (G) Kinetics of the BrdU+ HSC ratio. Data were analyzed using an unpaired Student t-test (twotailed) and two-way analysis of variance and are represented as mean ± SD (n=4 mice for each group). *P<0.05, **P<0.01, ***P<0.001.

haematologica | 2022; 107(1)

157

T. Wang et al.

Nupr1-/-: 74% vs. 86%) and less irradiation-induced apoptotic Nupr1-/- HSC compared with WT HSC (Online Supplementary Figure S4C). The proportion of cycling cells (G1-S-G2-M proliferative phases) in the residual Nupr1-/HSC was still significantly higher (P<0.001) than the WT

HSC 24 h after irradiation (WT vs. Nupr1-/-: 29% vs. 48%) (Online Supplementary Figure S4D). Thus, Nupr1-/- HSC were susceptible to irradiation-induced damage, but the surviving HSC proliferated, resulting in a fast recover of the HSC pool.

Figure 2. Nupr1-/- hematopoietic stem cells show a repopulating advantage in competitive transplantation experiments. (A) Schematic diagram of the competitive transplantation assay. Bone marrow nucleated cells (BMNC; 2.5x105) from Nupr1-/- (CD45.2) or littermate control (Vav-cre+, CD45.2) mice were mixed with equivalent wild-type (WT) (CD45.1) counterparts and injected into individual lethally irradiated recipients (CD45.1). Four months later, the recipients were sacrificed and 1x106 BMNC from primary transplanted recipients were transplanted into lethally irradiated secondary recipients. (B) Kinetic analysis of donor chimerism (CD45.2+) in peripheral blood (PB). Data were analyzed by two-way analysis of variance (ANOVA) and are represented as mean ± standard deviation (SD) (control group: n=5 mice; Nupr1-/- group: n=6 mice). *P<0.05, **P<0.01. (C) Kinetic analysis of donor-derived lineage chimerism in PB, including myeloid cells (CD11b+) (left), B lymphocytes (CD19+) (middle), and T lymphocytes (CD90.2+) (right). Data were analyzed using a paired Student t-test (two-tailed) and two-way ANOVA and are represented as mean ± SD (control group: n=5 mice, Nupr1-/- group: n=6 mice). *P<0.05. (D) Flow cytometry analysis of the hematopoietic stem cell (HSC) compartment in primary recipients 4 months after transplantation. Representative plots from one recipient mouse in each group are shown. (E) Cell number and percentage of donor-derived HSC in primary recipients 4 months after competitive transplantation. Data were analyzed using a Student t-test and are represented as mean ± SD (n=5) **P<0.01.

158

haematologica | 2022; 107(1)

Nupr1 regulates the quiescence threshold of HSC

Nupr1-deleted hematopoietic stem cells expand robustly in vitro We next examined whether the deletion of Nupr1 could enhance HSC expansion in vitro. Fifty HSC sorted from WT and Nupr1-/- mice were cultured in vitro for 10 days following a recently described protocol32 (Figure 4A). After 10 days of culture, the WT input cells yielded more than 2.2×104 cells, while Nupr1-/- HSC produced approximately 5×104 total cells (P<0.001) (Figure 4B). The colonies derived from Nupr1-/- HSC were much larger than those from WT HSC (Figure 4C). Furthermore, we analyzed the phenotypic HSC populations in the expanded cells and found that the absolute number of phenotypic HSC in individual Nupr1-/colonies was 3 times more than WT HSC (P=0.005) (Figure 4D, E). To determine whether the quantitative expansion of

phenotypic HSC contained net proliferation of functional HSC, we performed competitive repopulating-unit assays,33 using serial doses of limiting dilutions of the in vitro-expanded cells. The WT HSC frequency in the 10-day expanded cells was 1 in 371 cells, which is equivalent to 61 functional HSC, while the Nupr1-/- HSC frequency in the 10-day expanded cells was 1 in 190 cells (Figure 4F),34 which is equivalent to 263 functional HSC (P=0.045). Therefore, the deletion of Nupr1 induced an approximately 4-fold expansion of functional HSC numbers over WT HSC. Thus, deletion of Nupr1 enhances the expansion ability of HSC in vitro.

Reversion of p53 expression offsets the competitiveness of Nupr1-/- hematopoietic stem cells To further investigate the underlying molecular mechaFigure 3. Nupr1-/- hematopoietic stem cells continuously show a competitive advantage without losing their long-term self-renewing ability in secondary transplantation. (A) Kinetic analysis of donor chimerism (CD45.2+) in peripheral blood (PB) of secondary transplanted recipients. Data were analyzed by two-way analysis of variance and are represented as mean ± standard deviation (SD) (n=5 mice). ***P<0.001. (B) Flow cytometry analysis of donor Nupr1-/hematopoietic stem cells (HSC) in secondary recipients 16 weeks after transplantation. Representative plots from each group of mice are shown. (C) Cell number and percentage of donor-derived HSC in secondary recipients 4 months after competitive transplantation. Data were analyzed using an unpaired Student t-test (two-tailed) and are represented as mean ± SD (n=5 mice). ***P<0.001. BMNC: bone marrow nucleated cells.

haematologica | 2022; 107(1)

159

T. Wang et al.

nisms of Nupr1 in regulating HSC, we performed RNAsequencing analysis of Nupr1-/- HSC from 8-week-old Nupr1-/- mice. Analysis of gene expression indicated that there were 319 genes differentially expressed between WT

and Nupr1-/- HSC (>2-fold difference in expression; adjusted P value <0.05 [DESeq2 R package]). Gene-ontology analysis of these differentially expressed genes indicated enrichment of genes involved in regulation of mitotic cell cycle and neg-

Figure 4. Deletion of Nupr1 promotes hematopoietic stem cell expansion in vitro. (A) Schematic diagram of hematopoietic stem cell (HSC) expansion in vitro. Fifty CD150+KSL HSC from wild-type (WT) and Nupr1-/- mice were sorted into fibronectin-coated plate wells, containing albumin-free F12 medium supplemented with 1 mg/mL polyvinyl alcohol (PVA), 100 ng/mL thrombopoietin (TPO) and 10 ng/mL stem cell factor (SCF). HSC were cultured for 10 days and then analyzed by flow cytometry (FACS). For the limiting dilution assay, serial doses were transplanted into lethally irradiated recipients, together with 2×105 bone-marrow competitor cells. (B) Number of cells derived from 50 HSC after 10 days of culture in vitro. Data were analyzed using an unpaired Student t-test (two-tailed) and are represented as mean ± standard deviation (SD) (WT, n=10; Nupr1-/-, n=16). ***P<0.001. (C) Representative images of WT and Nupr1-/- HSC from freshly isolated HSC (day 0) and after 10 days of culture (day 10). Images of five representative colonies (biological replicates) are shown. (D) Representative flow cytometric plots of HSC from cultured WT and Nupr1-/- HSC at day 10. p-HSC: primary HSC from bone marrow. e-HSC: expanded HSC after 10 days of culture ex vivo. (E) Counts of phenotypic CD150+KSL HSC at day 10 after culture. The dashed line indicates the amount of the primary input cells. Data were analyzed using an unpaired Student t-test (two-tailed) and are represented as mean ± SD (WT, n=8; Nupr1-/-, n=11). **P<0.01. (F) Poisson statistical analysis after limiting-dilution analysis; plots were obtained to allow estimation of competitive repopulating units in each condition (n=10 mice transplanted at each dose per condition, *P<0.05). The plot shows the percentage of recipient mice containing less than 1% CD45.2+ cells in the peripheral blood at 16 weeks after transplantation versus the number of cells injected per mouse. *P<0.05.

160

haematologica | 2022; 107(1)

Nupr1 regulates the quiescence threshold of HSC

ative regulation of cell cycle (Online Supplementary Figure S5A, B). In addition, the positive regulatory genes of cell cycle, such as Cdk4, Cdk6, Akt1 and Akt2, were upregulated in the Nupr1-/- HSC. However, regulators of HSC quiescence, such as Gfi1, Pten, Hlf, Cdc42 and Foxo1 were downregulated in the Nupr1-/- HSC (Online Supplementary Figure S5C).43,44 Gene set enrichment analysis illustrated that genes related to p53 pathways feedback loops, including Trp53, Ccng1, Ctnnb1, Pten, and Pik3c2b, were enriched in WT HSC (Figure 5A). The p53 pathway regulates a series of target genes involving cell cycle arrest, apoptosis, senescence, DNA repair, and metabolism.45 Interestingly, the expression of p53 was significantly reduced (P<0.001) to one-third of the control value in Nupr1-/- HSC (Figure 5B). Therefore, we hypothesized that downregulation of p53 in Nupr1-/- HSC might account for the competitive advantage of the HSC. MDM2 is a ubiquitin ligase E3 for p53, which is a key repressive regulator of p53 signaling.46 Mdm2-deficient mice showed increased levels of active p53, which is an ideal substitute model of upregulating p53 since directly overexpressing p53 leads to cell death and blood malignancies in mice.27,47 Nupr1-/- mice were crossed with Mdm2+/- mice to achieve upregulation of p53 expression in Nupr1-/- HSC. As expected, the levels of p53 protein expression in Nupr1-/-

Mdm2+/- HSC were comparable to those in WT HSC (P>0.05) but significantly higher than those in Nupr1-/- HSC, as measured by indirect immunofluorescence (Figure 5C, D). In addition, most genes involved in the p53 pathway were upregulated in the Nupr1-/-Mdm2+/- HSC, indicated partial recovery of the p53 pathway (Online Supplementary Figure S6). We next examined phenotypic HSC in the Nupr1-/-Mdm2+/- mice. Flow cytometry analysis showed that the Nupr1-/-Mdm2+/- HSC pool was indistinguishable from the WT, Nupr1-/-, and Mdm2+/- counterparts in terms of ratios and absolute numbers (Figure 6A, B). Furthermore, we tested the competitiveness of Nupr1-/-Mdm2+/- HSC in parallel with WT, Nupr1-/-, and Mdm2+/- HSC. BMNC (2.5x105) from WT, Nupr1-/- Mdm2+/- mice (CD45.2), Nupr1-/- mice (CD45.2), or Mdm2+/- mice (CD45.2) were transplanted into lethally irradiated recipients (CD45.1) along with the same number of WT (CD45.1) BMNC. In the recipients of Nupr1-/-Mdm2+/donor cells, the contribution of Nupr1-/-Mdm2+/- cells was significantly reduced (P<0.001) to ~20%, which was far below the percentage of Nupr1-/- cells in recipients of Nupr1-/- donor cells, and Mdm2+/- cells accounted for less than 10% in the peripheral blood of recipients 16 weeks after transplantation (Figure 6C). Sixteen weeks after transplantation, we also analyzed the Nupr1-/-Mdm2+/- HSC in the

A B

haematologica | 2022; 107(1)

Figure 5. Loss of Nupr1 confers repopulating advantage on hematopoietic stem cells by regulating p53 check-point signaling. (A) Gene set enrichment analysis (GSEA) of p53 pathway feedback loops in wildtype (WT) hematopoietic stem cells (HSC) and Nupr1-/- HSC. One thousand HSC from the bone marrow of WT and Nupr1-/- mice were sorted as individual samples for RNA-sequencing. DESeq2 normalized values of the expression data were used for GSEA. Expression of the leading-edge gene subsets is shown. p53 pathway feedback loops that are downregulated in Nupr1-/- HSC (>1.2-fold difference in expression; adjusted P value <0.05). WT HSC, n=4 cell sample replicates (one per column); Nupr1-/- HSC, n=4 cell sample replicates (one per column). FDR: false discovery rate. (B) Expression level of p53 in WT HSC and Nupr1-/- HSC determined by RNA-sequencing. The Y-axis indicates the expression value (DESeq2 normalized values of the expression data. Data were analyzed using an unpaired Student t-test (two-tailed) and are represented as mean ± standard deviation (SD) (n=4 mice for each group). ***P<0.001. (C) Immunofluorescence measurement of p53 proteins in single HSC from WT, Nupr1-/-, Mdm2+/-Nupr1-/- and Mdm2+/- mice. Images of three representative single cells from each group are shown. (D) Mean intensity of p53 fluorescence in WT, Nupr1-/-, Mdm2+/-Nupr1-/- and Mdm2+/- HSC. Each dot represents a single cell. Data were analyzed by one-way analysis of variance and are represented as mean ± SD. WT, n=18; Nupr1-/-, Mdm2+/-Nupr1-/-, Mdm2+/-: n=25. ***P<0.001.

161

T. Wang et al.

chimeras. Surprisingly, only a few Nupr1-/-Mdm2+/- HSC were present in the HSC pool of the recipients, while the Nupr1-/- HSC dominantly occupied the HSC pool (Figure 6D, E). Overall, the reversal of p53 expression offset the competitive advantage of Nupr1-/- HSC.

Deletion of Nupr1 in adulthood in the Nupr1fl/fl Mx1-cre model also promoted hematopoietic stem cell engraftment In the Nupr1fl/fl Vav-Cre model, the Nupr1 locus was deleted at an embryonic stage. To exclude the possibility that the

effects of loss of Nupr1 observed in adulthood is a consequence of an effect coming from the embryo, Nupr1fl/fl mice were crossed with Mx1-Cre mice to generate induced Nupr1 knockout mice at an adult stage in the presence of polyinosinic-polycytidylic acid (pIpC). The deletion of the Nupr1 gene in the Nupr1fl/flMx1-cre mice was verified by PCR in HSC (Online Supplementary Figure S1D, E). Consistent with the observation of loss of Nupr1 in the Nupr1fl/fl Vav-Cre model, significantly more Nupr1fl/flMx1-cre HSC entered the G1-S-S2 and M phases (median value: 26.35%) than their counterparts from littermate control

Figure 6. Figure continued on following page.

162

haematologica | 2022; 107(1)

Nupr1 regulates the quiescence threshold of HSC

Figure 6. Reversion of p53 expression by allelic depletion of the Mdm2 gene offsets the repopulating advantage of Nupr1-/- hematopoietic stem cells. (A) Representative plots of hematopoietic stem cell (HSC) analysis by flow cytometry from wild-type (WT), Nupr1-/-, Nupr1-/-Mdm2+/- and Mdm2+/- mice bone marrow. (B) Statistical analysis of WT, Nupr1-/-, Nupr1-/-Mdm2+/- and Mdm2+/- HSC number. Data were analyzed by two-way analysis of variance (ANOVA). n=5. BMNC: bone marrow nucleated cells; n.s.: not significant. (C) Donor bone marrow cells (2.5×105) from WT (black) Nupr1-/- (red), Nupr1-/-Mdm2+/- (blue) (CD45.2) or Mdm2+/- (purple) mice were transplanted into lethally irradiated recipient mice (CD45.1) along with 2.5×105 recipient bone marrow cells. Data were analyzed using an unpaired Student ttest and are represented as mean ± standard deviation (SD). WT, n=5; Nupr1-/-, n=5 mice; Nupr1-/-Mdm2+/-, n=6 mice; Mdm2+/-: n=5. *P<0.05, ***P<0.001. PB: peripheral blood. (D) Flow cytometry analysis of donor-derived HSC and recipient HSC in bone marrow of recipient mice 4 months after transplantation. HSC were gated as Lin– (i.e., CD2–, CD3–, CD4–, CD8–, B220–, Gr1–, CD11b–, Ter119–) CD48– Sca1+ c-Kit+ CD150+ CD34– CD135– cells. Plots from one representative mouse of each group are shown. (E) Statistical analysis of the percentage and absolute number of donor-derived HSC in recipient mice 4 months after transplantation. Data were analyzed by one-way ANOVA and are represented as mean ± SD. WT, n=5; Nupr1-/-, n=5 mice; Nupr1-/-Mdm2+/-, n=6 mice; Mdm2+/-: n=5. ***P<0.001.

mice (median value: 14.13%, P=0.07) (Figure 7A-C). The competitive transplantation result showed that donor Nupr1fl/flMx1-cre cells were advantaged over WT competitors in the peripheral blood of recipients (60%-80%) (Figure 7D). To further investigate whether Nupr1fl/flMx1-cre HSC haematologica | 2022; 107(1)

dominantly occupy recipient bone marrow, we sacrificed the recipients and analyzed the HSC 16 weeks after transplantation. The proportion and absolute number of Nupr1fl/flMx1-cre HSC were significantly greater (~2-fold) than the control HSC competitors in primary recipients 163

T. Wang et al. A

Figure 7. Deletion of Nupr1 in adulthood promotes hematopoietic stem cell engraftment. (A) Cell cycle analysis of Nupr1fl/flMx1-cre hematopoietic stem cells (HSC) under homeostasis. Representative plots of cell cycle from representative wild-type (WT) and Nupr1fl/flMx1-cre mice (8 weeks old). WT littermates (8 weeks old) were used as controls. HSC (Lin– (i.e., CD2, CD3– CD4– CD8– B220– Gr1– CD11b– Ter119–) CD48– Sca1+ c-kit+ CD150+ CD34– CD135–) were analyzed by DNA content (DAPI) versus Ki-67. G0 (Ki-67lowDAPI2N), G1 (Ki-67highDAPI2N), G2-S-M (Ki-67highDAPI>2N-4N). (B) Statistical analysis of the number of long-term HSC from WT and Nupr1fl/flMx1cre mice. BMNC: bone marrow nucleated cells. (C) Statistical analysis of the cell cycle of HSC. Ctr: n=4, Nupr1fl/flMx1-cre, n=5. **P<0.01. (D) Kinetic analysis of donor chimerism (CD45.2+) in peripheral blood. Data were analyzed by two-way analysis of variance and are represented as mean ± SD (Ctr group: n = 5 mice, Nupr1fl/flMx1-cre group: n =7 mice). ***P<0.001. (E) Flow cytometry analysis of the HSC compartment in primary recipients 4 months after transplantation. (F) Statistical analysis of donor HSC number and percentage in the transplantation chimeras. Data were analyzed using an unpaired Student t-test and are represented as mean ± standard deviation, n=5. **P<0.01, ***P<0.001.

(Figure 7E, F). Thus, in the Nupr1fl/fl Mx1-cre model, deletion of Nupr1 in adulthood also promotes HSC engraftment.

Discussion The intrinsic networks regulating the quiescence of HSC are largely unknown. In this study, loss of Nupr1 (p8), a gene preferentially expressed in long-term HSC, mildly tuned the quiescence threshold of HSC in the state of homeostasis, without compromising their essential functions in hematopoiesis. Nupr1 coordinated with p53 to form a signaling machinery regulating HSC quiescence and turnover rates. For the first time, we revealed the new role of Nupr1 in controlling HSC quiescence. Nupr1-/- HSC replenished faster than WT HSC under homeostasis. However, the size of the Nupr1-/- HSC pool 164

was not altered. These findings imply that despite the existence of intrinsic machinery controlling HSC quiescence, the scale of the HSC pool is also restricted by the extrinsic bone marrow microenvironment.48 Conventionally, molecules activating HSC produce a transient phenotypic proliferation of HSC but eventually lead to their functional exhaustion and even tumors.36-40 Interestingly, Nupr1 signaling seemingly plays a unique role in regulating HSC quiescence and turnover rates, as deletion of Nupr1 maintained the hematopoietic features of HSC. Consistently, enforced CDK6 expression in HSC confers these cells a competitive advantage without impairing their stemness and multilineage potential.9 This evidence supports the concept that targeting the intrinsic machinery of balancing the threshold of HSC quiescence might safely promote engraftment. Loss of Nupr1 in HSC resulted in an engraftment advantage. In the setting of transplantation stress, the HSC haematologica | 2022; 107(1)

Nupr1 regulates the quiescence threshold of HSC

niche occupied by WT HSC was ablated, providing a niche vacuum into which donor Nupr1-/- HSC could enter. The dominance of Nupr1-/- HSC is a consequence of faster turnover rates of these cells over their WT counterparts. In a previous study, loss of Dnmt3a also led to clonal dominance of HSC, although accompanied by a failure of hematopoiesis due to a dramatic block in differentiation.4,49 Thus, the engraftment advantage caused by loss of Nupr1 might have prospective translational implications for HSC transplantation, since a faster recovery of hematopoiesis in HSC transplant hosts definitely reduces infection risks in patients.50,51 In our models, Nupr1 regulated hematopoietic homeostasis via targeting the p53 pathway. p53 is essential for regulating hematopoietic homeostasis.27 It is unknown whether NUPR1 interacts directly with p53 in the context of HSC, as commercial antibodies suitable for protein-protein interaction assays are not currently available. NUPR1 and p53 interacted directly in human breast epithelial cells.22 Knocking out p53 in HSC can promote HSC expansion, but directly targeting p53 caused HSC apoptosis and tumorigenesis.52 Thus, Nupr1 might behave as an upstream regulator of p53 signaling and uniquely regulate cell quiescence in the context of HSC. In a previous study, Mdm2 was found to be a key repressive regulator of p53 signaling. MDM2 degrades p53 protein by promoting p53 ubiquitination.46,53 Complete deletion of Mdm2 will lead to embryonic death because of the excess expression of p53.46 This embryonic lethality can, however, be rescued by a combination of Trp53-/-, indicating its essential role of negative regulation of p53. We, therefore, crossed the Nupr1-/- mice with Mdm2+/- mice in order to upregulate p53 expression indirectly. The level of p53 expression is expectedly elevated in Nupr1-/-Mdm2+/- HSC; however, it is even higher than

References 1. Cheshier SH, Morrison SJ, Liao X, Weissman IL. In vivo proliferation and cell cycle kinetics of long-term self-renewing hematopoietic stem cells. Proc Natl Acad Sci U S A. 1999;96(6):3120-3125. 2. Wilson A, Laurenti E, Oser G, et al. Hematopoietic stem cells reversibly switch from dormancy to self-renewal during homeostasis and repair. Cell. 2008;135(6): 1118-1129. 3. Rodrigues NP, Janzen V, Forkert R, et al. Haploinsufficiency of GATA-2 perturbs adult hematopoietic stem-cell homeostasis. Blood. 2005;106(2):477-484. 4. Challen GA, Sun D, Jeong M, et al. Dnmt3a is essential for hematopoietic stem cell differentiation. Nat Genet. 2011;44(1):23-31. 5. Mayle A, Yang L, Rodriguez B, et al. Dnmt3a loss predisposes murine hematopoietic stem cells to malignant transformation. Blood. 2015;125(4):629638. 6. Santaguida M, Schepers K, King B, et al. JunB protects against myeloid malignancies by limiting hematopoietic stem cell proliferation and differentiation without affecting self-renewal. Cancer Cell. 2009;15(4): 341-352. 7. Takubo K, Goda N, Yamada W, et al. Regulation of the HIF-1alpha level is essential for hematopoietic stem cells. Cell Stem Cell. 2010;7(3):391-402. 8. Tesio M, Tang Y, Mudder K, et al.

haematologica | 2022; 107(1)

that in WT mice (Figure 5C, D). A decreased level of MDM2 and increased p53 activity in HSC reduce the ability of competitiveness.26 Thus, it is possible that the downregulating effect of Nupr1 on p53 level is mild, while the upregulation of p53 level by haploid deletion of Mdm2 is dramatic. Consequently, the competitiveness of Nupr1-/Mdm2+/- HSC failed to reach WT level in the rescue assay. In conclusion, loss of Nupr1 in HSC promotes engraftment by tuning the quiescence threshold of HSC via regulation of the p53 checkpoint pathway. Our study unveils the prospect of shortening the engraftment time-window in HSC transplantation by targeting the intrinsic machinery controlling HSC quiescence. Disclosures No conflicts of interest to disclose. Contributions TJW and CXX performed research, analyzed data and wrote the paper; YD and QTW analyzed RNA-sequencing data; SH, FD, KTW, XFL, LJL, YG and YXG performed experiments; JD, TC and HC discussed the manuscript; JYW designed the research, and wrote the manuscript. Funding This work was supported by grants from the National Natural Science Foundation of China (31900814, 81925002, 81922002), Strategic Priority Research Program of the Chinese Academy of Sciences (XDA16010601), Key Research & Development Program of Guangzhou Regenerative Medicine and Health Guangdong Laboratory (2018GZR110104006), CAS Key Research Program of Frontier Sciences (QYZDB-SSW-SM057), and Science and Technology Planning Project of Guangdong Province (2017B030314056).

Hematopoietic stem cell quiescence and function are controlled by the CYLDTRAF2-p38MAPK pathway. J Exp Med. 2015;212(4):525-538. 9. Laurenti E, Frelin C, Xie S, et al. CDK6 levels regulate quiescence exit in human hematopoietic stem cells. Cell Stem Cell. 2015;16(3):302-313. 10. Mallo GV, Fiedler F, Calvo EL, et al. Cloning and expression of the rat p8 cDNA, a new gene activated in pancreas during the acute phase of pancreatitis, pancreatic development, and regeneration, and which promotes cellular growth. J Biol Chem. 1997;272(51):32360-32369. 11. Ree AH, Tvermyr M, Engebraaten O, et al. Expression of a novel factor in human breast cancer cells with metastatic potential. Cancer Res. 1999;59(18):4675-4680. 12. Ree AH, Pacheco MM, Tvermyr M, Fodstad O, Brentani MM. Expression of a novel factor, com1, in early tumor progression of breast cancer. Clin Cancer Res. 2000;6(5):1778-1783. 13. Ito Y, Yoshida H, Motoo Y, et al. Expression and cellular localization of p8 protein in thyroid neoplasms. Cancer Lett. 2003;201 (2):237-244. 14. Mohammad HP, Seachrist DD, Quirk CC, Nilson JH. Reexpression of p8 contributes to tumorigenic properties of pituitary cells and appears in a subset of prolactinomas in transgenic mice that hypersecrete luteinizing hormone. Mol Endocrinol. 2004;18(10): 2583-2593.

15. Brannon KM, Million Passe CM, White CR, Bade NA, King MW, Quirk CC. Expression of the high mobility group A family member p8 is essential to maintaining tumorigenic potential by promoting cell cycle dysregulation in LbetaT2 cells. Cancer Lett. 2007;254(1):146-155. 16. Jiang WG, Davies G, Martin TA, Kynaston H, Mason MD, Fodstad O. Com-1/p8 acts as a putative tumour suppressor in prostate cancer. Int J Mol Med. 2006;18(5):981-986. 17. Malicet C, Lesavre N, Vasseur S, Iovanna JL. p8 inhibits the growth of human pancreatic cancer cells and its expression is induced through pathways involved in growth inhibition and repressed by factors promoting cell growth. Mol Cancer. 2003;2:37. 18. Malicet C, Giroux V, Vasseur S, Dagorn JC, Neira JL, Iovanna JL. Regulation of apoptosis by the p8/prothymosin alpha complex. Proc Natl Acad Sci U S A. 2006;103(8):26712676. 19. Vasseur S, Hoffmeister A, Garcia-Montero A, et al. p8-deficient fibroblasts grow more rapidly and are more resistant to adriamycin-induced apoptosis. Oncogene. 2002;21(11):1685-1694. 20. Carracedo A, Lorente M, Egia A, et al. The stress-regulated protein p8 mediates cannabinoid-induced apoptosis of tumor cells. Cancer Cell. 2006;9(4):301-312. 21. Gironella M, Malicet C, Cano C, et al. p8/nupr1 regulates DNA-repair activity after double-strand gamma irradiation-

165

T. Wang et al. induced DNA damage. J Cell Physiol. 2009;221(3):594-602. 22. Clark DW, Mitra A, Fillmore RA, et al. NUPR1 interacts with p53, transcriptionally regulates p21 and rescues breast epithelial cells from doxorubicin-induced genotoxic stress. Curr Cancer Drug Targets. 2008;8(5):421-430. 23. Dumble M, Moore L, Chambers SM, et al. The impact of altered p53 dosage on hematopoietic stem cell dynamics during aging. Blood. 2007;109(4):1736-1742. 24. Lotem J, Sachs L. Hematopoietic cells from mice deficient in wild-type p53 are more resistant to induction of apoptosis by some agents. Blood. 1993;82(4):1092-1096. 25. Shounan Y, Dolnikov A, MacKenzie KL, Miller M, Chan YY, Symonds G. Retroviral transduction of hematopoietic progenitor cells with mutant p53 promotes survival and proliferation, modifies differentiation potential and inhibits apoptosis. Leukemia. 1996;10(10):1619-1628. 26. Bondar T, Medzhitov R. p53-mediated hematopoietic stem and progenitor cell competition. Cell Stem Cell. 2010;6(4):309322. 27. Liu Y, Elf SE, Miyata Y, et al. p53 regulates hematopoietic stem cell quiescence. Cell Stem Cell. 2009;4(1):37-48. 28. Chen J, Ellison FM, Keyvanfar K, et al. Enrichment of hematopoietic stem cells with SLAM and LSK markers for the detection of hematopoietic stem cell function in normal and Trp53 null mice. Exp Hematol. 2008;36(10):1236-1243. 29. Wang YV, Leblanc M, Fox N, et al. Fine-tuning p53 activity through C-terminal modification significantly contributes to HSC homeostasis and mouse radiosensitivity. Genes Dev. 2011;25(13):1426-1438. 30. Liu D, Ou L, Clemenson GD Jr, et al. Puma is required for p53-induced depletion of adult stem cells. Nat Cell Biol. 2010;12(10):993-998. 31. Yamashita M, Nitta E, Suda T. Regulation of hematopoietic stem cell integrity through p53 and its related factors. Ann N Y Acad Sci. 2016;1370(1):45-54.

166

32. Wilkinson AC, Ishida R, Kikuchi M, et al. Long-term ex vivo haematopoietic-stemcell expansion allows nonconditioned transplantation. Nature. 2019;571(7763): 117-121. 33. Yamamoto R, Morita Y, Ooehara J, et al. Clonal analysis unveils self-renewing lineage-restricted progenitors generated directly from hematopoietic stem cells. Cell. 2013;154(5):1112-1126. 34. Hu Y, Smyth GK. ELDA: extreme limiting dilution analysis for comparing depleted and enriched populations in stem cell and other assays. J Immunol Methods. 2009;347(1-2):70-78. 35. Kiel MJ, He S, Ashkenazi R, et al. Haematopoietic stem cells do not asymmetrically segregate chromosomes or retain BrdU. Nature. 2007;449(7159):238242. 36. Motoda L, Osato M, Yamashita N, et al. Runx1 protects hematopoietic stem/progenitor cells from oncogenic insult. Stem Cells. 2007;25(12):2976-2986. 37. Miyamoto K, Araki KY, Naka K, et al. Foxo3a is essential for maintenance of the hematopoietic stem cell pool. Cell Stem Cell. 2007;1(1):101-112. 38. Ficara F, Murphy MJ, Lin M, Cleary ML. Pbx1 regulates self-renewal of long-term hematopoietic stem cells by maintaining their quiescence. Cell Stem Cell. 2008;2(5):484-496. 39. Tipping AJ, Pina C, Castor A, et al. High GATA-2 expression inhibits human hematopoietic stem and progenitor cell function by effects on cell cycle. Blood. 2009;113(12):2661-2672. 40. Campbell TB, Basu S, Hangoc G, Tao W, Broxmeyer HE. Overexpression of Rheb2 enhances mouse hematopoietic progenitor cell growth while impairing stem cell repopulation. Blood. 2009;114(16):33923401. 41. Stobbe CC, Park SJ, Chapman JD. The radiation hypersensitivity of cells at mitosis. Int J Radiat Biol. 2002;78(12):1149-1157. 42. Yu H, Shen H, Yuan Y, et al. Deletion of Puma protects hematopoietic stem cells

and confers long-term survival in response to high-dose gamma-irradiation. Blood. 2010;115(17):3472-3480. 43. Hao S, Chen C, Cheng T. Cell cycle regulation of hematopoietic stem or progenitor cells. Int J Hematol. 2016;103(5):487-497. 44. Pietras EM, Warr MR, Passegue E. Cell cycle regulation in hematopoietic stem cells. J Cell Biol. 2011;195(5):709-720. 45. Li T, Kon N, Jiang L, et al. Tumor suppression in the absence of p53-mediated cellcycle arrest, apoptosis, and senescence. Cell. 2012;149(6):1269-1283. 46. Honda R, Tanaka H, Yasuda H. Oncoprotein MDM2 is a ubiquitin ligase E3 for tumor suppressor p53. FEBS Lett. 1997;420(1):25-27. 47. Abbas HA, Maccio DR, Coskun S, et al. Mdm2 is required for survival of hematopoietic stem cells/progenitors via dampening of ROS-induced p53 activity. Cell Stem Cell. 2010;7(5):606-617. 48. Anthony BA, Link DC. Regulation of hematopoietic stem cells by bone marrow stromal cells. Trends Immunol. 2014;35(1): 32-37. 49. Challen GA, Sun D, Mayle A, et al. Dnmt3a and Dnmt3b have overlapping and distinct functions in hematopoietic stem cells. Cell Stem Cell. 2014;15(3):350-364. 50. Young JH, Logan BR, Wu J, et al. Infections after transplantation of bone marrow or peripheral blood stem cells from unrelated eonors. Biol Blood Marrow Transplant. 2016;22(2):359-370. 51. Safdar A, Armstrong D. Infections in patients with hematologic neoplasms and hematopoietic stem cell transplantation: neutropenia, humoral, and splenic defects. Clin Infect Dis. 2011;53(8):798-806. 52. Orazi A, Kahsai M, John K, Neiman RS. p53 overexpression in myeloid leukemic disorders is associated with increased apoptosis of hematopoietic marrow cells and ineffective hematopoiesis. Mod Pathol. 1996;9(1):48-52. 53. Haupt Y, Maya R, Kazaz A, Oren M. Mdm2 promotes the rapid degradation of p53. Nature. 1997;387(6630):296-299.

haematologica | 2022; 107(1)

ARTICLE

Hematopoiesis

Human erythroid differentiation requires VDAC1-mediated mitochondrial clearance

Ferrata Storti Foundation

Martina Moras,1,2,3 Claude Hattab,1,2,3 Pedro Gonzalez-Menendez,3,4 Claudio M. Fader,5,6 Michael Dussiot,3,7 Jerome Larghero,8 Caroline Le Van Kim,1,2,3 Sandrina Kinet,3,4 Naomi Taylor,3,4,9 Sophie D. Lefevre1,2,3# and Mariano A. Ostuni1,2,3# 1 Université de Paris, UMR_S1134, BIGR, Inserm, Paris, France; 2Institut National de Transfusion Sanguine, Paris, France; 3Laboratoire d’Excellence GR-Ex, Paris, France; 4 Institut de Génétique Moléculaire de Montpellier, Université de Montpellier, CNRS, Montpellier, France; 5Laboratorio de Biología Celular y Molecular, Instituto de Histología y Embriología (IHEM), Universidad Nacional de Cuyo, CONICET, Mendoza, Argentina; 6Facultad de Odontología, Universidad Nacional de Cuyo, Mendoza, Argentina; 7Université de Paris, UMR_S1163, Laboratory of Cellular and Molecular Mechanisms of Hematological Disorders and Therapeutical Implication, INSERM, Paris, France; 8AP-HP, Hôpital Saint-Louis, Unité de Thérapie Cellulaire, Paris, France and 9Pediatric Oncology Branch, Center for Cancer Research (CCR), National Cancer Institute (NCI), National Institutes of Health (NIH), Bethesda, MD, USA #

Haematologica 2022 Volume 107(1):167-177

SDL and MAO contributed equally as co-senior authors

ABSTRACT

rythroblast maturation in mammals is dependent on organelle clearance throughout terminal erythropoiesis. We studied the role of the outer mitochondrial membrane protein voltage-dependent anion channel-1 (VDAC1) in human terminal erythropoiesis. We show that short hairpin (shRNA)-mediated downregulation of VDAC1 accelerates erythroblast maturation. Thereafter, erythroblasts are blocked at the orthochromatic stage, exhibiting a significant decreased level of enucleation, concomitant with an increased cell death. We demonstrate that mitochondria clearance starts at the transition from basophilic to polychromatic erythroblast, and that VDAC1 downregulation induces the mitochondrial retention. In damaged mitochondria from non-erythroid cells, VDAC1 was identified as a target for Parkin-mediated ubiquitination to recruit the phagophore. Here, we showed that VDAC1 is involved in phagophore’s membrane recruitment regulating selective mitophagy of still functional mitochondria from human erythroblasts. These findings demonstrate for the first time a crucial role for VDAC1 in human erythroblast terminal differentiation, regulating mitochondria clearance.

Correspondence: MARIANO A. OSTUNI mariano.ostuni@inserm.fr SOPHIE D. LEFEVRE sophie.lefevre@inserm.fr Received: April 29, 2020. Accepted: December 21, 2020. Pre-published: January 7, 2021.

Introduction https://doi.org/10.3324/haematol.2020.257121 Mature erythrocytes result from a finely regulated process called erythropoiesis, producing two million erythrocytes every second in healthy human adults.1 In mammals, during the late stages of terminal erythroid differentiation, erythroblasts expel their nuclei and lose their organelles including the Golgi apparatus, endoplasmic reticulum (ER), ribosomes and mitochondria.2 Organelles can be eliminated by the general process of macro-autophagy. This process is initiated by the formation of a double membrane structure called phagophore, engulfing cytoplasmic material to form the autophagosome which fuses with a lysosome for degradation.3 Alternatively, mitochondria can be eliminated by a selective autophagic process called mitophagy, in which outer mitochondrial membrane (OMM) proteins act as signals to recruit the phagophore membrane.4 In erythroid progenitors, much work has shown that autophagy is essential for the organelle removal that occurs during reticulocyte maturation.5,6 Notably though, the majority of these studies were performed in human progenitors that had already reached the reticulocyte stage of differentiation or in genetic mouse models that

haematologica | 2022; 107(1)

167

M. Moras et al.

were deleted for different proteins involved in phagophore formation (i.e., Ulk1, Atg4, Atg7). In these latter models, the importance of autophagy in murine erythroid differentiation was demonstrated by the finding that the delayed clearance of mitochondria and ribosomes is associated with anemia.7-11 Mitophagy results from the binding of adaptors to LC3 (microtubule-associated protein 1 light channel 3B, also known as ATG8) within the growing autophagosome. Once associated with phosphatidyl ethanolamine (PE), LC3-II (PE-conjugated LC3) enables the elongation and maturation of the phagophore that is recruited to ubiquitin-covered mitochondria through interaction with the LIR (LC3-interacting region) motif of mitophagy adaptors. An autophagosome is then built around mitochondria that are finally degraded following autophagosomelysosome fusion. Mitophagy can also occur in an ubiquitin-independent manner following accumulation of OMM proteins such as NIX/BNIP3L,12 bringing together autophagosomal membranes and mitochondria via their LIR motif.4 In this regard, it is interesting to note that the absence of NIX results in mitochondrial retention and anemia in mice.12-14 The PINK1/Parkin pathway is one of the best-characterized pathways of mitophagy. In healthy functional mitochondria, PTEN-induced putative kinase 1(PINK1) is addressed to the OMM. Once translocated to the inner membrane, this protein is cleaved and targeted for degradation by the proteasome. However, on defective mitochondria, there is an accumulation of PINK1 which results in the translocation of Parkin, an E3 ubiquitin ligase, and the subsequent clearance of damaged mitochondria by mitophagy.15-17 At the surface of damaged mitochondria, Parkin interacts with OMM proteins among which voltage-dependent anion channels (VDAC). Furthermore, in the absence of all three VDAC, mitophagy has been shown to be impaired, at least in certain cell types.18-20 VDAC1 is a ubiquitous OMM protein involved in different pathways including apoptosis,21,22 mitochondrial cristae scaffold,23 and nucleotide transport.24,25 Its role in erythropoiesis is unknown, but transcriptomic and proteomic data could suggest a role at the terminal differentiation.26,27 Using a short hairpin RNA (shRNA) approach, we find that downregulation of VDAC1 results in a blockage in erythroid progenitor differentiation at the orthochromatic erythroblast stage, exhibiting a significantly decreased level of enucleation and cell death. Furthermore, we show that the clearance of mitochondria from terminal erythroblasts is dependent on VDAC1, starting at the polychromatic erythroblast stage of differentiation. VDAC1 plays a crucial role to initiate the recruitment of phagophore’s membrane, necessary to achieve an efficient maturation of human red blood cells.

Methods CD34+ cells isolation and ex vivo erythroid differentiation CD34+ cells were isolated from cord blood and cultured following a human ex vivo differentiation protocol as previously described28 (see the Online Supplementary Appendix). The experimental protocol was approved by ethical committees from the Inspire H2020 program (agreement 655850) and from INTS

168

(2019-1), additional information is included in the Online Supplementary Appendix.

Transduction of CD34+ progenitors Cells undergoing erythroid differentiation were transduced at day 4 of erythroid differentiation with lentiviral vectors (MISSION® pLKO.1) containing scrambled shRNA (shSCR) or shRNA targeting VDAC1 (shVDAC1) (TRCN0000297481) upstream of the EGFP transgene, at a MOI of 10 (Online Supplementary Figure S1A). Transduction efficiency was evaluated by the percentage of EGFP+ cells after 72 hours (h) and EGFP+ cells were sorted using the cell sorter SONY SH800.

Flow cytometry-based analysis of erythroid differentiation Cells were analyzed for surface markers expression, mitochondrial content and the presence of a nucleus, starting at day 7. Briefly, 105 cells were stained with 250 nM MitoFluor and Hoechst 34580 in media for 30 minutes (min) at 37°C. Cells were washed and subsequently stained with fluorochrome-conjugated antibodies against glycophorin A (GPA), Band3 and a4integrin in phosphate-buffered saline (PBS) 2% bovine serum albumin (BSA), for 30 min at 4°C. Cells were washed once with PBS and 7-AAD was added prior to acquisition to exclude dead cells. Cells were analyzed using a BD FACSCanto™ (BD Biosciences), data were acquired with Diva software and analyzed using FCS express 6 Flow Research Edition software.

Enucleation assay by imaging flow cytometry Orthochromatic erythroblasts were sorted at day 12 as previously described.28 Cells were maintained in culture for 1 day in phase III medium before staining with 1 mg/mL Hoechst 34580 and anti-GPA antibody. Nucleus polarization was analyzed by Amnis® Imaging Flow Cytometer. Approximately 10,000 events were collected and the “delta centroid GPA-Hoechst” was calculated as the distance of the center of the GPA-labeled erythroblast from the center of the Hoechst-labeled nucleus. A threshold of 2 for the delta centroid (DC) was fixed in order to discriminate between a polarized and non-polarized nucleus. Analysis was performed using the IDEAS 6.2 software.29

Other standard methods Quantitative real-time polymerase chain reaction (qRT-PCR), antibodies and dyes, May-Grünwald-Giemsa (MGG)-based identification of erythroblasts stage, apoptosis assay, reactive oxygen species (ROS) detection, mitochondrial respiration analysis, immunoblotting, immunolabeling for confocal microscopy, ATP measurement, electron microscopy, K562 cell culture, image stream co-localization and statistical analysis are fully described in the Online Supplementary Appendix.

Results shRNA-mediated knockdown of VDAC1 results in an accelerated erythropoiesis until the orthochromatic erythroblast stage of differentiation In order to assess the specific effects of VDAC1 during erythroid differentiation, we pursued a shRNA-mediated knockdown approach. CD34+ cells were transduced at day 4 of differentiation with a shSCR or a shVDAC1 lentiviral vector, both harboring the EGFP transgene (Online Supplementary Figure S1A), and EGFP+ cells were sorted at day 7 (Figure 1A). Knockdown efficiency was assessed at day 10 and a significant downregulation of haematologica | 2022; 107(1)

VDAC-driven mitophagy in human erythropoiesis

VDAC1 expression, both at the mRNA (53.6 ± 7.6%) and protein level (78.5 ± 5.9%) was observed as compared to shSCR-transduced cells (Online Supplementary Figure S1B and C). Of note, no significant difference was observed in the mRNA expression of two other VDAC, VDAC2 and

VDAC3, confirming the specificity of the shRNA construct (Online Supplementary Figure S1D). The effect of VDAC1 downregulation on erythroblast maturation was assessed by quantifying the percentages of erythroblasts at different stages of differentiation, as

B D

Figure 1. VDAC1 knockdown in human erythroid progenitors accelerates erythroid differentiation. (A) Schematic of the ex vivo erythroid differentiation protocol following short hairpin RNA (shRNA)-mediated downregulation of VDAC1. CD34+ progenitors were isolated from cord blood and transduced at day 4 with a lentiviral vector harboring either the VDAC1 shRNA or scramble shRNA (shSCR), together with the EGFP transgene. EGFP+ cells were sorted at day 7 and differentiated until day 17. (B) Representative May-Grünwald-Giemsa (MGG) images of erythroid progenitors and quantification of cells at different stages of differentiation are shown for day 7 and (C) day 10 of differentiation (n=3). Scale bar =20 mm. (D) Representative a4-integrin/Band3 flow cytometry profiles of GPA+ cells at day 10 are shown, allowing different stages of differentiation to be distinguished. (E) Band3 and (F) a4-integrin surface expressions were evaluated on glycophorin A positive (GPA+) cells following introduction of shSCR (black) and shVDAC1 (red) at days 8, 10, 14, and 16 of erythroid differentiation. The percentages at each time point were quantified and means ± standard error are shown (n=9). (G) Erythroblast proliferation in control progenitors and shVDAC1-transduced progenitors was monitored at days 7, 10, 14 and 17 (n=4). *P<0.05, **P<0.01, ***P<0.001. CBMC: cord blood mononucleated cells; shVDAC1: VDAC1 shRNA; D: day. ProE: proerythroblast; BasoE: basophilic erythroblasts; PolyE: polychromatic erythroblasts; OrthoE: orthochromatic erythroblasts; Retic: reticulocytes.

haematologica | 2022; 107(1)

169

M. Moras et al.

monitored by MGG coloration. VDAC1 downregulation did not alter erythroid differentiation before 3 days posttransduction (D7) (Figure 1B). However, by day 10 of differentiation, shVDAC1 transduced cells exhibited an accelerated terminal differentiation as compared to control cells, shown by a significantly higher percentage of orthochromatic erythroblasts as compared to earlier basophilic and polychromatic erythroblasts (Figure 1C). These data were confirmed by evaluating a4integrin/Band3 profile by flow cytometry allowing the terminal stage of differentiation to be distinguished as previously described28 (Figure 1D). Consistent with the data above, shVDAC1-transduced progenitors showed no difference in Band3 expression (Figure 1E) but a significantly decreased levels of a4 integrin, a marker which is lost during terminal differentiation, as compared to shSCR-transduced cells (Figure 1F). Furthermore, we observed a decreased proliferation rate in shVDAC1transduced cells starting at day 10 (Figure 1G). Altogether, these data point to the importance of VDAC1 in regulating terminal erythroid differentiation.

VDAC1 downregulation impairs erythroblast enucleation The main morphological feature of terminal mammalian erythroid differentiation is the enucleation, leading to the production of a reticulocyte and a pyrenocyte from an orthochromatic erythroblast. Despite the accelerated differentiation observed in shVDAC1-transduced cells,

orthochromatic erythroblast percentage was increased whereas reticulocyte level was significantly decreased from day 14 until day 17 as evaluated by MGG coloration (Figure 2A and B) as well as Hoechst staining (Figure 2C). Enucleation starts with the polarization of the nucleus towards the plasma membrane, allowing the separation of the nucleus from the yet-to-be formed reticulocyte by a mechanism similar to cytokinesis.30 In order to assess whether the decrease in enucleation in shVDAC1-transduced cells was due to an impairment in the polarization of the nucleus, we sorted orthochromatic erythroblasts at day 12 of differentiation and analyzed these cells by imaging flow cytometry 24 hours (h) later. We measured the distance between the center of the cell and the center of the nucleus (delta centroid XY from the IDEAS software). No significant differences in delta centroid XY were observed between shVDAC1- and shSCR-transduced cells, suggesting that VDAC1 downregulation does not block the polarization of the nucleus (Online Supplementary Figure S2). In order to assess whether the block in enucleation affects cell viability, we evaluated the level of apoptosis as a function of annexin V staining of exposed phosphatidylserine. While annexin V staining was not altered in shVDAC1-transduced cells as compared to shSCRtransduced cells at days 7 and 10 of differentiation, levels were significantly higher at days 14 and 17, strongly suggesting that an attenuated enucleation was associated with apoptosis (Figure 2D).

Figure 2. Enucleation is impaired in short hairpin VDAC1-transduced erythroid progenitors. (A) The progression of terminal erythroid progenitors to the reticulocyte stage of differentiation was evaluated following scramble shRNA (shSCR) and VDAC1 shRNA (shVDAC1) transduction. Representative May-Grünwald-Giemsa (MGG) images (left) and quantification of the different progenitor stages (right) at day 14 and (A) day 17 (B) are presented and means ± standard error (SE) are shown. Scale bar =20 mm. (C) The percentages of enucleated cells (glycophorin A postive [GPA+], Hoechst- cells) were quantified by flow cytometry at day 10, 14 and 17 of differentiation (n=4) and means ± SE are shown. (D) Apoptosis was evaluated by annexin V staining and the quantification of annexin V+ cells is presented at day 7, 10, 14 and 17 (n=3). *P<0.05, **P<0.01, ****P<0.0001. BasoE: basophilic erythroblasts; PolyE: polychromatic erythroblasts; OrthoE: orthochromatic erythroblasts; Retic: reticulocytes.

170

haematologica | 2022; 107(1)

VDAC-driven mitophagy in human erythropoiesis

Mitochondrial biomass remains elevated in shVDAC1-transduced erythroblasts In mammals, mature red blood cells are devoid of mitochondria. However, mitochondria can still be found at the reticulocyte stage in many mammalian species such as rabbits, dogs, rats, mice and humans.5,31-33 As expected, we detected a progressive decrease in mitochondrial biomass during erythroblast maturation, with a first major decrease in the transition between basophilic and polychromatic stages and a second between polychromatic and orthochromatic stages (Figure 3A). At day 10 of differentiation the mitochondrial marker TOM40 was detected at significantly higher levels in shVDAC1-transduced cells by western blot analysis (2.6±0.6-fold, n=5; Figure 3B). As VDAC1 downregulation alters the kinetics of differentiation, difference on protein expression at day 10 could results from a different proportion of erythroblastic stages. For this reason, mitochondrial biomass was assessed by flow cytometry in the different stages based on a-4integrin/Band3 profile as previously described. Mitochondrial biomass was significantly higher in shVDAC1-polychromatic and orthochromatic erythroblasts as compared to shSCR-transduced erythroblasts (Figure 3C). Importantly, these differences were associated with late stages of differentiation as there was no change in mitochondrial biomass in earlier basophilic erythroblasts. Although shSCR transduced cells progressively eliminate mitochondria at each differentiation stage, our data point to a defect in mitochondrial clearance dur-

ing the transition between basophilic and polychromatic erythroblasts in conditions of VDAC1 downregulation.

VDAC1 regulates mitochondrial morphology and oxidative phosphorylation in terminal erythroblasts In order to evaluate mitochondrial function in shVDAC1-transduced erythroblasts, we first evaluated the structural morphology of mitochondria by transmission electron microscopy at day 10 of differentiation (Figures 4A; Online Supplementary Figure S3). In contrast with the normal ultrastructural morphology detected in shSCR-transduced cells, shVDAC1-transduced cells exhibited swollen mitochondrial cristae (Figure 4A and B) with a significantly increased percentage of rounded as compared to elongated mitochondria (P<0.0001, Figure 4A to C). These data are in agreement with a published role of VDAC1 OMM-IMM contacts sites in cristae structuration.23 Interestingly, we also observed a reduced number of ER-mitochondria contact sites in shVDAC1-transduced cells (Figure 4D). As these mitochondrial associated membranes have been shown to recruit signaling molecules such as mTOR, GSK3 and hexokinase I, thereby increasing oxidative phosphorylation,34 it was of interest to evaluate mitochondrial function in erythroblasts with downregulated VDAC1. Notably, shVDAC1-transduced erythroblasts exhibited a significantly decreased oxygen consumption rate, evaluated as a function of mitochondrial biomass (Figure 4E). This measure, quantifying oxidative phosphorylation

Figure 3. VDAC1 downregulation causes a retention of mitochondria in late stage erythroblasts. (A) Mitochondrial mass was measured by Mitofluor staining in erythroblasts at different stages of terminal differentiation and representative histograms (left) and quantification of the mean fluorescence intensity (MFI) are shown (n=5) (right). (B) Levels of the mitochondrial protein TOM40 were quantified by immunoblot at day 10 of differentiation and representative blots (left) as well as normalization to b-actin (right) are presented. Levels in control cells was arbitrarily set at “1” (n=5). (C) Mitochondrial content of erythroblasts at day 11 of differentiation was evaluated by Mitofluor staining on basophilic (BasoE), polychromatic (PolyE) and orthochromatic (OrthoE) erythroblasts and MFI in progenitors transduced with scramble shRNA (shSCR) (black dots) and VDAC1 shRNA (shVDAC1) (red dots) transduced cells are presented (n=5). Gates for erythroblasts populations are based on a4-integrin/Band3 profiles as shown in Figure 1D. *P<0.05, **P<0.01. ProE: proerythroblast.

haematologica | 2022; 107(1)

171

M. Moras et al.

(OXPHOS), was lower in shVDAC1-transduced erythroblasts for both basal levels and maximal respiration (Figure 4F). In accord with the lower OXPHOS, intracellular ATP levels were significantly lower in shVDAC1transduced cells than control cells (Figure 4G). We also measured mitochondria membrane potential and ROS accumulation at day 10 and showed no difference in cells transduced with shVDAC1 compared to the shSCR (Online Supplementary Figure S4), suggesting an adaptation of the cell to compensate for mitochondria changes of metabolism upon VDAC1 downregulation. Thus, decreased expression of VDAC1 attenuates mitochondr-

ial function, resulting in a diminished energetic state of the differentiating erythroblast.

Delayed mitochondrial clearance in late-stage erythroblasts with downregulated VDAC1 is due to a defective recruitment of autophagosomal membranes Given that VDAC1 has been shown to regulate mitophagy in non-erythroid cell lines,18-20 we hypothesized that VDAC1 downregulation could impair mitophagy in human erythroblasts as well. While the autophagy-mediated lipidation of LC3-I to LC3-II in the autophagosome membrane was not altered upon VDAC1 downregulation

Figure 4. Mitochondria morphology is altered and oxidative phosphorylation is attenuated upon VDAC1 downregulation. (A) Sample transmission electron microscopy (TEM) images of erythroblasts at day 10 (D10) of differentiation (black arrows show endoplasmic reticulum [ER]-mitochondria contact site) with swollen cristae indicated by red arrows. Scale bar =200 nm. (B) TEM-based quantification of the percentage of altered cristae and (C) number of rounded (white) and elongated (black) mitochondria by mm2 are presented. (D) The number of mitochondria-ER contact site by mm2 is presented. Quantification in 20 cells for panels B and C and nine different cells for panel D. *P<0.05, ***P<0.001, ****P<0.0001. (E) Oxygen consumption rates (OCR) were monitored in the indicated populations at D10 of differentiation by Seahorse technology and representative profiles are formed (n>3 technical replicates). Data were normalized to mitochondrial quantity using mean fluorescence intensity (MFI) values of mitotracker deep red previously measured by flow cytometry. (F) Basal and maximum respiration were quantified with levels in control cells arbitrarily set at “1” (n=3). (G) Intracellular ATP in scramble shRNA (shSCR) and VDAC1 shRNA (shVDAC1)-transduced erythroblasts was evaluated at D10 of differentiation by luminometry with levels in control cells arbitrarily set at “1” (n=3). Data are presented as means ± standard error. *P<0.05, ***P<0.001.

172

haematologica | 2022; 107(1)

VDAC-driven mitophagy in human erythropoiesis

Figure 5. Autophagy is impaired in VDAC1 scramble short hairpin RNA-transduced cells. (A) Schematic representation of the role of the PINK1/Parkin and NIX pathways in mitochondrial degradation. (B) LC3-I and LC3-II levels were evaluated in scramble shRNA (shSCR) and VDAC1 shRNA (shVDAC1)-transduced erythroblasts at day 10 (D10) and representative immunoblots are presented (left). LC3-II levels were normalized to b-actin with levels in control cells arbitrarily set at “1” (n=5) (right). (C) p62 levels were monitored and representative immunoblots (left) and quantifications (right) are presented. p62 levels were normalized to b-actin with levels in control cells arbitrarily set at “1” (n=6). (D) p62 protein level was evaluated by flow cytometry on basophilic (BasoE), polychromatic (PolyE) (n=8) and orthochromatic (OrthoE) (n=3) and mean fluorescence intensity (MFI) of shSCR (black dots) and shVDAC1 (red dots) are presented with levels in control cells arbitrarily set at “1”. Gates for erythroblasts populations are based on a4-integrin/Band3 profiles as shown in Figure 1D. *P<0.05. (E) Co-localization of LC3 and TOM22 was monitored by confocal microscopy and representative images are shown. Scale bar =3 mm. (left) The Pearson coefficient between LC3 and TOM22 are presented for 30 cells in a representative experiment (n=3) (right). *P<0.05, **P<0.01, ****P<0.0001. ns: non-significant.

haematologica | 2022; 107(1)

173

M. Moras et al.

(Figure 5B), there was a significant accumulation of p62, an adaptor protein that allows bridging between the ubiquitinated cargos and the phagophore membrane35 (Figure 5A to C). Furthermore, p62 expression was detected by flow cytometry in different stages on fixed cells stained with a4-integrin and Band3 antibodies. As well as for mitochondrial content, a significant increase in p62 was observed at polychromatic and orthochromatic stages in shVDAC1-transduced cells (Figure 5D). After engulfment of the mitochondria, the autophagosome usually fuses with a lysosome, leading to the degradation of the mitochondria along with the adaptor proteins (Figure 5A). In order to decipher whether p62 accumulation was due to a stabilization at MOM on ubiquitinated proteins or to a defect in mitophagy completion by lysosomal degradation, we studied the autophagosomal membrane recruitment to mitochondria by immunofluorescence. We observed a significant decreased co-localization between autophagosomal LC3-II and the mitochondrial marker TOM22 in shVDAC1-transduced cells (P<0.01; Figure 5E), suggesting a defect in the phagophore membrane recruitment and consequently a defect in mitophagy completion. In order to assess the specificity of our results, we constructed shSCR-K562 and shVDAC1-K562 stable cell lines and transfected shSCR-K562 cells with a small interfering RNA (siRNA) against VDAC1 (siVDAC1-K562). We confirmed VDAC1 downregulation by western blot (Online Supplementary Figure S5A), and showed, as for differentiating erythroblasts, no difference in LC3 activation (Online Supplementary Figure S5B) and an increase in p62 protein level (Online Supplementary Figure S5C). We also observed a decrease of LC3/TOM22 co-localization in shVDAC1-K562 and siVDAC1-K562, which indicates that the recruitment of the phagophore membrane to the mitochondria was impaired upon VDAC1 downregulation (Online Supplementary Figure S5D). These data confirmed that VDAC1 downregulation in K562 erythroid cells recapitulates the phenotype observed on VDAC1downregulated erythroblasts.

PINK1 protein level is reduced upon VDAC1 downregulation The phagophore recruitment is dependent on the stabilization of some specific MOM proteins and direct or indirect interaction with lipidated LC3 (Figure 5A). We first evaluated NIX levels, previously identified in mouse reticulocytes as being critical for the recruitment of autophagosomal membranes and appropriate mitophagy.36 Notably, we detected no significant decrease in neither NIX dimers nor NIX transcripts in human erythroblasts with decreased VDAC1 expression (Online Supplementary Figure S6). We then studied the second main mitophagy induction pathway and observed a significant decrease of full-length PINK1 protein level in shVDAC1-transduced bulk erythroblasts at D10 by western blotting (Figure 6A). PINK1 expression was then evaluated by flow cytometry on basophilic, polychromatic and orthochromatic erythroblasts based on the a4-integrin and Band3 expression pattern. We confirmed a significant decrease of PINK1 level starting from the polychromatic stage (Figure 6B), in line with the differences of mitochondrial mass and p62 expression. We next confirmed the reduction of PINK1 protein level upon VDAC1 downregulation in K562 cell lines (Figure 174

6C). Furthermore, we observed a reduction of the mitochondrial localization of PINK1, as shown by the reduction of PINK1/TOM22 co-localization in shVDAC1-K562 cells (Figure 6D). Finally, we observed a decrease of the phagophore recruitment to the mitochondria when downregulating PINK1 in shSCR-K562, as observed with the significant reduction of LC3/TOM22 co-localization (Figure 6E). In contrast, LC3/TOM22 co-localization did not worsen upon PINK1 downregulation in shVDAC1K562 cells (Figure 6E).

Discussion Mitochondrial clearance is a crucial process for the production of mature red blood cells and several studies have shown that mitophagy occurs at the reticulocyte stage, allowing the complete clearance of mitochondria and leading to final murine erythrocyte maturation.13,14 Mitophagy is a mitochondrial-specific form of general macroautophagy, which can be mediated by different well characterized pathways involving protein signaling as the PINK1/Parkin pathway, NIX and BNIP3. While autophagy related genes (ATG) have been shown to be critical for mitochondrial clearance during erythropoiesis,6-11 and the NIX-driven pathway has been identified as plying a role in murine erythropoiesis,12-14 the importance of these mitophagy pathways in human erythroid terminal differentiation is unknown. Here we focused on the role of VDAC1, an OMM protein, known to be involved in mitophagy in non-erythroid cells. We found that VDAC1 downregulation accelerates human erythroblast differentiation until the orthochromatic erythroblast stage, but then blocks enucleation and cumulates mitochondria with altered ultrastructure and decreased function, resulting in apoptosis. In agreement with previous data on the role of VDAC1 in the structuration of mitochondria,23 we observed altered mitochondria ultrastructure and a decrease in ERmitochondria contact sites in shVDAC1-transduced erythroblasts by electron microscopy. Notably, complexes between VDAC and hexokinase-1 at ER-mitochondrial contact sites have been shown to regulate mitochondrial metabolism.34 In our study, mitochondria in VDAC1downregulated erythroblasts showed a decrease in OCR and intracellular ATP content, a signature of reduced mitochondrial activity. Thus, perturbations in mitochondrial metabolism may influence the kinetics of erythroblast differentiation and their enucleation profiles. We demonstrated that VDAC1 is required for progressive mitochondrial clearance during the terminal phase of differentiation. This is in line with a previous study showing a decrease in mitochondrial mass after the proerythroblastic stage.37 We found that this is regulated through the PINK1/Parkin pathway, with VDAC1 modulating PINK1 accumulation at the mitochondria outer membrane and the recruitment of the phagophore. Accordingly, the attenuation of mitochondrial clearance was associated with increased levels of p62, an adaptor protein that allows bridging between the ubiquitinated cargos and the phagophore membrane.35 Our data therefore highlight the critical role of VDAC1 function in the recognition of mitochondria that are “to be degraded” during the process of red cell maturation. haematologica | 2022; 107(1)

VDAC-driven mitophagy in human erythropoiesis

We recently demonstrated that TSPO1 downregulation also affects mitochondria clearance with no effect on the erythroblasts’ differentiation kinetics,38 raising the possibility that this phenotype is mediated through the VDAC1-TSPO1 OMM complex.

In addition, our findings suggest that the canonical PINK1/Parkin pathway may play an important role during human erythropoiesis, confirming in humans a previous study on mice.39 Transcriptome analyses in human erythroblasts detected an upregulation, starting from the

Figure 6. Lower PINK1 protein levels upon VDAC1 downregulation. (A) PINK1 levels were evaluated by immunoblot in the indicated conditions at day 10 of differentiation and normalized to the quantity of mitochondria (housekeeping mitochondrial TOM40) on total number of cell (housekeeping b-actin) (top). Levels in control cells were arbitrarily set at “1” (n=5) (bottom). (B) PINK1 protein level was evaluated by flow cytometry on basophilic (BasoE), polychromatic (PolyE) (n=8) and orthochromatic (OrthoE) erythroblasts (n=3). PINK1 mean fluorescence intensity (MFI) was normalized to mitochondrial housekeeping TOM22 MFI in erythroblasts transduced with scramble shRNA (shSCR) (black dots) and VDAC1 shRNA (shVDAC1) (red dots), with levels in control cells arbitrarily set at “1”. Gates for erythroblasts populations are based on a4-integrin/Band3 profiles as shown in Figure 1D. (C) PINK1 protein levels were evaluated by immunoblot in the indicated conditions in shSCR-K562 and shVDAC1-K562 cells and normalized to the quantity of mitochondria marker TOM40 on total number of cell (β-actin). Levels in control cells were arbitrarily set at “1” (n=4). (D) PINK1/TOM22 co-localization (MPI ratio from PINK1 and TOM22 signals) assessed by imaging flow cytometry of shSCR-K562 and shVDAC1-K562 cells (n =3). (E) LC3/TOM22 co-localization assessed by imaging flow cytometry (myeloma prognostic index [MPI] ratio from LC3 and TOM22 signals) of shSCR-K562 and shVDAC1-K562 cells and shSCR-K562 cells transfected with a PINK1 siRNA (n=3). *P<0.05, ***P<0.001, ****P<0.0001. ns: non-significant; OMM: outer mitochondrial membrane.

haematologica | 2022; 107(1)

175

M. Moras et al.

polychromatic stage, of key proteins of mitophagy belonging to different pathways, such as NIX and PINK1.26 NIX upregulation was described by several works studying NIX-mediated mitophagy in murine reticulocytes,12-14 but no clear evidence was found to explain the upregulation of PINK1. Under the condition of VDAC1 downregulation, NIX protein levels were not altered, suggesting that mitochondrial retention was not due to a defect in the NIX pathway but in the PINK1 pathway, as described below. It was previously reported that VDAC is required to promote Parkin recruitment to the mitochondria.18 Our data suggest that this could be due to a lack in PINK1 accumulation at the OMM. Since VDAC1 is part of a multimeric complex with ATAD3A,23 we speculate that in the absence of VDAC1, ATAD3A is free to form alternative complexes.40 Recently, a new function of ATAD3A was described in hematopoietic progenitor cells, promoting the import of PINK1.41 In this context PINK1 is rapidly degraded by the mitochondrial peptidases MPP and Parl to prevent its accumulation in the surface prevent mitophagy. However, when ATAD3A dissociates from the translocase of the inner membrane (TIM) complex, the import of PINK1 is blocked, causing its accumulation.42 It is noteworthy that also hexokinase 2 (HK2) and adenine nucleotide transporter (ANT), both proteins known to be involved in a multimeric mitochondrial membrane protein complex, have been recently shown to be essential for PINK1 accumulation at the OMM,43,44 suggesting a role of this complex in PINK1/Parkin mitophagy. Further studies are necessary to assess this hypothesis and evaluate the crosstalk between VDAC1, ATAD3A, ANT and/or HK2 in the terminal phase of erythropoiesis. In conclusion, our data are the first demonstration of a fundamental role of mitochondrial-selective autophagy at the erythroblast stage. Loss of VDAC1 causes an ineffective elimination of mitochondria in the transition between basophilic and orthochromatic erythroblasts, demonstrating the critical role of VDAC1 in late stage erythropoiesis. We have taken account of recent reports highlighting the possible unspecific and strain-dependent effects of shRNA in knocking down mouse erythroblasts enucleation.45 However, we have observed the same effects after knocking down VDAC1 by two different approaches (shRNA and siRNA) and by using different experimental models, strongly supporting our conclusions.

References 1. Palis J. Primitive and definitive erythropoiesis in mammals. Front Physiol. 2014;5:3. 2. Moras M, Lefevre SD, Ostuni MA. From erythroblasts to mature red blood cells: organelle clearance in mammals. Front Physiol. 2017;8:1076. 3. Klionsky DJ. The molecular machinery of autophagy: unanswered questions. J Cell Sci. 2005;118(Pt 1):7-18. 4. Liu L, Sakakibara K, Chen Q, Okamoto K. Receptor-mediated mitophagy in yeast and mammalian systems. Cell Res. 2014;24(7):787-795. 5. Kent G, Minick OT, Volini FI, Orfei E. Autophagic vacuoles in human red cells. Am J Pathol. 1966;48(5):831-857. 6. Zhang J, Wu K, Xiao X, et al. Autophagy as a regulatory component of erythropoiesis. Int J Mol Sci. 2015;16(2): 4083-4094.

176

Over the last few years, several studies have demonstrated the association of mitophagy defects with various hematological syndromes, highlighting the importance of mitochondrial content and mitophagy during human erythropoiesis.46-53 Our work provides a better understanding of the mechanisms regulating mitochondrial autophagy in human erythroid progenitors, particularly during the terminal phase of erythropoiesis. Elucidating the different pathways involved in mitochondrial clearance and their kinetics, will promote the development of novel therapeutic approaches for treating hematological disorders that involve defective erythroid maturation. Disclosures No conflicts of interest to disclose Contributions MM designed experiments, performed experiments, analyzed data and wrote the paper; CH, PGM, MD, SK and CF performed experiments; JL evaluated protocols and supervised blood samples purchasing; CLVK critically evaluated experiments and obtained funding; SK and NT critically evaluated experiments, provided supervision, wrote the paper and obtained funding; SDL and MAO conceived the project, obtained funding, designed and critically evaluated experiments, provided supervision and wrote the paper; SDL and MAO contributed equally to this paper. All authors read and commented on the paper and approved the final version. Acknowledgments The authors would like to thank Mrs Corentine Chrysostome et Mr Abdellah Nait for their secretarial and technical assistance, respectively. We acknowledge the Cell Sorting facility of the Institut Imagine; the ImagoSeine core facility of the Institut Jacques Monod, member of IBiSA and France-BioImaging (ANR-10-INBS-04) infrastructure; the Laboratory of Excellence GR-Ex (Grant ANR-11-LABX-0051) and the Guests Researcher program from Paris Diderot University. Funding MM is funded by the European Union’s Horizon 2020 research and innovation program under the Marie SkłodowskaCurie grant agreement No. 665850, the Club du Globule Rouge et du Fer (CGRF) and Société Française d’Hématologie (SFH); PG-M is founded by the CLARIN-COFUND program from the Principado de Asturias and the European Union.

7. Honda S, Arakawa S, Nishida Y, Yamaguchi H, Ishii E, Shimizu S. Ulk1-mediated Atg5independent macroautophagy mediates elimination of mitochondria from embryonic reticulocytes. Nat Commun. 2014;5:4004. 8. Betin VM, Singleton BK, Parsons SF, Anstee DJ, Lane JD. Autophagy facilitates organelle clearance during differentiation of human erythroblasts: evidence for a role for ATG4 paralogs during autophagosome maturation. Autophagy. 2013;9(6):881-893. 9. Zhang J, Randall MS, Loyd MR, et al. Mitochondrial clearance is regulated by Atg7-dependent and -independent mechanisms during reticulocyte maturation. Blood. 2009;114(1):157-164. 10. Kundu M, Lindsten T, Yang CY, et al. Ulk1 plays a critical role in the autophagic clearance of mitochondria and ribosomes during reticulocyte maturation. Blood. 2008;112(4): 1493-1502.

11. Mortensen M, Ferguson DJ, Edelmann M, et al. Loss of autophagy in erythroid cells leads to defective removal of mitochondria and severe anemia in vivo. Proc Natl Acad Sci U S A. 2010;107(2):832-837. 12. Novak I, Kirkin V, McEwan DG, et al. Nix is a selective autophagy receptor for mitochondrial clearance. EMBO Rep. 2010;11 (1):45-51. 13. Sandoval H, Thiagarajan P, Dasgupta SK, et al. Essential role for Nix in autophagic maturation of erythroid cells. Nature. 2008;454 (7201):232-235. 14. Schweers RL, Zhang J, Randall MS, et al. NIX is required for programmed mitochondrial clearance during reticulocyte maturation. Proc Natl Acad Sci U S A. 2007;104(49):19500-19505. 15. Matsuda N, Sato S, Shiba K, et al. PINK1 stabilized by mitochondrial depolarization recruits Parkin to damaged mitochondria

haematologica | 2022; 107(1)

VDAC-driven mitophagy in human erythropoiesis

and activates latent Parkin for mitophagy. J Cell Biol. 2010;189(2):211-221. 16. Ordureau A, Paulo JA, Zhang W, et al. Dynamics of PARKIN-dependent mitochondrial ubiquitylation in induced neurons and model systems revealed by digital snapshot proteomics. Mol Cell. 2018;70(2):211-227. 17. Narendra DP, Jin SM, Tanaka A, et al. PINK1 is selectively stabilized on impaired mitochondria to activate Parkin. PLoS Biol. 2010;8(1):e1000298. 18. Sun Y, Vashisht AA, Tchieu J, Wohlschlegel JA, Dreier L. Voltage-dependent anion channels (VDACs) recruit Parkin to defective mitochondria to promote mitochondrial autophagy. J Biol Chem. 2012;287(48): 40652-40660. 19. Geisler S, Holmstrom KM, Skujat D, et al. PINK1/Parkin-mediated mitophagy is dependent on VDAC1 and p62/SQSTM1. Nat Cell Biol. 2010;12(2):119-131. 20. Gatliff J, East D, Crosby J, et al. TSPO interacts with VDAC1 and triggers a ROS-mediated inhibition of mitochondrial quality control. Autophagy. 2014;10(12):2279-2296. 21. Shoshan-Barmatz V, Keinan N, Abu-Hamad S, Tyomkin D, Aram L. Apoptosis is regulated by the VDAC1 N-terminal region and by VDAC oligomerization: release of cytochrome c, AIF and Smac/Diablo. Biochim Biophys Acta. 2010;1797(6-7):12811291. 22. Shimizu S, Narita M, Tsujimoto Y. Bcl-2 family proteins regulate the release of apoptogenic cytochrome c by the mitochondrial channel VDAC. Nature. 1999;399(6735): 483-487. 23. Rone MB, Midzak AS, Issop L, et al. Identification of a dynamic mitochondrial protein complex driving cholesterol import, trafficking, and metabolism to steroid hormones. Mol Endocrinol. 2012;26(11):18681882. 24. Lee AC, Xu X, Colombini M. The role of pyridine dinucleotides in regulating the permeability of the mitochondrial outer membrane. J Biol Chem. 1996;271(43):2672426731. 25. Rostovtseva TK, Komarov A, Bezrukov SM, Colombini M. VDAC channels differentiate between natural metabolites and synthetic molecules. J Membr Biol. 2002;187(2):147156. 26. An X, Schulz VP, Li J, et al. Global transcriptome analyses of human and murine terminal erythroid differentiation. Blood. 2014; 123(22):3466-3477. 27. Gautier EF, Ducamp S, Leduc M, Salnot V, Guillonneau F, Dussiot M, et al. Comprehensive proteomic analysis of human erythropoiesis. Cell Rep. 2016;16(5): 1470-1484.

haematologica | 2022; 107(1)

28. Hu J, Liu J, Xue F, et al. Isolation and functional characterization of human erythroblasts at distinct stages: implications for understanding of normal and disordered erythropoiesis in vivo. Blood. 2013;121(16): 3246-3253. 29. Konstantinidis DG, Pushkaran S, Johnson JF, et al. Signaling and cytoskeletal requirements in erythroblast enucleation. Blood. 2012;119(25):6118-6127. 30. Wang J, Ramirez T, Ji P, Jayapal SR, Lodish HF, Murata-Hori M. Mammalian erythroblast enucleation requires PI3K-dependent cell polarization. J Cell Sci. 2012;125(Pt 2):340-349. 31. Gasko O, Danon D. Deterioration and disappearance of mitochondria during reticulocyte maturation. Exp Cell Res. 1972;75(1):159-169. 32. Simpson CF, Kling JM. The mechanism of mitochondrial extrusion from phenylhydrazine-induced reticulocytes in the circulating blood. J Cell Biol. 1968;36(1):103-109. 33. Beams HW, Kessel RG. Electron microscope and ultracentrifugation studies on the rat reticulocyte. Am J Anat. 1966;118(2):471507. 34. Bantug GR, Fischer M, Grahlert J, et al. Mitochondria-endoplasmic reticulum contact sites function as immunometabolic hubs that orchestrate the rapid recall response of memory CD8+ T cells. Immunity. 2018;48(3):542-555.e6. 35. Pankiv S, Clausen TH, Lamark T, et al. p62/SQSTM1 binds directly to Atg8/LC3 to facilitate degradation of ubiquitinated protein aggregates by autophagy. J Biol Chem. 2007;282(33):24131-24145. 36. Vo MT, Smith BJ, Nicholas J, Choi YB. Activation of NIX-mediated mitophagy by an interferon regulatory factor homologue of human herpesvirus. Nat Commun. 2019;10(1):3203. 37. Liu X, Zhang Y, Ni M, et al. Regulation of mitochondrial biogenesis in erythropoiesis by mTORC1-mediated protein translation. Nat Cell Biol. 2017;19(6):626-638. 38. Moras M, Hattab C, Gonzalez-Menendez P, et al. Downregulation of mitochondrial TSPO inhibits mitophagy and reduces enucleation during human terminal erythropoiesis. Int J Mol Sci. 2020;21(23):9066. 39. Yang C, Hashimoto M, Lin QXX, Tan DQ, Suda T. Sphingosine-1-phosphate signaling modulates terminal erythroid differentiation through the regulation of mitophagy. Exp Hematol. 2019;72:47-59. 40. Gilquin B, Taillebourg E, Cherradi N, et al. The AAA+ ATPase ATAD3A controls mitochondrial dynamics at the interface of the inner and outer membranes. Mol Cell Biol. 2010;30(8):1984-1996.

41. Jin G, Xu C, Zhang X, et al. Atad3a suppresses Pink1-dependent mitophagy to maintain homeostasis of hematopoietic progenitor cells. Nat Immunol. 2018;19(1):2940. 42. Greene AW, Grenier K, Aguileta MA, et al. Mitochondrial processing peptidase regulates PINK1 processing, import and Parkin recruitment. EMBO Rep. 2012;13(4):378385. 43. Heo JM, Harper NJ, Paulo JA, et al. Integrated proteogenetic analysis reveals the landscape of a mitochondrial-autophagosome synapse during PARK2-dependent mitophagy. Sci Adv. 2019;5(11):eaay4624. 44. Hoshino A, Wang WJ, Wada S, et al. The ADP/ATP translocase drives mitophagy independent of nucleotide exchange. Nature. 2019;575(7782):375-379. 45. Traxler EA, Thom CS, Yao Y, Paralkar V, Weiss MJ. Nonspecific inhibition of erythropoiesis by short hairpin RNAs. Blood. 2018;131(24):2733-2736. 46. Lupo F, Tibaldi E, Matte A, et al. A new molecular link between defective autophagy and erythroid abnormalities in chorea-acanthocytosis. Blood. 2016;128(25):2976-2987. 47. Jagadeeswaran R, Vazquez BA, Thiruppathi M, et al. Pharmacological inhibition of LSD1 and mTOR reduces mitochondrial retention and associated ROS levels in the red blood cells of sickle cell disease. Exp Hematol. 2017;50:46-52. 48. Jiang H, Yang L, Guo L, et al. Impaired mitophagy of nucleated erythroid cells leads to anemia in patients with myelodysplastic syndromes. Oxid Med Cell Longev. 2018;2018:6328051. 49. Sbardella D, Tundo GR, Campagnolo L, et al. Retention of mitochondria in mature human red blood cells as the result of autophagy impairment in Rett syndrome. Sci Rep. 2017;7(1):12297. 50. Wu L, Xu W, Xu L, Kong Q, Fang J. Mitophagy is increased during erythroid differentiation in beta-thalassemia. Int J Hematol. 2017;105(2):162-173. 51. Li-Harms X, Milasta S, Lynch J, et al. Mitoprotective autophagy is impaired in erythroid cells of aged mtDNA-mutator mice. Blood. 2015;125(1):162-174. 52. Houwerzijl EJ, Pol HW, Blom NR, van der Want JJ, de Wolf JT, Vellenga E. Erythroid precursors from patients with low-risk myelodysplasia demonstrate ultrastructural features of enhanced autophagy of mitochondria. Leukemia. 2009;23(5):886891. 53. Koschade SE, Brandts CH. Selective autophagy in normal and malignant hematopoiesis. J Mol Biol. 2020;432(1):261282.

177

ARTICLE Ferrata Storti Foundation

Haematologica 2022 Volume 107(1):178-186

Correspondence: ELLIOT STIEGLITZ elliot.stieglitz@ucsf.edu

Myelomonocytic Leukemia

Molecular and phenotypic diversity of CBLmutated juvenile myelomonocytic leukemia Anna Hecht,1,2 Julia A. Meyer,1 Astrid Behnert,1 Eric Wong,1 Farid Chehab,3 Adam Olshen,4,5 Aaron Hechmer,4 Catherine Aftandilian,6 Rukhmi Bhat,7 Sung Won Choi,8 Satheesh Chonat,9 Jason E. Farrar,10 Mark Fluchel,11 Haydar Frangoul,12 Jennifer H. Han,13 Edward A. Kolb,14 Dennis J. Kuo,13 Margaret L. MacMillan,15 Luke Maese,16 Kelly W. Maloney,17 Aru Narendran,18 Benjamin Oshrine,19 Kirk R. Schultz,20 Maria L. Sulis,21 David Van Mater,22 Sarah K. Tasian,23 Wolf-Karsten Hofmann,2 Mignon L. Loh1,4 and Elliot Stieglitz1,4 Department of Pediatrics, Benioff Children’s Hospital, University of California San Francisco, San Francisco, CA, USA; 2Department of Hematology/Oncology, University Hospital Mannheim, Heidelberg University, Heidelberg, Germany; 3Department of Laboratory Medicine, University of California San Francisco, San Francisco, CA, USA; 4 Helen Diller Family Comprehensive Cancer Center, University of California San Francisco, San Francisco, CA, USA; 5Department of Epidemiology and Biostatistics, University of California San Francisco, San Francisco, CA, USA; 6Pediatric Hematology/Oncology, Stanford University, Stanford, CA, USA; 7Northwestern University Feinberg School of Medicine, Ann & Robert H. Lurie Children’s Hospital of Chicago, Chicago, IL, USA; 8Blood and Marrow Transplantation Program, University of Michigan, Ann Arbor, MI. USA; 9 Department of Pediatrics, Emory University School of Medicine, Aflac Cancer and Blood Disorders Center, Children’s Healthcare of Atlanta, Atlanta, GA, USA; 10Arkansas Children’s Research Institute, Department of Pediatrics, University of Arkansas for Medical Sciences, Little Rock, AR, USA; 11University of Utah, Department of Pediatrics, Division of Pediatric Hematology-Oncology, Salt Lake City, UT, USA; 12The Children's Hospital at TriStar Centennial and Sarah Cannon Research Institute, Nashville, TN, USA; 13Division of Pediatric Hematology-Oncology, University of California San Diego/Rady Children's Hospital San Diego, CA, USA; 14Nemours Center for Cancer and Blood Disorders/Alfred I. DuPont Hospital for Children, Wilmington, DE, USA; 15Blood and Marrow Transplant Program, Department of Pediatrics, University of Minnesota Medical School, Minneapolis, MN, USA; 16Department of Pediatrics and Huntsman Cancer Institute, University of Utah, Salt Lake City, UT, USA; 17Children's Hospital Colorado, Aurora, CO, USA; 18Pediatric Hematology and Oncology, Alberta Children's Hospital, Calgary, Alberta, Canada; 19Johns Hopkins All Children’s Hospital, St. Petersburg, FL, USA; 20British Columbia Children’s Hospital and Research Institute, Vancouver, British Columbia, Canada; 21Department of Pediatrics, Memorial Sloan Kettering Cancer Center, New York, NY, USA; 22Department of Pediatrics, Duke University Medical Center, Durham, NC, USA and 23Division of Oncology and Center for Childhood Cancer Research, Children’s Hospital of Philadelphia; Department of Pediatrics and Abramson Cancer Center, University of Pennsylvania Perelman School of Medicine, Philadelphia, PA, USA. 1

ABSTRACT Received: August 25, 2020. Accepted: December 21, 2020. Pre-published: December 30, 2020. https://doi.org/10.3324/haematol.2020.270595

178

utations in the CBL gene were first identified in adults with various myeloid malignancies. Some patients with juvenile myelomonocytic leukemia (JMML) were also noted to harbor mutations in CBL, but were found to have generally less aggressive disease courses compared to patients with other forms of Ras pathway-mutant JMML. Importantly, and in contrast to most reports in adults, the majority of CBL mutations in JMML patients are germline with acquired uniparental disomy occurring in affected marrow cells. Here, we systematically studied a large cohort of 33 JMML patients with CBL mutations and found that this disease is highly diverse in presentation and overall outcome. Moreover, we discovered somatically acquired CBL mutations in 15% of pediatric patients who presented with more aggressive disease. Neither clinical features nor methylation profiling were able to distinguish patients with somatic CBL mutations from those with germline CBL mutations, highlighting the need for germline testing. Overall, we demonstrate that disease courses are quite heterogeneous even among patients with germline CBL mutations. Prospective clinical trials are warranted to find ideal treatment strategies for this diverse cohort of patients.

haematologica | 2022; 107(1)

The spectrum of CBL-mutated JMML

Introduction Juvenile myelomonocytic leukemia (JMML) is a rare and aggressive disease of young children that presents with features of both myelodysplasia and myeloproliferation. To date, hematopoietic stem cell transplantation (HSCT) is the only curative therapy and results in long-term survival of about 50-60% of patients.1,2 The mutational landscape of JMML has been well described in several large studies.3-6 Nearly all patients have initiating mutations in genes that activate the Ras pathway, most commonly, PTPN11, NRAS, KRAS, NF1 or CBL. Secondary mutations are found in approximately 30% of patients at diagnosis and are associated with inferior survival. CBL mutations were first described in JMML patients in 2009 by our group and others7-10 and account for approximately 15% of all JMML cases.3,11,12 Germline tissue including cord blood and buccal samples were heterozygous for mutations, while leukemia cells were homozygous, indicating that loss of heterozygosity was required for leukemogenesis. The majority of these children presented with syndromic features including facial dysmorphia, growth retardation, cryptorchidism and autoimmune phenomena, specifically vasculitis – a condition that is now called “CBLsyndrome”. Patients with CBL-syndrome are at risk of developing JMML. Surprisingly, several of these patients with confirmed JMML experienced spontaneous resolution of their leukemia. Based on this observation and in stark contrast to other forms of JMML, CBL-syndrome JMML has been thought to be associated with relatively good prognosis. Treatment has evolved accordingly for these patients, who undergo a period of close observation of blood counts and assessment of splenomegaly, with HSCT utilized only after more conventional therapies have failed.13 In overlapping adult myelodysplastic syndromes/myeloproliferative neoplasms, somatically acquired CBL mutations occur in about 10-15% of cases and are associated with a poor prognosis. Several studies in chronic myelomonocytic leukemia (CMML), a disease of adults with many similarities to JMML, have shown that CBL mutations are associated with inferior survival.14-16 In our experience, we have observed that the spectrum of clinical courses of patients with CBL-mutated JMML is wide, ranging from spontaneous resolution to aggressive disease with transformation to acute myeloid leukemia. This observation prompted us to perform a systematic study of clinical presentations and molecular features to identify potential factors that can identify patients who may be observed and those who need immediate therapy.

Methods Patients We collected samples and clinical data from 33 patients diagnosed with JMML according to international diagnostic criteria17 who were found to have a clonal CBL mutation. The patients were diagnosed between 2006 and 2019 in 27 different centers across the USA and Canada. Treatment decisions were at the discretion of the treating physicians and were independent of this study. Written consent and specimens were obtained from all patients at the time of routine clinical assessments. The study design was reviewed and approved by the institutional review board of University of California San Francisco, in accordance with the Declaration of Helsinki.

haematologica | 2022; 107(1)

Material and studies We received bone marrow and/or peripheral blood from all patients, collected at the time of initial diagnosis or shortly thereafter. Mononuclear cells were isolated using a Ficoll method and DNA was extracted using the Gentra Puregene kit (Qiagen, Germany). For identification of genomic mutations, DNA was extracted from bone marrow or peripheral blood mononuclear cells. DNA from peripheral blood was used for DNA methylation analysis. The germline status of CBL mutations was determined using material from either a buccal swab, cord blood or after sorting CD3+ T cells from peripheral blood. A somatic mutation was defined as a mutation that was identified in leukemia tissue (peripheral blood or bone marrow), but had a minimum variant allele fraction (VAF) of <5% in any germline tissue (epithelial cells from buccal swab, cord blood, or T cells).

Next-generation sequencing of juvenile myelomonocytic leukemia-associated mutations Genomic DNA samples were sequenced using a custom amplicon-based sequencing approach (Paragon Genomics, Hayward, CA, USA) targeting 26 genes that are known to be recurrently mutated in JMML18 (Online Supplementary Table S1 and Online Supplementary Methods). A VAF of 0.05 (=5%) at diagnosis was required for reporting.

Targeted MethylSeq library preparation, sequencing and hierarchical clustering Genomic DNA (300 ng) was bisulfite converted and 100 ng of converted single-strand DNA were used as the input for a custom 3000 CpG loci targeted MethylSeq assay (Tecan). Single-index library pools were sequenced on the Illumina NovaSeq with paired-end 150 base-pair (bp) mode averaging 20x106 reads per library preparation. MethylSeq hierarchical clustering and classification were performed according to the international consensus definition12 as described in the Online Supplementary Methods.

Statistical analyses Spontaneous resolution of disease was defined as normalization of monocyte counts in the peripheral blood and reduction of spleen to normal size without any JMML-directed therapy. Persistent disease was defined as not achieving complete remission according to the International Working Group definition19 with or without therapy excluding HSCT. Overall survival was defined as time from initial diagnosis to death from any cause and was estimated using the Kaplan-Meier method. The median follow-up time was estimated using the reverse Kaplan-Meier method. Differences in clinical features between patients with germline CBL mutations and those with somatic-only CBL mutations were tested for significance using the Fisher exact test for categorical variables and the Mann-Whitney U test for continuous variables. The level of statistical significance for all tests was a Pvalue less than 0.05. All calculations were performed using Microsoft Excel (version 16.16.3), GraphPad Prism software (version 8.0) and R (version 3.4.1).

Results Patients’ characteristics The cohort consisted of 33 patients with a median age at diagnosis of 1.1 years (range, 1 month - 25 years). The median follow-up in this study was 3.5 years (range, 0.212.1 years). The patients’ characteristics at diagnosis are shown in Table 1. Additional clinical information for each patient can be found in Online Supplementary Table S2. 179

A. Hecht et al.

Mutational spectrum of CBL-mutated juvenile myelomonocytic leukemia and occurrence of secondary mutations Clonal CBL mutations were confirmed for every patient in the cohort and were exclusively located in the linker and ring finger domains, as previously reported7,16,20 (Figure 1 and Online Supplementary Table S3). Twenty-eight patients had germline alterations of CBL, including six patients with deletions affecting these domains. All but two of these patients with germline CBL mutations showed a higher VAF of the same mutation in their tumor as a result of heterozygosity (all VAF are shown in Online Supplementary Table S3). Of the two patients without heterozygosity, one had a deletion and the other a splice site mutation, compatible with previous reports in other patients.21 Interestingly, we identified five patients (15%) in our cohort with somatic-only CBL mutations. The somatic-only mutations were located at the same hotspot positions as those in the germline population and were homozygous in four patients and heterozygous in one patient. We then conducted further experiments in these five patients with somatic-only CBL mutations. To confirm that these mutations were in fact somatic in nature, buccal swabs were tested in four patients and a cord blood sample in the remaining patient. No exonic or intronic alterations of CBL or any other Ras pathway genes were found in the germline samples of these five patients by targeted deep sequencing. In one of the patients with a somatic-only CBL mutation, we discovered a heterozygous RUNX1 p.R166Q germline mutation, which has been reported in familial platelet disorder with propensity to myeloid malignancy syndrome. Overall, missense CBL mutations at residue Y371 were the most common alteration in both groups. One patient with a germline CBL mutation was found to have a secondary somatic CBL mutation with a lower VAF, leading to loss of heterozygosity. No other somatic secondary mutations were found in either cohort using our targeted 26-gene JMML panel at diagnosis.

Comparison of patients with germline or somatic-only CBL mutations Table 2 describes the clinical presentation of patients with germline CBL mutations compared to those with somaticonly CBL mutations. There were no significant differences between the two cohorts in the patients’ characteristics at diagnosis (Table 2). Of note, patients in both cohorts presented at a median age of 1 year. Clinical signs of dysmorphia could not be used to distinguish between the groups,

as 13 of 20 patients with germline CBL mutations had no overtly syndromic features (65%, data not available for 8 patients). None of the patients with somatic CBL mutations presented with obvious signs of dysmorphia. In eight of 15 (53%) patients with germline CBL mutations for whom information was available, the mutation was documented to be inherited from a parent with equal rates of maternal and paternal inheritance.

Outcome and diversity of clinical courses of CBL-mutated juvenile myelomonocytic leukemia The overall survival rate of the whole cohort after a median follow-up of 3.7 years was 41% (95% confidence interval [95% CI]: 7-75%; median survival: 7.9 years) (Figure 2A). However, 45% of the patients (15 of 33) underwent HSCT at a median time of 0.5 years after initial diagnosis (range, 0.3-2.6 years). Overall survival of patients who underwent HSCT was 69% (95% CI: 36-87%; median survival not reached) compared to 31% (95% CI: 1-74%;

Table 1. Patients’ characteristics at diagnosis.

Whole cohort (n=33) Median age at diagnosis, years 1.1 Range 0.1-25.3 Gender Male 14 Female 19 Median WBC at diagnosis, x109/L 34.2 Range 6.9-196.0 Median absolute monocyte count at diagnosis, x109/L 5.8 Range 0.8-31.1 Median platelet count at diagnosis, x109/L 68 Range 10-204 Median hemoglobin at diagnosis, g/dL 9.9 Range 6.2-11.8 Elevated hemoglobin F for age 24% (6 of 25 with data available) Abnormal cytogenetics 3% (1 of 30 with data available) Monosomy 7 0 Splenomegaly 97% (29 of 30 with data available) Dysmorphic features present 35% (7 of 20 with data available) Germline CBL mutation inheritance 53% (8 of 15 with data available) Maternal origin 4 Paternal origin 4 WBC: white blood cell count.

Figure 1. Germline and somatic CBL mutations in patients with juvenile myelomonocytic leukemia. Each dot represents one mutation found at the specified codon. The resulting change in amino acid is coded by colors. The stars represent six patients found to have deletions. 4H: four helix bundle; EF: EF-hand like domain; SH2: Src homology 2 domain; L: linker domain; RF: ring finger domain; Pro-rich: proline-rich domain; UBA/LZ: ubiquitin-associated/leucine zipper domain.

180

haematologica | 2022; 107(1)

The spectrum of CBL-mutated JMML

median survival: 7.9 years) for patients who did not undergo HSCT (Figure 2B, C). Of note, all three deaths in the non-transplanted cohort were due to other organ failure associated with CBL-syndrome unrelated to JMML. There was no difference in overall survival between the patients with germline CBL mutations and those with somatic CBL mutations (Online Supplementary Figure S1). We observed that all five patients with somatic CBL mutations were refractory to moderately-intense myeloidbased chemotherapy (i.e., 2 g/m2/day cytarabine alone or in combination with 30 mg/m2/day fludarabine).1 Three of

these five patients received an allogeneic HSCT, and two are alive and disease-free at the time of last follow-up. One relapsed with frank acute myeloid leukemia following allogeneic HSCT for JMML and died of infectious complications (Vignette 1). The clinical courses of the 28 patients with germline CBL mutations were heterogeneous. Three patients experienced spontaneous resolution of JMML without any treatment. Two patients underwent splenectomy as their only treatment but both died several years later of CBL-syndromeassociated organ failure (Vignette 2). Twelve patients received allogeneic HSCT following different pretransplant therapies. Nine of the transplanted patients went into remission after the transplant (2 after graft failure [Vignette 5 in the Online Supplementary Material and a separate case previously described by Oshrine.22]). Among the remaining three patients who were transplanted, two died of relapsed JMML and one died of transplant-related complications. Of the 11 patients who were not transplanted and did not experience spontaneous resolution, nine have persistent disease without having been transplanted to date. The median follow-up of these patients is 3.5 years and their treatments included low-dose chemotherapy (Vignette 3), moderately-intense acute myeloid leukemia-directed chemotherapy, azacitidine, and/or targeted agents. Several patients received no therapy, while others were refractory to multiple lines of therapy. One patient with multiple congenital abnormalities died from complications of multiorgan failure in infancy. One patient was lost to follow up. An overview of all disease courses is shown in Figure 3.

Methylation-based clustering of patients’ samples In order to investigate whether DNA methylation profiles are capable of distinguishing patients who will experience spontaneous resolution from those with more aggressive disease, we assessed the methylation of approximately 3,000 CpG loci using peripheral blood of each patient at diagnosis. Patients were designated as having low, intermediate or high methylation based on minimum distance to the centroid of the international consensus definition Table 2. Comparison of clinical characteristics of patients with germline or somatic-only CBL mutations.

Germline CBL Somatic CBL (n=28) (n=5)

Figure 2. Kaplan-Meier survival curves for CBL-juvenile myelomonocytic leukemia. (A) Overall survival of the whole cohort (n=33). (B) Overall survival of patients who underwent hematopoietic stem cell transplantation (n=15). (C) Overall survival of patients who did not undergo hematopoietic stem cell transplantation (n=17). One patient who was lost to follow-up was not considered in analyses for (B) or (C). HSCT: hematopoietic stem cell transplantation; w/o: without.

Median age at diagnosis, years Range Gender Male Female Median WBC at diagnosis, x109/L Range Median absolute monocyte count at diagnosis, x109/L Range Median platelet count at diagnosis, x109/L Range Median hemoglobin at diagnosis, dg/L Range Elevated hemoglobin F for age Abnormal cytogenetics Splenomegaly

P-value

1.1 0.1-25.3

1.0 0.6-3.5

0.79

12 16 31.7 6.9-196.0 5.5

2 3 43.0 24.1-88.0 6.7

0.85

0.96

0.8-31.1 62

4.4-12.0 102

0.56

10-204 9.7 6.2-11.8 18% 4% 96%

42-167 10.4 8.9-11.7 20% 0% 100%

0.58

0.34 0.91 N/A N/A

WBC: white blood cell count; N/A: not applicable.

haematologica | 2022; 107(1)

181

Figure 3. Swimmer plot showing the clinical course of each patient over time. Each bar represents one patient color-coded based on the presence of germline CBL (blue) or somatic-only CBL (orange). Dates of hematopoietic stem cell transplantation (HSCT), relapse, death or resolution of splenomegaly are depicted by symbols. The current ongoing status of the patient is depicted as a color-coded arrow. Therapeutic agents received by the patient (before HSCT if applicable) are shown on the left side with pattern-coded dots. Clinical features (age, hemoglobin F levels and platelet counts at diagnosis are depicted as filled (if true) or empty (if false) boxes. Dashed boxes are used if data were not available. Information on treatment was missing for patients UPN1125, UPN2949 and UPN 2357; patient UPN2949 was lost to follow-up. HbF: hemoglobin F; PLT: platelet count.

A. Hecht et al.

182

haematologica | 2022; 107(1)

Figure 4. DNA methylation profiles of patients with CBL-mutated juvenile myelomonocytic leukemia. (A) CBL mutant samples profiled by MethylSeq classified according to the international juvenile myelomonocytic leukemia (JMML) consensus signature. (B) DNA methylation groups low, intermediate, and high were defined using an international cohort of JMML samples described by Schönung et al.12 Each CBL mutant sample was classified into one of the three methylation groups based on minimum distance to the nearest centroid. Heatmaps (A and B) show the b values of 1,386 CpG loci used for methylation classification. LOH: Loss of heterozygosity. Two patients in the low methylation group are not depicted in (A) because their methylation data were generated with an Illumina 450k array.

The spectrum of CBL-mutated JMML

183

A. Hecht et al.

(Figure 4).1,5,12,23 Thirty-one of the 33 CBL-mutated patients were classified as having low methylation irrespective of whether their mutation was germline or somatic in nature, while the other two patients, one with a germline mutation, the other with a somatic mutation, were classified as having intermediate methylation. Notably, both patients who were classified as having intermediate methylation relapsed after HSCT. Unsupervised hierarchical clustering did not reveal any differences in methylation patterns among patients with germline CBL mutations who experienced spontaneous remission, survived more than 4 years without HSCT or died without receiving treatment. To demonstrate the wide spectrum of clinical presentations and the variability in treatment decisions we share the following vignettes.

Vignette 1: UPN2903 A previously healthy 12-month-old female infant was noted to have thrombocytopenia with a normal hemoglobin concentration and white blood cell count on a routine full blood count. The patient was asymptomatic and was observed. At 3 years of age, leukoerythroblastosis was noted on her peripheral blood smear, and her fetal hemoglobin concentration was noted to be elevated for age. Her bone marrow was normocellular for age with trilineage hematopoiesis and no dysplasia. Genetic sequencing revealed a CBL p.Y371H mutation at 70% VAF. The patient was noted to have possible facial dysmorphology. A presumptive diagnosis of germline CBL-associated JMML was made, although no germline testing was performed at that time. The patient was observed for nearly 2 years at which point her white cell count increased to 40x109/L. A repeat bone marrow examination was performed, which revealed transformation to acute myeloid leukemia with 20% blasts and a der(16)t(1;16) on cytogenetics resulting in partial deletion of 16q and partial trisomy for 1q. Repeat tumor testing revealed that the CBL mutation had decreased to 52% VAF, and additional NRAS p.G13D (22%) and KRAS p.T58I (3%) mutations were noted. The patient went on to receive various chemotherapeutic regimens, including daunorubicin, cytarabine and etoposide, mitoxantrone and cytarabine, 6-mercaptopurine, hydroxyurea, trametinib and sorafenib, but was never able to achieve remission. The patient was transplanted with 10% bone marrow blasts using busulfan, cyclophosphamide, and melphalan conditioning, but died of infection and graft-versus-host disease shortly after the transplant. Post-mortem, a cord blood sample was analyzed and confirmed the absence of a germline CBL mutation. However, a heterozygous RUNX1 p.R166Q mutation was detected in the cord blood, indicating that the patient had an autosomal dominant familial platelet disorder with propensity to myeloid malignancy syndrome.

Vignette 2: UPN2056 A 6-month-old patient with several congenital anomalies including hypoplastic kidneys, midgut malrotation, webbed neck, and choanal atresia was suspected of having Kabuki or Noonan syndrome. Over time, he displayed significant developmental delay and short stature. At 3 years of age, this patient developed a myeloproliferative disorder with features suggestive of JMML, including monocytosis, anemia, and juvenile xanthogranulomas, and fewer than 5% blasts in his bone marrow. At this time, tumor sequencing revealed a novel homozygous frameshift mutation in exon 7 of CBL. Germline testing revealed the identical heterozygous deletion. Of note, the patient had an identical twin brother who developed a myeloproliferative disorder at nearly the same time with similar congenital anomalies 184

and was found to carry the same mutation. The patient had a splenectomy at age 4 years, but his monocytosis persisted. The patient had multiple medical comorbidities, including chronic renal disease, seizures and hypothyroidism, and died of complications unrelated to JMML.

Vignette 3: UPN2977 A previously healthy 9-month-old female was evaluated for fatigue and splenomegaly. A full blood count revealed a white blood count of 52x109/L, hemoglobin concentration of 9.0 g/dL, and platelet count of 68x109/L. Bone marrow evaluation demonstrated monocytosis with normal cytogenetics. Genetic testing revealed a germline heterozygous CBL p.Y371H mutation with loss of heterozygosity in the bone marrow. She was initially observed, but subsequently failed to thrive and developed worsening splenomegaly and leukocytosis (peak WBC of 121x109/L). She was started on 6-mercaptopurine with a transient improvement in her white blood cell count. She experienced recurrent splenomegaly, leukocytosis, and thrombocytopenia, so went on to receive two cycles of fludarabine/cytarabine and underwent splenectomy. Despite these interventions, she had persistent leukocytosis. Repeat bone marrow evaluation was consistent with persistent JMML by morphology, but demonstrated new cytogenetic changes with a 3q deletion. Given her refractory JMML and cytogenetic evolution, she received four cycles of azacitidine followed by a haplo-identical peripheral blood stem cell transplant from her mother with busulfan, cyclophosphamide, and melphalan conditioning. She had engraftment failure, so received a stem cell boost and then ultimately received 5/6 unrelated umbilical cord blood transplant with fludarabine, cyclophosphamide, and total body irradiation as conditioning. Her post-transplant course was complicated by recurrent bacteremia and cytomegalovirus viremia. She is now in clinical remission more than 1 year after transplantation and is doing well.

Discussion In this study, we comprehensively analyzed the largest cohort of CBL-mutated JMML patients described to date. The clinical courses and outcomes of the 33 patients analyzed varied dramatically, and a typical “watch and wait” approach was not appropriate for all patients. Interestingly, we found that five of the 33 patients had somatically acquired CBL mutations. To our knowledge, this is the first time somatic CBL mutations have been described in JMML. All mutations, whether germline or somatic in nature, were located at the known hotspot loci in the CBL gene. No secondary mutations were found in any other JMML-associated genes in any of these patients at diagnosis. Moreover, patients with germline or somatic CBL mutations did not differ in their clinical presentation or laboratory features at diagnosis. Surprisingly, the overall survival rate of patients with germline CBL mutations in our cohort was lower than that in previous reports. One question that remains unanswered is whether transplantation for germline CBL-mutant patients can effectively decrease or prevent the known non-hematologic complications of CBL-syndrome and potentially increase overall survival. There was a trend towards improved overall survival in patients with germline mutations who underwent HSCT but transplant-related mortality remains a concern. In a prior study, five of six patients with CBL-syndrome haematologica | 2022; 107(1)

The spectrum of CBL-mutated JMML

JMML who were not transplanted experienced spontaneous resolution of their hematologic diseases.8 In our cohort, we observed spontaneous resolution in only three of the 18 non-transplanted CBL-mutated JMML patients. In particular, none of the patients with somatic-only CBL mutations experienced spontaneous resolution, which is consistent with other JMML subtypes harboring somatic Ras pathway mutations. In adults with chronic myelomonocytic leukemia, somatic CBL mutations are found in about 10-15% of cases and are associated with a higher rate of transformation to acute myeloid leukemia and inferior overall outcomes.14-16 Similarly, while germline PTPN11 mutations are associated with Noonan syndrome and a typically self-limited form of JMML,24 patients with somatic PTPN11 mutations experience aggressive leukemia and rarely, if ever, have spontaneous remission.23,25 In our cohort, all five patients with somaticonly CBL mutations were refractory to medium-intensity myeloid-based chemotherapy and required HSCT for definitive disease control. We thus maintain that patients with somatic-only CBL-mutated JMML should not be observed without therapeutic intervention, but instead should be treated with HSCT. Furthermore, somatic-only testing of JMML samples is insufficient for proper identification of these patients, and dedicated germline testing should also be performed in all patients with CBL mutations. Aberrant DNA methylation profiles have shown to be a promising tool for risk stratification in JMML.5,11,23 We therefore assessed the methylome of each patient at diagnosis. Four of the patients had been assessed for methylation in previous methylation studies (UPN1333, UPN2056, UPN2178, UPN2309) that utilized an arraybased approach. The current study assessed methylation status using a sequencing-based approach and all four patients were found to have the same methylation designation as in the previous studies. All but two of the 33 patients clustered with the low-methylation group and both of the patients with intermediate methylation relapsed after HSCT. However, methylation analysis was not able to distinguish between patients who experienced spontaneous remission and those with persistent disease. At this time, methylation analysis does not appear to be a biomarker capable of predicting outcome in patients with CBL-mutated JMML, in contrast to other JMML subtypes. The optimal treatment for patients who require therapy due to splenomegaly or cytopenias is still unclear. The patients reported in this manuscript were diagnosed between 2006 and 2019. The optimal management of CBL-mutated patients with JMML has evolved over time, and no consensus currently exists given the relatively recent identification of this JMML subtype and the heterogeneity of its presentation. Several experimental studies have hinted at the potential use of targeted therapies in CBL-mutated disease. Mutant CBL is known to engage LYN/PI3K/AKT and JAK2 proteins, and targeted kinase inhibitors of these signaling pathways have been proposed as potential therapeutic strategies.26,27 Preclinical

haematologica | 2022; 107(1)

studies have specifically investigated the SRC/LYN inhibitor dasatinib and have shown decreased sensitivity of CBL-mutant JMML cells to granulocyte-macrophage colony-stimulating factor and increased chemotherapy sensitivity.28,29 Another preclinical study using patientderived induced pluripotent stem cells with CBL mutations reported potential efficacy of JAK2 and PI3K/mTOR signaling inhibition and decreased granulocytemacrophage colony-stimulating factor hypersensitivity.30 To date, there have been no prospective clinical trials using these agents in JMML. Further research and international collaboration are warranted to investigate the potential use of targeted therapies in patients with CBLmutated JMML, as a “watch and wait” approach is likely to be effective in only a minority of patients. Disclosures No conflicts of interest to disclose. Contributions ES and MLL designed the study. AH, JAM, AB and EW performed experiments. AH, JAM, AO and AH analyzed the data and performed statistical tests. FC provided additional germline samples. CA, RB, SWC, SC, JEF, MF, HF, JHH, EAK, DJK, MLM, LM, KWM, AN, BO, KRS, MLS, DVM and SKT provided patients’ samples and clinical data. CA and MLS wrote additional patient vignettes. AH, SKT, WKH, MLL and ES wrote the manuscript. ES supervised the study. All authors read and agreed to the final version of the manuscript. Acknowledgments We thank the patients and parents who participated in this study, who made this research possible. We also acknowledge the following physicians who referred patients to our study and without whom this work would not have been possible: Dr. David Becton and Dr. Kimo Stine (Arkansas Children’s Hospital), Dr. Elizabeth Raetz (New York University, Langone), Dr. Paul Shaughnessy (Texas Transplant Institute) and Dr. Alan Ikeda (Children's Specialty Center of Nevada). Funding This work was supported by the National Institutes of Health, National Cancer Institute grants 1U54CA196519 (to MLL, ES); 1U01CA232486 (SKT); 1K08CA184418 (SKT); National Institutes of Health, National Heart, Lung, and Blood Institute grant K08HL135434 (ES); the Leukemia Lymphoma Society (MLL, ES); Cookies for Kids Cancer (MLL); Pediatric Cancer Research Foundation (ES); the V Foundation (ES); the UCSF Catalyst Program (ES); the California Cancer League (ES); the Frank A. Campini Foundation (MLL and ES); the Children’s Hospital of Philadelphia Center for Childhood Cancer Research (SKT) and the German Research Foundation HE 7682/1-1 (AH). Next-generation sequencing was supported by the Center for Advanced Technologies at UCSF and the Computational Biology and Informatics group at the UCSF HDFCCC (supported by NCI grant: 5P30CA082103). MLL is the Benioff Chair of Children’s Health and the Deborah and Arthur Ablin Endowed Chair for Pediatric Molecular Oncology at Benioff Children’s Hospital.

185

A. Hecht et al.

References 1. Stieglitz E, Ward AF, Gerbing RB, et al. Phase II/III trial of a pre-transplant farnesyl transferase inhibitor in juvenile myelomonocytic leukemia: a report from the Children's Oncology Group. Pediatr Blood Cancer. 2015;62(4):629-636. 2. Dvorak CC, Satwani P, Stieglitz E, et al. Disease burden and conditioning regimens in ASCT1221, a randomized phase II trial in children with juvenile myelomonocytic leukemia: a Children's Oncology Group study. Pediatr Blood Cancer. 2018;65(7): e27034. 3. Stieglitz E, Taylor-Weiner AN, Chang TY, et al. The genomic landscape of juvenile myelomonocytic leukemia. Nat Genet. 2015;47(11):1326-1333. 4. Caye A, Strullu M, Guidez F, et al. Juvenile myelomonocytic leukemia displays mutations in components of the RAS pathway and the PRC2 network. Nat Genet. 2015;47(11):1334-1340. 5. Lipka DB, Witte T, Toth R, et al. RAS-pathway mutation patterns define epigenetic subclasses in juvenile myelomonocytic leukemia. Nat Commun. 2017;8(1):2126. 6. Murakami N, Okuno Y, Yoshida K, et al. Integrated molecular profiling of juvenile myelomonocytic leukemia. Blood. 2018;131 (14):1576-1586. 7. Loh ML, Sakai DS, Flotho C, et al. Mutations in CBL occur frequently in juvenile myelomonocytic leukemia. Blood. 2009;114 (9):1859-1863. 8. Niemeyer CM, Kang MW, Shin DH, et al. Germline CBL mutations cause developmental abnormalities and predispose to juvenile myelomonocytic leukemia. Nat Genet. 2010;42(9):794-800. 9. Shiba N, Kato M, Park MJ, et al. CBL mutations in juvenile myelomonocytic leukemia and pediatric myelodysplastic syndrome. Leukemia. 2010;24(5):1090-1092. 10. Perez B, Mechinaud F, Galambrun C, et al. Germline mutations of the CBL gene define a new genetic syndrome with predisposition to juvenile myelomonocytic leukaemia. J Med Genet. 2010;47(10):686-691. 11. Murakami N, Okuno Y, Yoshida K, et al.

186

Integrated molecular profiling of juvenile myelomonocytic leukemia. Blood. 2018;131 (14):1576-1586. 12. Schönung M, Meyer J, Nöllke P, et al. International consensus definition of DNA methylation subgroups in juvenile myelomonocytic leukemia. Clin Cancer Res. 2021;27(1):158-168. 13. Locatelli F, Niemeyer CM. How I treat juvenile myelomonocytic leukemia. Blood. 2015;125(7):1083-1090. 14. Duchmann M, Yalniz FF, Sanna A, et al. Prognostic role of gene mutations in chronic myelomonocytic leukemia patients treated with hypomethylating agents. EBioMedicine. 2018;31:174-181. 15. Itzykson R, Kosmider O, Renneville A, et al. Prognostic score including gene mutations in chronic myelomonocytic leukemia. J Clin Oncol. 2013;31(19):2428-2436. 16. Jankowska AM, Makishima H, Tiu RV, et al. Mutational spectrum analysis of chronic myelomonocytic leukemia includes genes associated with epigenetic regulation: UTX, EZH2, and DNMT3A. Blood. 2011;118(14): 3932-3941. 17. Chan RJ, Cooper T, Kratz CP, Weiss B, Loh ML. Juvenile myelomonocytic leukemia: a report from the 2nd International JMML Symposium. Leuk Res. 2009;33(3):355-362. 18. Hecht A, Meyer J, Chehab FF, et al. Molecular assessment of pretransplant chemotherapy in the treatment of juvenile myelomonocytic leukemia. Pediatr Blood Cancer. 2019;66(11):e27948. 19. Niemeyer CM, Loh ML, Cseh A, et al. Criteria for evaluating response and outcome in clinical trials for children with juvenile myelomonocytic leukemia. Haematologica. 2015;100(1):17-22. 20. Kales SC, Ryan PE, Nau MM, Lipkowitz S. Cbl and human myeloid neoplasms: the Cbl oncogene comes of age. Cancer Res. 2010;70(12):4789-4794. 21. Strullu M, Caye A, Cassinat B, et al. In hematopoietic cells with a germline mutation of CBL, loss of heterozygosity is not a signature of juvenile myelo-monocytic leukemia. Leukemia. 2013;27(12):2404-2407. 22. Oshrine B. Primary graft failure but treatment success: a case of reversion to het-

erozygosity after allogeneic hematopoietic cell rransplantation with autologous hematopoietic recovery in a child with CBLrelated juvenile myelomonocytic leukemia. J Pediatr Hematol Oncol. 2021;43(3):e426e428. 23. Stieglitz E, Mazor T, Olshen AB, et al. Genome-wide DNA methylation is predictive of outcome in juvenile myelomonocytic leukemia. Nat Commun. 2017;8(1):2127. 24. Choong K, Freedman MH, Chitayat D, Kelly EN, Taylor G, Zipursky A. Juvenile myelomonocytic leukemia and Noonan syndrome. J Pediatr Hematol Oncol. 1999;21(6):523-527. 25. Matsuda K, Nakazawa Y, Iwashita C, et al. Myeloid progenitors with PTPN11 and nonRAS pathway gene mutations are refractory to treatment with 6-mercaptopurine in juvenile myelomonocytic leukemia. Leukemia. 2014;28(7):1545-1548. 26. Javadi M, Richmond TD, Huang K, Barber DL. CBL linker region and RING finger mutations lead to enhanced granulocytemacrophage colony-stimulating factor (GMCSF) signaling via elevated levels of JAK2 and LYN. J Biol Chem. 2013;288(27):1945919470. 27. Belizaire R, Koochaki SHJ, Udeshi ND, et al. CBL mutations promote activation of PI3K/AKT signaling via LYN kinase. bioRxiv. 2020 Apr 14. doi: 10.1182/ blood.2020006528[preprint, not peerreviewed]. 28. Bunda S, Kang MW, Sybingco SS, et al. Inhibition of SRC corrects GM-CSF hypersensitivity that underlies juvenile myelomonocytic leukemia. Cancer Res. 2013;73(8):2540-2550. 29. Bunda S, Qin K, Kommaraju K, Heir P, Ohh M. Juvenile myelomonocytic leukaemiaassociated mutation in Cbl promotes resistance to apoptosis via the LynPI3K/AKT pathway. Oncogene. 2015;34 (6):789-797. 30. Tasian SK, Casas JA, Posocco D, et al. Mutation-specific signaling profiles and kinase inhibitor sensitivities of juvenile myelomonocytic leukemia revealed by induced pluripotent stem cells. Leukemia. 2019;33(1):181-190.

haematologica | 2022; 107(1)

ARTICLE

Non Hodgkin-Lymphoma

Micro-RNA networks in T-cell prolymphocytic leukemia reflect T-cell activation and shape DNA damage response and survival pathways

Ferrata Storti Foundation

Till Braun,1* Markus Glaß,2* Linus Wahnschaffe,1* Moritz Otte,1 Petra Mayer,1 Marek Franitza,3 Janine Altmüller,3 Michael Hallek,1 Stefan Hüttelmaier,2# Alexandra Schrader,1# and Marco Herling1# Department I of Internal Medicine, Center for Integrated Oncology (CIO), Aachen-BonnCologne-Duesseldorf, Excellence Cluster for Cellular Stress Response and AgingAssociated Diseases (CECAD), Center for Molecular Medicine Cologne (CMMC), University of Cologne (UoC), Cologne; 2Institute of Molecular Medicine, Section for Molecular Cell Biology, Faculty of Medicine, Martin Luther University Halle-Wittenberg, Charles Tanford Protein Center, Halle, and 3Cologne Center for Genomics, Center for Molecular Medicine Cologne (CMMC), University of Cologne, Cologne, Germany 1

Haematologica 2022 Volume 107(1):187-200

*TB, MG and LW contributed equally as co-first authors. #

SH, AS and MH contributed equally as co-senior authors.

ABSTRACT

-cell prolymphocytic leukemia (T-PLL) is a poor-prognostic mature T-cell malignancy. It typically presents with exponentially rising lymphocyte counts, splenomegaly, and bone marrow infiltration. Effective treatment options are scarce and a better understanding of TPLL’s pathogenesis is desirable. Activation of the TCL1 proto-oncogene and loss-of-function perturbations of the tumor suppressor ATM are TPLL’s genomic hallmarks. The leukemic cell reveals a phenotype of active T-cell receptor (TCR) signaling and aberrant DNA damage responses. Regulatory networks based on the profile of microRNA (miR) have not been described for T-PLL. In a combined approach of small-RNA and transcriptome sequencing in 46 clinically and moleculary well-characterized T-PLL, we identified a global T-PLL-specific miR expression profile that involves 34 significantly deregulated miR species. This pattern strikingly resembled miR-ome signatures of TCR-activated T cells. By integrating these T-PLL miR profiles with transcriptome data, we uncovered regulatory networks associated with cell survival signaling and DNA damage response pathways. Despite a miR-ome that discerned leukemic from normal T cells, there were also robust subsets of T-PLL defined by a small set of specific miR. Most prominently, miR-141 and the miR200c-cluster separated cases into two major subgroups. Furthermore, increased expression of miR-223-3p as well as reduced expression of miR-21 and the miR-29 cluster were associated with more activated Tcell phenotypes and more aggressive disease presentations. Based on the implicated pathobiological role of these miR deregulations, targeting strategies around their effectors appear worth pursuing. We also established a combinatorial miR-based overall survival score for T-PLL (miROS-T-PLL), that might improve current clinical stratifications.

Introduction T-cell prolymphocytic leukemia (T-PLL) is a neoplasm of post-thymic T cells.1 It represents the most frequent mature T-cell leukemia in Western countries, however, with an incidence of 2 per million per year, it is still classified as an orphan disease.2 T-PLL patients typically present at a median age of about 65 years, with exponentially rising lymphocytosis, marked bone marrow infiltration, and splenomegaly.3 Phenotypically, T-PLL cells resemble mature, antigenexperienced T cells.4,5 The aggressive growth of T-PLL cells is paralleled by a refractory behavior towards most conventional chemotherapies.6 The current

haematologica | 2022; 107(1)

Correspondence: MARCO HERLING Marco.Herling@medizin.uni-leipzig.de Received: July 20, 2020. Accepted: November 6, 2020. Pre-published: November 19, 2020. https://doi.org/10.3324/haematol.2020.267500

187

T. Braun et al.

treatment of choice, the CD52-antibody alemtuzumab, is efficient in inducing initial responses, but nearly all patients relapse within 12-24 months thereafter.7,8 Second-line options are even less efficient and the median overall survival (OS) of T-PLL patients is <2 years.1,3,9,10 Activating translocation of the T-cell leukemia/lymphoma 1 (TCL1) proto-oncogene is the most prevalent genetic aberration in T-PLL.11 In addition, loss-of-function perturbations of the tumor suppressor ataxia telangiectasia mutated (ATM) are reported for >80% of T-PLL patients.1 Both alterations contribute towards a phenotype of enhanced T-cell receptor signaling (TCR) and an aberrant DNA damage response, including resistance to p53-mediated cell death, deregulated cell cycle control, and deficient DNA repair mechanisms.1,11 Activating lesions in janus kinase (JAK) and signal transducer and activator of transcription (STAT) molecules as well as epigenetic aberrations have emerged as further hallmarks of T-PLL pathology, resulting in a sustained survival signaling and pro-oncogenic cell cycle deregulation.1,12–15 Despite these recent advances, a better understanding of T-PLL's pathobiology is of importance in order to identify novel treatment options. MicroRNA (miR) have increasingly been recognized as relevant in the pathogenesis of hematopoetic and solid tumors. They are small non-coding RNA with an average length of 22 nucleotides. By targeting specific mRNA, miR function as posttranscriptional repressors.16 Importantly, most miR regulate a large set of genes, often resulting in a cooperative effect on a given cellular pathway, rather than a specific effect on a single gene. Both onco miR and tumor-suppressive miR have been causally implicated in mature B- and T-lymphoid malignancies.17– 19 As a prominent example, chronic lymphocytic leukemia (CLL) harbors an unique miR expression signature with miR-181b downregulation as the best investigated miR deregulation.20 When overexpressing miR181b in the Em-TCL1A CLL mouse model, leukemic expansion is decelerated.21 Moreover, miR-181b as well as the miR-29 and miR-34b/c were shown to target the proto-oncogene TCL1A, reflected by an association of their downregulation with oncogenic TCL1A overexpression in CLL.22 Likewise, specific miR have been identified to be involved in the pathogenesis of mature T-cell tumors such as cutaneous T-cell lymphoma (CTCL) (e.g., deregulation of miR-29 and miR-200)17,23 or NK/T-cell lymphoma (downregulated miR-150).24 In T-PLL, frequent genomic aberrations of argonaute RISC catalytic component 2 (AGO2), a master regulator of miR processing,25 provide first hints for altered miR activity and miR expression signatures (miR-omes).1 However, global miR deregulations, likely involved in T-PLL’s pathophysiology, have not been reported. In the presented study, we performed small-RNA sequencing to investigate the spectrum of differential miR expression in T-PLL. We identify global, T-PLL-specific miR alterations associated with gene signatures affiliated to functional categories of survival signaling and DNA damage response pathways. In addition, we show that the miR-omes of T-PLL cells and of activated T cells are remarkably similar. Finally, we identify associations of miR alterations with cellular activation, clinical tumor burden, and patient outcome, all underlining the impact of miR deregulations in T-PLL.

188

Methods Patient cohort Primary isolates of 48 well-annotated T-PLL patients and of T cells from six age-matched healthy donors were studied (banked 2009-2019; patient characteristics in Table 1). The diagnosis of TPLL was confirmed according to World Health Organization criteria26 and consensus guidelines.2 All patients (median age 68 years) provided informed consent according to the Declaration of Helsinki. Collection and use of the samples have been approved for research purposes by the ethics committee of the University Hospital of Cologne (#11–319). Most samples (82.6%) were collected prior to any first-line treatment (n=38 of 46).

Sequencing and data processing RNA from peripheral blood mononuclear cells (PBMC) of T-PLL patients (median purity 95.4%) and CD3+ pan-T cells of six agematched healthy controls (median purity 90.2%) was subjected to library preparation and sequenced on the NovaSeq 6000 (n=48 TPLL) and the HiSeq4000 platform (n=46 T-PLL, Illumina, San Diego, USA) according to the manufacturer’s instructions for polyA-RNA and small-RNA sequencing, respectively. Details on cell isolation, stimulation, RNA isolation, library preparation, sequencing, and data processing are given in the Online Supplementary Methods.

Gene set enrichment analysis Gene set enrichment analysis (GSEA) were performed on preranked lists using the GSEA-software (v3.0)27 and MSigDB (v7.0)28 HALLMARK gene sets. For each considered miR, Spearman correlation coefficients for this miR and all protein-coding genes were determined by comparing the respective gene's count per million (CPM) and fragments per kilobase of million (FPKM) mapped reads. Sorted lists of correlation coefficients were then used as input for the GSEA.

MiR target prediction In order to obtain putative mRNA targets for each miR, predicted miR bindings were first determined using the R-package multiMiR (v1.6.0, Database Version 2.3.0),29 all of eight prediction databases (diana_microt, elmmo, microcosm, miranda, mirdb, pictar, pita, targetscan), and a 20% default prediction cutoff. All bindings predicted by <2 different databases were removed. From the remaining predicted genes, we chose those as putative miR targets that showed a negative Spearman correlation (rho<0; P<0.05; false discovery rate [FDR] <0.25) of their expression values with the expression of the respective miR.

Correlations with clinical data In order to test for associations of miR-223-3p, miR-21, miR-29, and miR-200c/141 with clinical characteristics, cytogenetics, immunophenotypes, and outcome data, cases were divided into groups by the mean or tertiles as cutoffs according to the distribution of expression values within the patient cohort. Further details on statistics are provided in the Online Supplementary Methods.

Survival score In order to develop a survival score, we (i) randomly divided our cohort into a training set (n=22) and a validation set (n=22). We then (ii) identified miR that were expressed in at least 80% of TPLL samples and (iii) that were highly associated with OS in the training set (upper tertile of patients with highest vs. tertile with the lowest expression) using log-rank tests. Several other parameters that had been described to be prognostically relevant in T-PLL (e.g., leukocyte counts, TCL1 mRNA expression) were added to

haematologica | 2022; 107(1)

T-cell-receptor-activity shaped miR-omes of T-PLL

Table 1. Clinical, cytogenetic and immunophenotypic characteristics of analyzed T-cell prolymphocytic leukemia (n=46 cases analyzed by smallRNA sequencing).

Patients‘ characteristics Median age, years (range)1 Sex Median OS from diagnosis, months (range)1

68 (32-88); n = 46 Male = 25 Female = 21 17.1 (0.4-98.7); n = 44

Clinical presentation

At diagnosis

Median WBC count, x10 /L (range) Median hemoglobin, g/dL (range)1 Median platelet count, x109/L (range)1 Median LDH, U/L (range)1 Splenomegaly (%)2 Hepatomegaly (%)2 Lymphadenopathy (%)2 9

At sample

91.4 (16.7-825.2); n = 38 12.6 (6.6-16.6); n = 34 129 (33-438); n = 34 567 (178-9423); n = 29 n = 18/29 (62.07) n = 5/27 (18.52) n = 15/26 (57.69)

114.1 (20.8-756.2); n = 43 12.6 (6.2-15.6); n = 36 118 (48-394); n = 36 795 (226-8634); n = 36

Cytogenetic features inv(14)(q11;q32) (%)2 t(14;14)(q11;q32) (%)2 t(X;14)(q28;q11) (%)2 TCR gene rearrangement3 (%)2 MYC amplification3 (%)2 ATM deletion3 (%)2

n = 27/36; (75.00) n = 3/35; (8.57) n = 2/36; (5.56) n = 37/39; (94.87) n = 23/27; (85.19) n = 15/33; (45.45)

Immunophenotype TCL1 (%)2 CD3+ (%)2 CD5+ (%)2 CD7+ (%)2 CD4+/CD8- (%)2 CD4-/CD8+ (%)2 CD4+/CD8+ (%)2

n = 35/38; (92.11) n = 36/40; (90.00) n = 40/40; (100.00) n = 39/40; (97.50) n = 27/39; (69.23) n = 7/39; (17.95) n = 5/39; (12.82)

1 Range reaches from lowest value to highest value in the cohort; 2percentages are out of total cases with sufficient data; 3evaluated by fluorescence in situ hybridization. OS: overall survival; TCR: T-cell receptor; WBC: white blood cell; LDH: serum lactate dehydrogenase.

the resulting miR with the strongest single OS associations (lowest P-values) towards a multi-parameter training model. Next, (iv) we subjected all parameters to a recursive partitioning algorithm using the rpart R package (v.4.1-15) to identify optimum individual cutoffs in our training set. Such deprioritizations finally retained the four factors miR-200a-3p, miR-223-3p, miR-424-5p, and TCL1A, for which optimum cutoffs best allowed discriminations of T-PLL patients with shorter versus longer OS. We then (v) built multivariate scores for all possible combinations of these four parameters, adding one point to the total score if the respective expression cutoff was passed, and calculated optimum thresholds for these scores. We (vi) selected the score which allowed best discrimination of OS in our training set and (vii) verified our score in the validation set as well as in the total cohort of 44 T-PLL patients.

Results Global T-cell prolymphocytic leukemia-specific microRNA (miR) deregulations highlight differential expression of miR-200c and miR-141 clusters In order to investigate the spectrum of cellular miR expressed in T-PLL, small-RNA sequencing was performed of peripheral blood (PB)-isolated tumor cells from 46 T-PLL patients and of pan-T cells of PB from six healthy donors. Thaematologica | 2022; 107(1)

PLL patient characteristics are presented in Table 1 and sample purities in the Online Supplementary Figure S1. As T-PLL cases show a spectrum of (often nonconventional) memory T-cell phenotypes and of small naïve subsets,5 we chose age-matched CD3+ pan-T cells as controls (reflecting a representative mix of populations) in these global profiling analyses. In total, we identified 2,094 miR, of which 37 miR displayed a differential expression in T-PLL versus healthy donor T cells (q<0.05, Online Supplementary Table S1). Of these, 14 miR sequences were upregulated (0.7% of all identified miR) and 23 were downregulated in T-PLL (1.1%, Figure 1A). While miR-6724-5p (fold change [fc]=0.18, q<0.0001) and miR-206 (fc=0.04, q<0.0001) showed the strongest downregulation, miR-5699-3p (fc=122, q=0.02), miR-200c-3p (fc=38.2, q=0.005), and miR-141-3p (fc=43.2, q=0.005) were the most upregulated. Considering all T-PLL cases, miR-141-3p and miR-21-5p showed the highest absolute abundance while miR-206, miR-651-3p and miR6774-5p displayed the lowest absolute expression (Online Supplementary Figure S2). Among the 37 deregulated miR, miR-6724-5p was annotated four times due to its expression from four different genomic loci. The following analyses are, therefore, based on 34 miR, containing a sum 189

T. Braun et al.

Figure 1. Global microRNA-deregulations highlight differential expression of miR-141 and miR-200c clusters in T-cell prolymphocytic leukemia. (A) MicroRNA (miR) profiles were evaluated using small-RNA sequencing of primary human peripheral blood-derived T-cell prolymphocytic leukemia (T-PLL) cells (n=46 cases) and CD3+ healthy donor-derived pan-T cell controls (n=6 donors). Volcano plot showing fold changes (fc) and false discovery rates (FDR) of all identified miR (n=2,094). Differentially expressed miR (n=37, Online Supplementary Table S1) are highlighted in blue (downregulation) and red (upregulation). Relative proportions are based on all identified miR. (B) Principle component analysis (PCA) based on miR differentially expressed comparing T-PLL cases (n=46) to healthy donor controls (n=6). Separate clustering of cases and controls indicates global differences in miR expression profiles. (C) Heatmaps of differentially expressed miR (n=34; FDR<0.05) in T-PLL samples vs. CD3+ pan-T cells of healthy donors. Left two-column heatmap: mean counts per million (CPM) values compared between healthy donor controls (n=6) and T-PLL (n=46; red=higher expression; blue=lower expression). Right heatmap: colors represent z-scores of respective CPM values calculated for each miR (red=higher z-score; blue=lower z-score). Control samples with slightly lower purities (77% and 85%) and T-PLL cases presenting a rather unique transcriptome, which clusters closer to control T cells (Online Supplementary Figure S4C), are indicated by asterixis (brown: T-PLL cases; green: control cases).

190

haematologica | 2022; 107(1)

T-cell-receptor-activity shaped miR-omes of T-PLL

expression for miR-6724-5p. In order to validate our results from small-RNA sequencing, we performed quantitative real-time polymerase chain reaction (qRT-PCR) analyses of three highly deregulated miR (miR-223-3p, miR-200c-3p, miR-141-3p, Online Supplementary Figure S3A to C) in eight T-PLL and four healthy donor T-cell controls. A strong correlation between the results from small-RNA sequencing and qRT-PCR (r2=0.84, P<0.0001, Pearson, Online Supplementary Figure S3D) confirmed differential (over)expression of these miR in T-PLL and underlines the robustness of the sequencing data. Using a previously published independent data set from single nucleotide polymorphism (SNP) arrays of 83 T-PLL1 (overlap of 23 cases with the cohort presented here), we identified only small fractions of cases to carry genomic losses of significantly downregulated miR: miR-140-3p, miR-196b-5p (both CN<1.5 in 4.82% of cases), miR-339-3p, and miR-589-5p (both CN<1.5 in 8.43%, Online Supplementary Table S2). Unsupervised clustering by principal component analysis (PCA) based on miR differentially expressed in T-PLL versus healthy donor T cells indicated a homogeneity across T-PLL samples and confirmed the global differences in the miR profiles between T-PLL and healthy controls (Figure 1B). Interestingly, unsupervised hierarchical clustering analysis of miR expression revealed two clusters of T-PLL cases, which were distinguished by expression of miR-200c and miR-141 family members (Figure 1C): 23 T-PLL cases showed low miR-141/-200c expression as compared to CD3+ pan-T cell controls whereas 23 T-PLL samples had higher than normal T-cell expression of miR-141 and -200c. Besides higher serum lactate dehydrogenase (LDH) levels (P=0.03, Mann-Whitney-Wilcoxon test [MWW]) and a lower incidence of TP53 deletions (by fluorescence in situ hybridisation [FISH], P=0.02, Fisher’s exact test) in the miR141/-200c high-expressing cohort, we did not find other differences between these two subsets (Online Supplementary Table S3). Comparing the transcriptomes of cases allocated to these two separate clusters, we identified 356 genes to be differentially expressed (Online Supplementary Table S4). In line with GSEA based on miR141/200c-correlated genes (see following analyses), we identified the HALLMARK pathways E2F TARGETS (normalized enrichment score [NES]=4.38, q<0.0001) and G2M CHECKPOINT (NES=2.68, q<0.0001) as significantly altered between the two clusters of T-PLL. Furthermore, global miR-ome profiles were not associated with distinct cellular immunophenotypes, e.g., neither with CD45RA/RO expression (i.e., “memory-like” vs. “naïve-like” T-PLL) nor with CD4/8 expression (Online Supplementary Figures S4A and B).

MicroRNA-profiles of T-cell prolymphocytic leukemia resemble those of T-cell receptor signaling-activated T cells and form regulatory networks around nodes of DNA damage response and prosurvival signaling In order to align the miR-ome data with those of global transcriptome alterations, polyA-RNA sequencing was performed on PB-isolated tumor cells from 41 miR-characterized T-PLL patients and seven additional T-PLL patients as well as on CD3+ PB pan-T cells from six age-matched healthy donors. In total, we detected 948 protein-encoding mRNA to be differentially expressed (q<0.05, Online Supplementary Figure S5A; Online Supplementary Table S5). Using this set of deregulated genes, PCA corroborated homogeneity among T-PLL cases and a clear distinction to normal T-cell controls (Online Supplementary Figure S5B). In haematologica | 2022; 107(1)

accordance with published data, TCL1A (fc=1843, q<0.0001) and CTLA4 (fc=0.06, q<0.0001) were among the most differentially expressed genes in T-PLL versus T-cell controls (Online Supplementary Figure S5C). In their transcriptome profiles two T-PLL clustered closer to control T cells than the bulk of cases. However, their miR expression signature (Figure 1C) and clinical presentation did not differ from the overall cohort. GSEA (HALLMARK gene sets27) determined 34 gene sets as upregulated in T-PLL when compared to T-cell controls, of which 19 gene sets were significantly enriched at a FDR of <5%. Sixteen gene sets were downregulated in T-PLL (11 with FDR <0.05, Online Supplementary Table S6). The identified significantly altered pathways associated with cancer and/or immunology are presented in Figure 2A. These included several HALLMARK gene sets reflecting dysregulations in DNA damage response pathways (e.g., DNA REPAIR: NES=-2.18, q=0.005; E2F TAGETS: NES=2.05, q=0.01) and prosurvival signaling (e.g., INFLAMMATORY RESPONSE, NES=3.12, q<0.0001; TNFA SIGNALING VIA NFKB, NES=2.44, q=0.001), in line with previously published data.1 As T-PLL cells generally display a mature, T-cell activated phenotype,1,11 we investigated whether T-PLL cell miRomes resemble those of TCR-activated healthy donorderived T cells. For that, PBMC (to avoid direct manipulation of T cells) of four healthy donors were cultured for 72 hours with and without stimulation by anti-CD3/CD28 crosslinking, followed by miR sequencing of CD3+-enriched cells. Sample purities and experimental controls are shown in the Online Supplementary Figure S6. We identified 56 miR which are differentially expressed in response to TCR activation (q<0.05, Online Supplementary Figure S7A; Online Supplementary Table S7). PCA indicated homogeneity within both groups (T-PLL and normal PBMC) as well as global differences between their TCR-induced miR profiles (Online Supplementary Figure S7B). We identified miR known to be affected by TCR activation (e.g., miR-150-5p)31 as well as previously unreported miR (e.g., miR-18a-5p; Online Supplementary Figure S7C). Integrative PCA based on differentially expressed miR in T-PLL versus healthy controls (Figure 1C) showed that the stimulated (over unstimulated) T cells clustered closer to T-PLL (Figure 2B). Fittingly, unsupervised clustering comparing TCR-stimulated T cells to unstimulated controls confirmed that the miR profiles of TPLL cells resemble the miR-ome of TCR-activated healthy donor-derived T cells (Figure 2C). We next assessed implicated functional relationships by predicted mRNA targets for each deregulated miR in T-PLL. For that, we (i) ranked mRNA based on their degree of correlation with a specific miR and (ii) performed HALLMARK set GSEA on these ranked mRNA. Pathways reflecting dysregulations of DNA damage response and prosurvival signaling emerged as predominantly associated with the alterations of miR expression. Exemplary HALLMARK pathways are shown in Figure 3A, a full list of gene sets is displayed in Online Supplementary Figure S8. For example, we obtained highly significant NES for the E2F TARGET and the IL2 STAT5 SIGNALING HALLMARK gene sets for (i) the transcriptome of T-PLL as compared to the one of healthy controls (E2F TARGETS: NES=1.92, q=0.02; IL2 STAT5 SIGNALING: NES=-2.86, q<0.0001, Online Supplementary Table S6) and (ii) for most of the miR differentially expressed in T-PLL (Figure 3A). Overall, there was a striking similarity of the miR pro191

T. Braun et al.

Figure 2. Global alterations of T-cell prolymphocytic leukemia microRNA expression signatures/ transcriptome networks resemble those of T-cell receptor-activated T cells. (A) Gene set enrichment analysis (GSEA) of differentially expressed mRNA in T-cell prolymphocytic leukemia (T-PLL) using HALLMARK gene sets (n=948 genes, Online Supplementary Table S6; Online Supplementary Figure S5; total n=48 gene sets). Gene expression profiles were assessed using global mRNA sequencing of primary TPLL cells (n=48 cases) and healthy donor-derived CD3+ pan-T cells (n=6 donors). Color code represents assignment to dysregulation to either altered DNA damage response pathways (black) or prosurvival signaling (grey). Exemplary GSEA plots are presented (DNA REPAIR: normalized enrichment score [NES]=-2.18, q=0.008, E2F TARGETS: NES=2.05, q=0.01, INFLAMMATORY RESPONSE: NES=3.12, q<0.0001). (B and C) T-PLL miR-omes resembled those of activated healthy donor T cells: age-matched healthy donor-derived peripheral blood mononuclear cells (PBMC) were isolated via density gradient centrifugation. T-cell activation was achieved via antibody-mediated CD3/CD28 crosslinking. After 72 hours (hrs), CD3+ primary T cells were isolated by magnetic-activated cell sorting (MACS) (negative selection; see Method section for details; cell purities are given in Online Supplementary Figure S6A, see Online Supplementary Figure S6B and C for control experiments on stimulation). MiR profiles were generated using small-RNA sequencing. Unstimulated cultured controls clustered together with directly isolated control samples, indicating that there was a negligible cell culture effect on miR expression profiles. Color code: T-PLL in brown, controls in green colors (light green: CD3+ T cells cultured in vitro for 72 hrs. without stimulation; green: CD3+ T cells submitted to miR-ome sequencing directly after enrichment; dark-green: CD3+ T-cell cultures at 72 hrs. subsequent to T-cell receptor [TCR] activation). (B) PCA based on differentially expressed miR comparing T-PLL to healthy donor-derived T cells (n=37 miR, Online Supplementary Table S1). TCR-activated healthy donor-derived T cells clustered with T-PLL cases. (C) Heatmap and clustering based on differentially expressed miR comparing healthy donor-derived T cells: unstimulated controls versus TCR-activated condition (n=56 miR, FDR<0.05). Colors represent z-scores of CPM values calculated for each miR (blue=lower z-score; red=higher z-score).

192

haematologica | 2022; 107(1)

T-cell-receptor-activity shaped miR-omes of T-PLL

Figure 3. Functional affiliations of predicted microRNA-targets reflect associations with processes of altered DNA damage response and prosurvival signaling. (A) Gene set enrichment analysis (GSEA) heatmap based on microRNA (miR) associated mRNA. GSEA were conducted for all significantly deregulated miR (n=34 miR; Online Supplementary Table S1) using ranked correlation indices between mRNA and miR expression in 41 T-PLL and six healthy donor-derived T-cell samples. Exemplary HALLMARK pathways are displayed. Full figure is displayed in Online Supplementary Figure S8. Color code summarizes normalized enrichment scores (NES, blue=negative NES; red=positive NES). Statistical significance is summarized via asterisks (*P<0.05; **P<0.01 ; ***P<0.001 ; Kolmogorov-Smirnov-test). (B) Differential expression of miR-223-3p as analyzed by small-RNA sequencing shows significant upregulation in T-PLL (n=46) over normal T-cell controls (n=6); (fc=9.85; P=0.0002). (C) Exemplary GSEA plots of miR-223-3p correlated mRNA: P53 PATHWAY: NES=2.59, q<0.0001, G2M CHECKPOINT: NES=-2.34, q=0.003). (D) Predicted targets (by seed sequences, see Methods section for details) that correlated negatively with miR-223-3p expression in all analyzed cases and controls represent regulatory networks involved in DNA damage response and prosurvival signaling. Font color represents differential expression of mRNA comparing T-PLL cells (n=48 cases) and healthy donor-derived CD3+ pan-T cells (n=6 donors; for description of global mRNA sequencing results refer to Online Supplementary Figure S5 and Online Supplementary Table S5, blue= lower expression; red= higher expression). Color of highlighted boxes represents assignment of genes to functional groups of DNA damage response pathways (black) and prosurvival signaling (grey). (E) Groups of low and high miR-233-3p expression were assigned by results of small-RNA sequencing via comparison of the lower versus upper tertile of cases. Primary T-PLL cases were evaluated for CD38 and CD69 surface expression using flow cytometry (see Online Supplementary Table S8 for the comprehensive dataset). T-PLL with high miR-223-3p expression levels presented with a more activated T-cell phenotype (median CD38 expression: 65.5% vs. 3.6%, P=0.006; median CD69 expression: 8.5% vs. 1.0%, P=0.1; Mann–Whitney–Wilcoxon [MWW] test).

haematologica | 2022; 107(1)

193

T. Braun et al.

files of T-PLL cells with those of TCR-activated T cells. By integrating T-PLL’s miR profiles with transciptome data via GSEA (based on differentially expressed mRNA and on mRNA ordered by their correlation to the respective miR), we uncovered prominent regulatory networks around DNA damage response and prosurvival pathways in T-PLL.

Increased miR-223-3p expression is linked to activated T-cell phenotypes and associates with signatures of altered DNA damage response and cell cycle deregulation We next focused on phenotypic and clinical associations of miR (i) differentially expressed in our cohort and (ii) already linked to B- and / or T-cell leukemogenesis. In our sequencing analysis of small-RNA, miR-223-3p was significantly upregulated in T-PLL over CD3+ pan-T cells from age-matched healthy donors (fc=9.85, P=0.0002, Figure 3B). GSEA based on mRNA ranked by their correlation to miR223-3p expression revealed an association of miR-223-3p with signatures of altered DNA damage responses (e.g., P53 PATHWAY: NES=2.59, q<0.0001) and deregulated cell-cycle mediators (e.g., G2M CHECKPOINT: NES=-2.34, q=0.003, Figure 3C). We further evaluated potential miR-223-3p target mRNA, by combining target prediction databases and miR-ome – transcriptome correlations. Defining criteria of mRNA as putative miR targets in T-PLL are outlined in the Methods section. In total, we identified eight putative targets of miR-223-3p in T-PLL (Figure 3D). Notably, FOXO1 (rho=-0.41, P=0.005) was identified as one of them and was significantly downregulated in T-PLL (n=32 of 48 cases with fc<0.5 compared to normal T-cell controls). Tumor suppressive FOXO1 is a prominent regulator of redox balances and DNA insult-mediated cell death.32 When T-PLL cases were divided into three subgroups based on miR-223-3p expression (high, medium, low), we found that miR-223-3p expression correlated significantly with surface expression levels of the TCR activation markers CD38 (mean surface expression: 65.5% vs. 3.6%, P=0.006, MWW) and CD69 (mean 8.5% vs. 1.0%, P=0.1, MWW, Figure 3E, Online Supplementary Table S8 with summary of clinical data).

Increased levels of miR-200c and miR-141 species are associated with deregulated cell cycle molecules, activated phenotypes, and more aggressive presentations Another miR family, miR-200c/-141, was significantly upregulated in a subset of 23 T-PLL cases (Figures 4A and 1C). Upregulation for miR-141-3p was 43.2-fold (P<0.0001), for miR-141-5p 29.0-fold (P=0.0001), for miR200c-3p 38.2-fold (P<0.0001), and for miR-200c-5p 56.6fold (P=0.003) over all cases. MiR-141-3p showed the highest absolute CPM values among all deregulated miR in the entire cohort of T-PLL (mean CPMmiR-141-3p =26,561; mean CPM of all significantly deregulated miR in T-PLL =1,440, fc=18.4; Online Supplementary Figure S2). GSEA based on mRNA ranked by their correlation to miR-200c/-141 family members revealed significant enrichments of the HALLMARK E2F TARGET (NES=3.64, q<0.0001) and HALLMARK G2M CHECKPOINT gene sets (NES=3.05, q<0.0001, both based on miR-141-3p correlated mRNA, Figure 4B). Additionally, we identified 93 mRNA as potential targets of miR-200c/-141 (Figure 4C) in T-PLL. While nine mRNA showed an overlap between miR-141-3p and 194

miR-200c-3p target mRNA, we found only a small set of putative targets which were shared between either miR141-3p or miR-200c-3p and miR-141-5p. Exemplarily, KAT2B, a known tumor suppressor affecting DNA damage and cell cycle regulation,33 emerged as a potential target of miR-200c-3p (rho=-0.41, P=0.005, Spearman). Surface expression of the T-cell activation marker CD40L was elevated in cases with high miR-141-3p (mean 16.5% vs. 0.05%, P=0.03, MWW), high miR-200c-3p (mean 16.5% vs. 0.05%, P=0.03, MWW), and high miR-200c-5p expression (mean 20.0% vs. 0.0%, P=0.009, MWW, Figure 4D). Serum LDH levels at the time of sample correlated with increased miR-141-3p (mean 898 U/L vs. 509 U/l, P=0.03, MWW, Figure 4E) and elevated miR-200c-3p expression (mean 917 U/L vs. 509 U/L, P=0.02, MWW, Online Supplementary Table S9 with summary of clinical data).

Reduced miR-21 expression is linked to features of advanced or aggressive disease The small-RNA sequencing analysis also revealed a 3.7fold reduction of miR-21-3p expression (fc=0.27, P<0.0001) and a 3.2-fold reduction of miR-21-5p expression (fc=0.31, P<0.0001) in T-PLL (Online Supplementary Figure S9A). Interestingly, absolute expression (CPM) values of miR-215p were the second most altered among all deregulated miR in T-PLL (mean CPM of all significantly deregulated miR in T-PLL of 1440 vs. mean CPMmiR-21-5p of 15,526, fc=10.8, Online Supplementary Figure S2), suggesting a highly T-PLL-specific loss of expression. GSEA of miR-21-associated mRNA implicated relevance of this miR in apoptosis and cell cycle regulation, as gene sets like HALLMARK P53 PATHWAY and G2M CHECKPOINT were significantly deregulated (NES of APOPTOSIS PATHWAY considering miR-21-5p-associated mRNA=3.89, q<0.0001; NES of G2M CHECKPOINT considering miR-21-3p-associated mRNA=2.39, q=0.001, Online Supplementary Figure S9B). Furthermore, we assessed potential target mRNA of miR-21-3p and miR-21-5p as described above (n=42, e.g., MAP3K1 as a putative target of both miR-21-3p and miR-21-5p, Online Supplementary Figure S9C). In contrast to the current concept of miR-21 being a potent suppressor of cell cycle arrest and apoptosis induction, we did not find negative correlations with mRNA mediating these published effects (e.g., PDCD4: rho=0.04, P=0.79; BTG2: rho=0.28, P=0.05; correlations based on miR-21-5p expression, Spearman).34 Dichotomized by mean miR-21-5p expression, T-PLL with low miR-21-5p levels revealed higher white blood cell (WBC) counts (mean 154 G/L vs. 90.0 G/l, P=0.02, MWW) and lower platelet counts (mean 110 G/L vs. 153 G/L, P=0.03, MWW, Online Supplementary Figure S9D) at the time of sampling, indicating a more active growth behavior of T-PLL with low miR21 expression. Fittingly, serum levels of LDH (mean 933 U/L vs. 522 U/L, P=0.02, MWW, Online Supplementary Figure S9E) were elevated in patients with low cellular miR-21-5p expression (Online Supplementary Table S10 with summary of clinical data).

Reduced expression of miR-29 clusters is associated with alterations of survival signaling and cell cycle regulators reflected in features of a more active disease As analyzed by small-RNA sequencing, the miR-29 family members miR-29a-3p (fc=0.29, P<0.0001), miR-29b-15p (fc=0.47, P=0.001), and miR-29c-3p (fc=0.29, P<0.0001) showed a homogenous downregulation in T-PLL over haematologica | 2022; 107(1)

T-cell-receptor-activity shaped miR-omes of T-PLL

Figure 4. Increased expression of miR-200c and miR-141 clusters is associated with deregulation of cell cycle regulators reflected in more activated phenotypes and more aggressive disease course. (A) Differential expression of miR-141 and miR-200 family members as analyzed by small-RNA sequencing showed significant upregulation of miR-141-3p (fold change [fc]=43.2; P<0.0001), miR-141-5p (fc=29.0; P=0.0001), miR-200c-3p (fc=38.2; P<0.0001) and miR-200c-5p (fc=56.6; P=0.003, n=46 T-cell prolymphocytic leukemia [T-PLL], n=6 controls). (B) Exemplary gene set enrichment analysis (GSEA) plots of miR-200/-miR-141 correlated mRNA: E2F TARGETS: NES=3.64, q<0.0001, G2M CHECKPOINT: NES=3.05, q<0.0001 (both based on miR-141-3p correlated mRNA). (C) Predicted targets (by seed sequences, see Methods section for details) correlating negatively with miR-141/-miR-200 expression in all analyzed cases and controls showed regulatory networks involved in DNA damage response and prosurvival signaling. Font color represents differential expression of mRNA comparing T-PLL cells (n=48 cases) and healthy donor-derived CD3+ pan-T cells (n=6 donors; for description of global mRNA sequencing results refer to Online Supplementary Figure S5 and Online Supplementary Table S5, blue= lower expression; red= higher expression). Color of highlighting boxes represents assignment of genes to functional groups of DNA damage response pathways (black) and prosurvival signaling (grey). (D) Groups of low and high miR-141/-200 expression were assigned by results of small-RNA sequencing: after division into three tertiles, cases of the lower were compared to those of the upper tertile. Cases were evaluated for CD40L surface expression using flow cytometry. TPLL with higher miR-141-5p, miR-200c-3p, and miR-200c-5p expression presented with a more activated T-cell phenotype (median CD40L expression: 16.5% vs. 0.05%, P=0.03; 16.5% vs. 0.05%, P=0.03, 20.0% vs. 0.0%, P=0.009; MWW). (E) Higher serum LDH levels (see Online Supplementary Table S9 for a summary of clinical data) were associated with high expression of miR-141-3p and miR-200c-3p (median 898 U/L vs. 509 U/L; P=0.03; median 917 G/L vs. 509 G/L; P=0.02; MWW). Groups were divided into three tertiles and the lower was compared against the upper tertile.

haematologica | 2022; 107(1)

195

T. Braun et al.

healthy donor T cells (Online Supplementary Figure S10A). GSEA of associated mRNA showed an enrichment of genes of the HALLMARK E2F TARGET (NES=-3.52, q<0.0001) and of the TNFA SIGNALING VIA NFKB gene sets (NES=2.08, q=0.01, both miR-29a-3p-correlated mRNA) in low miR-29 expressing T-PLL (Online Supplementary Figure S10B). Furthermore, we identified 79 putative-target mRNA of the miR-29a-3p / miR-29b-1-5p / miR-29c-3p cluster in T-PLL by the above described strategy (Online Supplementary Figure S10C). Indicating an association of reduced miR-29 species with aberrant survival signaling, surface TCR activation markers were elevated in T-PLL with low miR-29a-3p (Online Supplementary Figure S10D), namely CD38 (mean expression 43.1% vs. 12.9%, P=0.02, MWW) and CD69 (mean 27.9% vs. 0.89%, P=0.02, MWW). A more active disease state as indicated by lower platelet counts (mean 110 G/L vs. 186 G/L, P=0.15, MWW) and higher LDH serum levels (mean 1,850 U/L vs. 708 U/l, P=0.06, MWW) at sampling was linked to low miR-29b-15p expression (Online Supplementary Figure S10E and F). Lower miR-29c expression tended to be associated with a higher incidence of effusions (six of 12 over one of ten with high miR-29c-3p-expression; P=0.07, Fisher’s exact test, Online Supplementary Figure S11A). In addition, genomic ATM deletions were more frequent in cases with low miR29b-1-5p expression (nine of 11 vs. one of 11 with high miR-29b-1-5p; P=0.002, Fisher’s exact test, Online Supplementary Figure S11B, Online Supplementary Table S11 with summary of clinical data).

A combinatorial miR-based overall survival score for T-cell prolymphocytic leukemia Based on the observed correlation of miR expression with clinical parameters, we aimed to establish a prognostic score that stratifies T-PLL patients according to miR expression levels. In order to identify best candidates in an unbiased fashion, we first associated miR expression levels with OS for all miR detected in at least 80% of T-PLL samples (n>36 cases) and compared T-PLL patients with highest expression levels (upper tertile) to those with lowest expression of the respective miR (lower tertile; Online Supplementary Table S12 with summary of clinical data). In this analysis, miR-98-3p (median OS in high vs. low expression: 16.3 months vs. 29.4 months, P=0.0008, log-rank, Figure 5A), miR-200a-3p (52.7 months vs. 19.1 months, P=0.001, log-rank, Figure 5B), miR223-3p (14.9 months vs. 26.0 months, P=0.001, log-rank, Figure 5C), and miR-424-5p (14.4 months vs. 26.0 months, P=0.0007, log-rank, Figure 5D) were most significantly correlated with OS. We subjected a training model composed of these four miR (miR-98-3p, miR-200a-3p, miR-223-3p, miR-424-5p) and of selected factors that had shown to be of prognostic relevance in T-PLL (e.g., WBC counts, TCL1A mRNA level)1,11,30 to parameter shaving by recursive partitioning. This algorithm identified optimum individual thresholds stratifying OS in the randomly created 22-case training set. Using these cutoffs, three miR (200a-3p, miR223-3p, miR-424-5p) and TCL1A expression remained as most significant discriminators for OS. Multiple combinatorial scores of these four parameters were built and for these scores optimum thresholds for discrimination of OS were calculated (recursive partitioning, see Methods). Best separation was obtained using a miR-exclusive 3-tier score: miR200a-3p fc<2.21, miR-223-3p fc≥9.8, miR-424-5p fc≥0.91; relative to healthy donor T cells) with a cutoff of ≥2 sum points (Table 2). Finally, we verified the miROS-T-PLL score 196

in the 22-case validation set (P=0.0004, log-rank) and in the total cohort of 44 T-PLL (median OS high vs. low miROS-TPLL: 14.4 months vs. 29.4 months, P<0.0001, log-rank, Figure 5E). In order to identify variables underlying (as potential confounders) the miR-based prognostic separation, we associated the expression of miR, which we used for the score, as well as the miROS-T-PLL score itself, with genomic, mRNA expression, immunophenotypic, and clinical data (Online Supplementary Tables S13 and S14). We did not detect significant differences in the distribution of these parameters between the groups determined by expression of the three miR or by the miROS-T-PLL score, further validating the newly established score.

Discussion Here, we report a pilot analysis of cellular miR expression in a cohort of 46 T-PLL patients. We identified 34 miR to be significantly deregulated in comparison to PB-derived T cells from age-matched healthy donors. These miR included those which had already been reported as altered in Tcell acute lymphoblastic leukemia (T-ALL, e.g., miR-2233p)35 and in CTCL (e.g., miR-29 and miR-200).17,23 They also contained miR that had not been described in the neoplastic context (e.g., miR-10395-5p). The global profiles of deregulated miR in T-PLL showed a rather uniform pattern across the analyzed cases. Together with the integrated information from transcriptome sequencing, this set of data allows for the first time insights into miR-based regulatory networks in T-PLL. It is important to mention, that four of the 34 differentially expressed miR presented with low CPM values (<1), either being unspecific background in the sequencing technology or representing biological relevant miR expressed at low levels. In addition, for two of the small-RNA identified to be differentially expressed in T-PLL (miR-6724-5p, miR5699-3p), miR-base36 assigned questionable confidence in their annotation, although their expression was previously reported in other entities (e.g., bladder cancer).37 In line with T-PLL’s phenotype of augmented TCR activation,1,5,11 our comparative profiling revealed a resemblance of T-PLL’s miR-ome to the one of TCR-activated healthy donor-derived T cells. A specific remodeling of the miR repertoire upon TCR activation had been shown,38 however, analysis of full spectrum miR expression by small RNA sequencing upon TCR activation in healthy donor-derived pan-T cells had not been reported before. We identified here previously unknown miR (e.g., upregulation of miR18a-5p) to be altered upon TCR activation in addition to those that had been described (miR-17-5p or miR-1505p).31,39 We conclude that constitutive TCR activation shapes the characteristic miR-ome of T-PLL cells. As hallmarks of T-PLL’s miR-ome, we identified miR-2233p, miR-21, the miR-29 family, and the miR-200c/-141 cluster as significantly deregulated. These miR have previously emerged as either onco miR or tumor-suppressive miR in other T- or B-cell malignancies.23,40–43 We further identified putative target signatures, potentially mediating the postulated effects of prosurvival signaling and aberrant DNA damage responses (e.g., downregulation of FOXO1 upon miR-223 upregulation). Limiting, the postulated target genes are, although predicted through multiple robust algorithms, based on associations without proven biological haematologica | 2022; 107(1)

T-cell-receptor-activity shaped miR-omes of T-PLL

effects, which has to be addressed in future experiments. Deregulation of these hallmark miR in T-PLL was further associated with either a pronounced cellular activation phenotype or more aggressive clinical presentations. MiR-21 stood out as one of the most abundant miR in TPLL. In contrast to the current concept of miR-21 being an onco miR,44 we detected significantly downregulated miR21 levels in T-PLL samples as compared to healthy donor T cells. Our integrative correlations further revealed an asso-

ciation of low miR-21 expression with more aggressive disease presentations. We identified SKP2 and MAP3K1 as predicted targets of miR-21 in T-PLL, potentially mediating these features. Notably, we did not find significant negative correlations of miR-21 with those mRNA that were previously described to mediate the effect of this miR as a potent suppressor of cell cycle inhibition and apoptosis.34 This indicates a T-cell specific and context-dependent function of miR-21.

Figure 5. miR-based overall survival score for T-cell prolymphocytic leukemia (miROS-T-PLL) stratifies patients based on miR-200a-3p, miR-223-3p, and miR-4245p expression. (A to D) Associations of microRNA (miR) expression with overall survival (OS) from diagnosis (see Online Supplementary Table S12 for the comprehensive dataset). (A) miR-98-3p (median OS high vs. low expression: 16.3 months vs. 29.4 months, P=0.0008, log-rank test); (B) miR-200a-3p (median OS high vs. low expression: 52.7 months vs. 19.1 months, P=0.001, log-rank test); (C) miR-223-3p (median OS high vs. low expression: 14.9 months vs. 26.0 months, P=0.001, log-rank test), and (D) miR-424-5p (median OS high vs. low expression: 14.4 months vs. 26 months, P=0.0007, log-rank test) were most significantly associated with OS, when comparing the tertile of T-PLL patients with the highest expression to the tertile with the lowest expression of the respective miR. (E) Analysis via training (n=22) and validation data sets (n=22): optimum thresholds were calculated by recursive partitioning for all possible combinations of miR-200a-3p, miR-223-3p, and miR-424-5p expression levels, adding one point to the total score if the respective expression cutoff was passed. Best results were obtained using the following thresholds: miR-200a-3b: fold change (fc)<2.21, miR-223-3p: fc ≥9.8, miR-424-5p: fc≥0.91 (fc relative to CD3+ pan-T cells derived from healthy donors) and a cutoff of ≥2 points. A significant OS association was observed in the validation set (P=0.0004, log-rank test) as well as (E) in the total cohort of 44 T-PLL patients (median OS high vs. low expression: 14.4 months vs. 29.4 months, P<0.0001, log-rank test).

haematologica | 2022; 107(1)

197

T. Braun et al.

Our computational analyses of mRNA targeted by deregulated miR in T-PLL suggest a strong impact of altered miR clusters on activation, death resistance, and aberrant DNA damage responses. Abnormal activity of these pathways, triggered by TCL1A overexpression and damaging ATM aberrations, has emerged as a hallmark of T-PLL Table 2. Prognostic score (miROS-T-PLL) including miR-200a-3p, miR223-3p, and miR-424-5p expression levels.

Parameter miR-200a-3p expression (FC, rel. to healthy CD3+ pan-T cells) miR-223-3p expression1 (FC, rel. to healthy CD3+ pan-T cells) miR-424-5p expression1 (FC, rel. to healthy CD3+ pan-T cells)

0 Points

1 Point

≥ 2.21

< 2.21

< 9.80

≥ 9.80

< 0.91

≥ 0.91

Prognostic Groups Lower Risk Higher Risk

0-1 points 2-3 points

evaluated by small-RNA sequencing and compared to the mean expression of CD3+ pan-T cells of six healthy donors: FC: fold change; miR: microRNA; T-PLL: T-cell prolymphocytic leukemia; miROS-T-PLL: miR-based overall survival score for T-PLL.

pathobiology.1,11 We propose that protumorigenic miR networks in their function as posttranscriptional regulators may further enhance the effects of these key genomic lesions, contributing substantially to the pathogenesis of TPLL. Similar cooperative miR-mRNA networks were postulated for T-ALL and CLL.35,45 The causes of the miR deregulations we observed here, remain unknown and are, besides TCR activation, likely multifactorial. As T-PLL is characterized by a strong genomic instability and high burdens of reactive oxygen species,1,46 mutations or copy number alterations of miR-encoding genes provide possible explanations, in addition to epigenetic mechanisms. Notably, incidences of genomic losses of the downregulated miR-1403p, miR-196b-5p, miR-339-3p, and miR-589-5p were in the order of 4-9% of T-PLL cases in our cohort. In summary, we identified a T-PLL-specific miR-ome, with 34 differentially deregulated miR, that appears instructed by TCR activation. By integrating altered miR expression with the information derived from transcriptome analyses, we postulate that the miR-ome of T-PLL shapes (dys)regulated networks towards apoptotic resistance, cell cycle abrogation, and defective DNA damage repair (Figure 6). Highlighting the pathobiological impact of the discovered miR deregulations, we developed the first

Figure 6. Graphical summary of postulated microRNA/mRNA-based deregulated networks in T-cell prolymphocytic leukemia. T-cell activity shaped microRNA expression signatures (miR-omes)/ transcriptome networks are displayed as identified by our combinatorial approach of small-RNA and transcriptome sequencing analyses in 41 clinically well characterized T-cell prolymphocytic leukemia (T-PLL) cases. MiR differentially expressed in T-PLL and mRNA associated with these miR (P<0.05) are presented. The background color of miR and mRNA (dots) indicates the fold changes of differential expression as compared to age-matched healthy donor-derived CD3+ pan-T cells (blue=downregulation, red=upregulation). MiR-223-3p, the miR-21 family, the miR-29 family, and the miR-200c/141 family emerged as hallmarks of the T-PLL miR-ome, as they were (i) significantly deregulated among T-PLL cases, (ii) presented putative targets involved in oncogenic pathways of TPLL’s pathobiology, (iii) showed associations with prognostic parameters, and (iv) were already described in the leukemogenesis of other B- and T-cell malignancies. Deregulations of these four miR-families were associated with cooperative effects on DNA damage response pathways as well as on pro-proliferative and cell survival signaling (as revealed by gene set enrichment analysis based on correlated mRNA). Genes previously described as hallmarks of T-PLL (e.g., TCL1A, CTLA4, and MYC) were found within the network of the identified miR-associated mRNA.

198

haematologica | 2022; 107(1)

T-cell-receptor-activity shaped miR-omes of T-PLL

clinical survival score that predicts T-PLL patients’ outcomes, based on expression levels of miR-200a-3p, miR223-3p, and miR-424-5p. Clinical management of T-PLL remains challenging and current studies mainly focus on the development of targeted treatments directed against anti-apoptotic factors like BLC247 or sustained prosurvival signaling mediated via JAK/STAT pathways.48,49 Our study suggests that developing a miR-directed targeting strategy could allow tackling a combination of those pathways and might therefore be a promising approach to successfully eradicate T-PLL cells. However, these concepts have to be tested in subsequent preclincal studies. Furthermore, the presented miROS-TPLL survival score might help in better discriminating rather indolent versus aggressive T-PLL disease phases at the time of diagnosis. Disclosures No conflicts of interest to disclose.

References 1. Schrader A, Crispatzu G, Oberbeck S, et al. Actionable perturbations of damage responses by TCL1/ATM and epigenetic lesions form the basis of T-PLL. Nat Commun. 2018;9(1):697. 2. Staber PB, Herling M, Bellido M, et al. Consensus criteria for diagnosis, staging, and treatment response assessment of Tcell prolymphocytic leukemia. Blood. 2019;134(14):1132-1143. 3. Braun T, von Jan J, Wahnschaffe L, Herling M. Advances and perspectives in the treatment of T-PLL. Curr Hematol Malig Rep. 2020;15(2):113-124. 4. Matutes E, Brito-Babapulle V, Swansbury J, et al. Clinical and laboratory features of 78 cases of T-prolymphocytic leukemia. Blood. 1991;78(12):3269-3274. 5. Oberbeck S, Schrader A, Warner K, et al. Non-canonical effector functions of the Tmemory-like T-PLL cell are shaped by cooperative TCL1A and TCR signaling. Blood. 2020 Dec 10;136(24):2786-2802. 6. Hopfinger G, Busch R, Pflug N, et al. Sequential chemoimmunotherapy of fludarabine, mitoxantrone, and cyclophosphamide induction followed by alemtuzumab consolidation is effective in T-cell prolymphocytic leukemia. Cancer. 2013;119(12):2258-2267. 7. Cross M, Dearden C. B and T cell prolymphocytic leukaemia. Best Pract Res Clin Haematol. 2019;32(3):217-228. 8. Dearden CE, Khot A, Else M, et al. Alemtuzumab therapy in T-cell prolymphocytic leukemia: comparing efficacy in a series treated intravenously and a study piloting the subcutaneous route. Blood. 2011;118(22):5799-802. 9. Dearden C. How I treat prolymphocytic leukemia? Blood. 2012;120(3):538-551. 10. Pflug N, Cramer P, Robrecht S, et al. New lessons learned in T-PLL: results from a prospective phase-II trial with fludarabine– mitoxantrone–cyclophosphamide–alemtuzumab induction followed by alemtuzumab maintenance. Leuk Lymphoma. 2019;60(3):649-657. 11. Herling M, Patel KA, Teitell MA, et al. High TCL1 expression and intact T-cell receptor signaling define a hyperproliferative subset of T-cell prolymphocytic leukemia. Blood.

haematologica | 2022; 107(1)

Contributions TB, AS, and MH developed the concept of the research; TB, PM, MF, MH and JA acquired the data; TB, MG, LW and MO performed the formal analysis; MG and LW acquired and analyzed the data; TB, LW and AS prepared the original draft and wrote the manuscript; SH and MH wrote, reviewed and edited the mansucript; AS, SH, MH and MHa supervised the work; TB and MO did the project administration. All authors reviewed and approved the final version of the manuscript. Funding This research was funded by the DFG Research Unit FOR1961 (Control-T; HE3553/4-2), the Köln Fortune program, and the Fritz Thyssen Foundation (10.15.2.034MN). This work was also funded by the EU Transcan-2 consortium ‘ERANET-PLL’ and by the ERAPerMed consortium ‘JAKSTAT-TARGET’. A.S. was supported by a scholarship of the German José Carreras Leukemia Foundation (DJCLS 03 F/2016). We gratefully acknowledge patients with their families for their invaluable contributions.

2008;111(1):328-337. 12. Wahnschaffe L, Braun T, Timonen S, et al. JAK/STAT-activating genomic alterations are a hallmark of T-PLL. Cancers (Basel). 2019;11(12):1833. 13. Bellanger D, Jacquemin V, Chopin M, et al. Recurrent JAK1 and JAK3 somatic mutations in T-cell prolymphocytic leukemia. Leukemia. 2014;28(2):417-419. 14. Kiel MJ, Velusamy T, Rolland D, et al. Integrated genomic sequencing reveals mutational landscape of T-cell prolymphocytic leukemia. Blood. 2014;124(9):14601472. 15. Stengel A, Kern W, Zenger M, et al. Genetic characterization of T-PLL reveals two major biologic subgroups and JAK3 mutations as prognostic marker. Genes Chromosomes Cancer. 2016;55(1):82-94. 16. Jonas S, Izaurralde E. Towards a molecular understanding of microRNA-mediated gene silencing. Nat Rev Genet. 2015;16(7): 421-433. 17. Kohnken R, Wen J, Mundy-Bosse B, et al. Diminished microRNA-29b level is associated with BRD4-mediated activation of oncogenes in cutaneous T-cell lymphoma. Blood. 2018;131(7):771-781. 18. Zenz T, Mohr J, Eldering E, et al. miR-34a as part of the resistance network in chronic lymphocytic leukemia. Blood. 2009;113 (16):3801-3808. 19. Wallaert A, Van Loocke W, Hernandez L, Taghon T, Speleman F, Van Vlierberghe P. Comprehensive miRNA expression profiling in human T-cell acute lymphoblastic leukemia by small RNA-sequencing. Sci Rep. 2017;7(1):7901. 20. Calin GA, Ferracin M, Cimmino A, et al. A microRNA signature associated with prognosis and progression in chronic lymphocytic leukemia. N Engl J Med. 2005;353(17): 1793-1801. 21. Bresin A, Callegari E, D’Abundo L, et al. miR-181b as a therapeutic agent for chronic lymphocytic leukemia in the Eµ-TCL1 mouse model. Oncotarget. 2015;6(23): 19807-19818. 22. Pekarsky Y, Santanam U, Cimmino A, et al. Tcl1 Expression in chronic lymphocytic leukemia is regulated by miR-29 and miR-181. Cancer Res. 2006;66(24):11590-11593. 23. Shen X, Wang B, Li K, et al. microRNA signatures in diagnosis and prognosis of cuta-

neous T cell lymphoma. J Invest Dermatol. 2018;138(9):202. 24. Wu SJ, Chen J, Wu B, Wang YJ, Guo KY. MicroRNA-150 enhances radiosensitivity by inhibiting the AKT pathway in NK/T cell lymphoma. J Exp Clin Cancer Res. 2018;37(1):18. 25. Ye Z, Jin H, Qian Q. Argonaute 2: a novel rising star in cancer research. J Cancer. 2015;6(9):877-882. 26. Herling M, Khoury JD, Washington LT, Duvic M, Keating MJ, Jones D. A systematic approach to diagnosis of mature T-cell leukemias reveals heterogeneity among WHO categories. Blood. 2004;104(2):328335. 27. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci. 2005;102(43):15545-15550. 28. Liberzon A, Birger C, Thorvaldsdóttir H, Ghandi M, Mesirov JP, Tamayo P. The Molecular Signatures Database hallmark gene set collection. Cell Syst. 2015;1(6): 417-425. 29. Ru Y, Kechris KJ, Tabakoff B, et al. The multiMiR R package and database: integration of microRNA–target interactions along with their disease and drug associations. Nucleic Acids Res. 2014;42(17):e133. 30. Jain P, Aoki E, Keating M, et al. Characteristics, outcomes, prognostic factors and treatment of patients with T-cell prolymphocytic leukemia (T-PLL). Ann Oncol. 2017;28(7):1554-1559. 31. Rodríguez-Galán A, Fernández-Messina L, Sánchez-Madrid F. Control of immunoregulatory molecules by miRNAs in T cell activation. Front Immunol. 2018;9:2148. 32. Dansen TB, Burgering BMT. Unravelling the tumor-suppressive functions of FOXO proteins. Trends Cell Biol. 2008;18(9):421429. 33. Zhu C, Qin Y-R, Xie D, et al. Characterization of tumor suppressive function of P300/CBP-associated factor at frequently deleted region 3p24 in esophageal squamous cell carcinoma. Oncogene. 2009;28(31):2821-2828. 34. Buscaglia LEB, Li Y. Apoptosis and the target genes of microRNA-21. Chin J Cancer. 2011;30(6):371-380. 35. Mavrakis KJ, Van Der Meulen J, Wolfe AL,

199

T. Braun et al. et al. A cooperative microRNA-tumor suppressor gene network in acute T-cell lymphoblastic leukemia (T-ALL). Nat Genet. 2011;43(7):673-678. 36. Kozomara A, Griffiths-Jones S. miRBase: annotating high confidence microRNAs using deep sequencing data. Nucleic Acids Res. 2013;42(D1):D68-D73. 37. Usuba W, Urabe F, Yamamoto Y, et al. Circulating miRNA panels for specific and early detection in bladder cancer. Cancer Sci. 2019;110(1):408-419. 38. Bronevetsky Y, Villarino AV, Eisley CJ, et al. T cell activation induces proteasomal degradation of Argonaute and rapid remodeling of the microRNA repertoire. J Exp Med. 2013;210(2):417-432. 39. Sousa IG, do Almo MM, Simi KCR, et al. MicroRNA expression profiles in human CD3(+) T cells following stimulation with anti-human CD3 antibodies. BMC Res Notes. 2017;10(1):124.

200

40. Pomari E, Lovisa F, Carraro E, et al. Clinical impact of miR-223 expression in pediatric T-Cell lymphoblastic lymphoma. Oncotarget. 2017;8(64):107886-107898. 41. Kwon JJ, Factora TD, Dey S, Kota J. A systematic review of miR-29 in cancer. Mol Ther Oncolytics. 2018;12:173-194. 42. Humphries B, Yang C. The microRNA-200 family: small molecules with novel roles in cancer development, progression and therapy. Oncotarget. 2015;6(9):6472-6498. 43. Garzon R, Heaphy CEA, Havelange V, et al. MicroRNA 29b functions in acute myeloid leukemia. Blood. 2009;114(26):5331-5341. 44. Feng YH, Tsao CJ. Emerging role of microRNA-21 in cancer (Review). Biomed Rep. 2016;5(4):395-402. 45. Li J, Qin Y, Zhang H. Identification of key miRNA-gene pairs in chronic lymphocytic leukemia through integrated analysis of mRNA and miRNA microarray. Oncol Lett. 2018;15(1):361-367.

46. Prinz C, Vasyutina E, Lohmann G, et al. Organometallic nucleosides induce nonclassical leukemic cell death that is mitochondrial-ROS dependent and facilitated by TCL1-oncogene burden. Mol Cancer. 2015;14:114. 47. Boidol B, Kornauth C, van der Kouwe E, et al. First-in-human response of BCL-2 inhibitor venetoclax in T-cell prolymphocytic leukemia. Blood. 2017;130(23):24992503. 48. Gomez-Arteaga A, Margolskee E, Wei MT, van Besien K, Inghirami G, Horwitz S. Combined use of tofacitinib (pan-JAK inhibitor) and ruxolitinib (a JAK1/2 inhibitor) for refractory T-cell prolymphocytic leukemia (T-PLL) with a JAK3 mutation. Leuk Lymphoma. 2019;60(7):16261631. 49. Bailey NG, Elenitoba-Johnson KSJ. What is new in mature T-cell leukemias? Semin Diagn Pathol. 2020;37(1):72-78.

haematologica | 2022; 107(1)

ARTICLE

Non Hodgkin-Lymphoma

Phenogenomic heterogeneity of post-transplant plasmablastic lymphomas

Ferrata Storti Foundation

Rebecca J. Leeman-Neill,¹ Craig R. Soderquist,1,2 Francesca Montanari,3 Patricia Raciti,4 David Park,¹ Dejan Radeski,5 Mahesh M. Mansukhani,2 Vundavalli V. Murty,6 Susan Hsiao,² Bachir Alobeid¹ and Govind Bhagat¹ Department of Pathology and Cell Biology, Division of Hematopathology, Columbia University Irving Medical Center, NY Presbyterian Hospital, New York, NY, USA; 2 Department of Pathology and Cell Biology, Division of Personalized Genomic Medicine, Columbia University Irving Medical Center, NY Presbyterian Hospital, New York, NY, USA; 3Department of Pathology and Cell Biology, Division of Hematology/Oncology, Columbia University Irving Medical Center, NY Presbyterian Hospital, New York, NY, USA; 4Department of Pathology, Greenwich Hospital, Greenwich, CT, USA; 5Department of Haematology, Sir Charles Gairdner Hospital, Perth, Australia; 6Department of Medicine, Division of Cytogenetics, Columbia University Irving Medical Center, NY Presbyterian Hospital, New York, NY, USA 1

Haematologica 2022 Volume 107(1):201-210

ABSTRACT

lasmablastic lymphoma (PBL) is a rare and clinically aggressive neoplasm that typically occurs in immunocompromised individuals, including those infected with human immunodeficiency virus (HIV) and solid organ allograft recipients. Most prior studies have focused on delineating the clinico-pathological features and genetic attributes of HIVrelated PBL, in which MYC deregulation, Epstein-Barr virus (EBV) infection and, more recently, mutations in JAK/STAT, MAP kinase, and NOTCH pathway genes have been implicated in disease pathogenesis. The phenotypic spectrum of post-transplant (PT)-PBL is not well characterized and data on underlying genetic alterations are limited. This led us to perform comprehensive histopathological and immunophenotypic evaluation and targeted sequencing of 18 samples from 11 patients (8 males, 3 females; age range, 12-76 years) with PT-PBL; eight de novo and three preceded by other types of post-transplant lymphoproliferative disorders. Post-transplant PBL displayed morphological and immunophenotypic heterogeneity and some features overlapped those of plasmablastic myeloma. Six (55%) cases were EBV positive and five (45%) showed MYC rearrangement by fluorescence in situ hybridization. Recurrent mutations in epigenetic regulators (KMT2/MLL family, TET2) and DNA damage repair and response (TP53, mismatch repair genes, FANCA, ATRX), MAP kinase (KRAS, NRAS, HRAS, BRAF), JAK/STAT (STAT3, STAT6, SOCS1), NOTCH (NOTCH1, NOTCH3, SPEN), and immune surveillance (FAS, CD58) pathway genes were observed, with the mutational profiles of EBV+ and EBV– cases exhibiting both similarities and differences. Clinical outcomes also varied, with survival ranging from 0-15.9 years after diagnosis. Besides uncovering the biological heterogeneity of PT-PBL, our study highlights similarities and distinctions between PT-PBL and PBL occurring in other settings and reveals potentially targetable oncogenic pathways in subsets of the disease.

Introduction Plasmablastic lymphoma (PBL) is an uncommon and aggressive B-cell nonHodgkin lymphoma characterized by a proliferation of cells with immunoblastic or plasmablastic morphology, occasionally with a component of mature plasma cells, and an immunophenotype indicative of terminal B-cell differentiation.1 PBL usually occurs in the context of immune dysregulation, which may be due to human immunodeficiency virus (HIV) infection, iatrogenic (e.g., after organ transplantation), congenital, or age-related (immune senescence).2 Prior studies of mostly HIV-associ-

haematologica | 2022; 107(1)

Correspondence: REBECCA LEEMAN-NEILL rjl2165@cumc.columbia.edu GOVIND BHAGAT gb96@cumc.columbia.edu Received: July 17, 2020. Accepted: November 25, 2020. Pre-published: December 3, 2020. https://doi.org/10.3324/haematol.2020.267294

201

R.J. Leeman-Neill et al.

ated PBL have highlighted the prognostic significance of disease stage and Epstein-Barr virus (EBV) status2-4 and genomic analyses have revealed frequent MYC rearrangements, heterogeneous chromosome/DNA copy number abnormalities,5-7 variable transcriptional and microRNA profiles2,8-10 and, recently, recurrent mutations in JAK/STAT, MAPK and NOTCH pathway genes.7,11 PBL occurring after solid organ transplantation (PT-PBL) is an uncommon type of post-transplant lymphoproliferative disorder (PTLD), accounting for 6-7% of PTLD and constituting a minor fraction (5-14%) of all PBL.2-4,12,13 Only limited data regarding the pathological and molecular features of PT-PBL have been reported.2-4,12,14 In order to clarify the pathogenetic bases of PT-PBL, we performed morphological, immunophenotypic and molecular analyses, including targeted genomic sequencing, of a series of PT-PBL, comprising de novo PBL, both primary and recurrent tumors, and those preceded by other types of PTLD.

Methods

Results Clinical characteristics We analyzed 18 samples from 11 patients (8 males, 3 females; median age 61 years; range, 12-76 years) with PTPBL, accounting for 11/177 (6%) of ‘destructive’B-cell PTLD and 11/98 (11%) of monomorphic B-cell PTLD diagnosed at our institution during the study period. PT-PBL occurred in recipients of heart (4/11, 36%), kidney (3/11, 27%), lung (3/11, 27%) and combined liver/kidney (1/11, 9%) allografts at a median of 9.6 years after transplantation (range, 0.6-11.9 years). The intestines were the most common sites of disease (6/11, 55%). Three patients had recurrent PBL and in three patients the PBL was preceded by another type of PTLD. Staging marrow biopsies, performed in seven patients, including three of five with EBV– PBL, showed no evidence of PBL/PTLD. Therapy and outcome data are summarized in Table 1 and details are provided in the Online Supplementary Data. Serum protein electrophoresis revealed low-level monoclonal paraproteins in four of seven (57%) patients with available results; none had lytic bone lesions on imaging. Results of pertinent laboratory tests and imaging studies are listed in Online Supplementary Table S2.

Case selection We searched our departmental database for cases of PTLD diagnosed over the past 18 years (2002-2019) to select those fulfilling morphological and immunophenotypic features of PBL according to the current World Health Organization classification.1 Other types of PTLD preceding PT-PBL were also identified. Clinical and laboratory data were retrieved from electronic health records. This study was performed according to the principles of the Declaration of Helsinki and a protocol approved by the Institutional Review Board of Columbia University.

Morphologic and immunophenotypic features

G-band karyotyping was performed on metaphase preparations obtained after unstimulated overnight culture. Fluorescence in situ hybridization (FISH) was performed on metaphase spreads or FFPE sections using TP53/CEP 17 and MYC/IGH/CEP8 probes (Abbott Molecular, Des Plaines, IL, USA) using standard methods. Two hundred cells per hybridization were evaluated. For interphase FISH analysis, the cut-off was 1% for IGH/MYC and 4% for TP53/CEP 17 alterations.

All PT-PBL showed diffuse infiltrates of large immunoblastic or plasmablastic cells (Figure 1). A minor component of small, more mature plasma cells (plasmacytic differentiation), comprising 10-20% of the neoplastic infiltrate, was seen in five (45%) cases (Figure 1A, Table 2). Some PBL had numerous tingible body macrophages, imparting a ‘starry sky’ appearance (Figure 1B) or multinucleated/anaplastic cells (Figure 1C). Foci of necrosis were observed in six of 11 (55%) cases. Details of the immunophenotypes of all cases are listed in Table 2 and flow cytometry results in Online Supplementary Table S3. Representative cases are illustrated in Figure 2. All PBL expressed MUM1/IRF4, nine of 11 (82%) were CD138+ and subsets showed B-cell antigen, CD10, CD56, PD-1 or PD-L1 expression. IgG, IgA or IgM was expressed by five (45%), two (18%), and two (18%) of the 11 cases. All evaluable PBL were positive for EMA and negative for Cyclin D1, CD117, HHV8 and ALK. Variable CD30 positivity was noted in seven of the 11 (64%) cases. The Ki-67 proliferation index ranged from 20 to >90% (median 90%). MYC expression ranged from <10% to 90% (median 45%), with six of ten (60%) cases showing ≥40% MYC expression. Two of the latter cases expressed BCL2 in ≥50% of cells (‘double expressors’). P53 overexpression was observed in five of ten (50%) cases. The immunoprofiles and/or proportions of cells expressing certain antigens differed in some PBL on recurrence. A mild to moderate infiltrate of reactive PD-1+ lymphocytes was observed in all ten cases evaluated and five of ten (50%) had PD-L1+ macrophages admixed. Three PBL were preceded by other forms of PTLD; nodal EBV+ monomorphic PTLD (plasmacytoma) (case 5), duodenal EBV+ P-PTLD (case 8), and intestinal EBV– monomorphic PTLD (diffuse large B-cell lymphoma, DLBCL) (case 10).

Targeted genomic sequencing

Epstein-Barr virus status and latency profiles

DNA was extracted from tumors and matched non-tumor tissue for sequencing a panel of 465 cancer-associated genes, as described previously16 (Online Supplementary Methods). Microsatellite instability (MSI) was also analyzed (Online Supplementary Methods).

The neoplastic cells were positive for EBER in six of 11 (55%) PBL, with four of these six (67%) showing latency II and one case each displaying latency 0/I and latency III profiles. Four EBV+ PBL and two EBV– PBL occurred in patients

Morphology and immunohistochemistry Formalin-fixed, paraffin-embedded (FFPE) tissue sections were stained with hematoxylin and eosin for cytomorphological evaluation and semi-quantitative assessment of the percentage of mature plasma cells. Immunohistochemistry/in situ hybridization was performed to analyze expression of B- and plasma-cell antigens, the cellular microenvironment, a variety of biomarkers and the EBV status/latency profiles (see Online Supplementary Methods).

Immunoglobulin heavy chain (IGH) gene rearrangement analysis Polymerase chain reaction analysis for immunoglobulin heavy chain (IGH) gene rearrangement was performed on DNA extracted from fresh or FFPE tissue using the BIOMED-2 primers, as described previously.15

Cytogenetic analysis

202

haematologica | 2022; 107(1)

Phenogenomic features of post-transplant PBL

Table 1. Clinical features of post-transplant plasmablastic lymphomas. Patient 1 2 3 4

Age, sex

61, M

70, M

76, M

65, M

22, M

12, F

72, M

57, F

64, M

15, M

60, F

Indication for tx

ETOH cirrhosis

NIDCM

CAD

PCKD

Obstructive uropathy

Kawasaki disease

IPF

LAM

Alport syndrome

HLHS

Pulmonary fibrosis

Type of tx

Liver and kidney

Heart

Kidney

Heart

Lung

Kidney

Heart

Lung

Azathioprine, Cyclosporine, Prednisone

MMF, Cyclosporine, Prednisone

Nasal

Skin, spinal epidural tissue

Skin, LN

Neg

Immunosuppression at diagnosis Site of involvement

Azathioprine. MMF, Tacrolimus, Cyclosporine, Prednisone Prednisone Pleural and Small peritoneal intestine, fluid pulmonary valve

Marrow involvement

Neg

Cyclosporine, Azathioprine, Prednisone Tacrolimus, Prednisone

Azathioprine, Azathioprine, Tacrolimus, Tacrolimus, Prednisone Prednisone

Small intestine, Small intestine, Small pleura peritoneal intestine fluid Neg

Azathioprine, Cyclosporine, MMF, Tacrolimus, Prednisone Tacrolimus, Prednisone Prednisone Small intestine, Small pelvic soft intestine tissue Neg

Neg

Liver

Stage

Time to development of PTLD (years)

0.6

7.0

11.1

1.3

7.8

10.3

11.9

0.9

9.6

10.8

10.7

None

R-EPOCH

Radiotherapy

Radiotherapy, BortezomibDexamethasone

R-CP, FCM

CP, GemOx, Surgery

R-EPOCH

None

BortezomibEPOCH, ASCT

R-EPOCH, Rituximab

0.03, dead

0.5, dead

4.1, dead

5.4, dead

0.3, dead

15.9, alive

2.0, dead

0.4, dead

0, dead

6.0, alive

1.4, dead

PTLD/sepsis

Unknown

Cardiac arrest

PTLD

ICH

Sepsis

Therapy

Dx to last follow up ( years) and status Cause of death

M: male; F: female; tx: transplant; dx: diagnosis; ETOH: alcohol-related; NIDCM non-ischemic dilated cardiomyopathy; CAD coronary heart disease; PCKD: polycystic kidney disease; IPF: idiopathic pulmonary fibrosis; LAM: lymphangioleiomyomatosis; HLHS: hypoplastic left heart syndrome; MMF mycophenolate mofetil; LN lymph node;; PTLD: post-transplant lymphoproliferative disorder; R-EPOCH: rituximab, etoposide, vincristine, doxorubicin, cyclophosphamide, prednisone; R-CP: rituximab cyclophosphamide, prednisone; FCM: fludarabine, cyclophosphamide, mytoxantrone; CP: cyclophosphamide, prednisone; GemOx: gemcitabine, oxaliplatin; ASCT autologous stem cell transplant; Dx: diagnosis; NA: not analyzed; ICH: intracranial hemorrhage.

seropositive for EBV at the time of transplantation. EBV viremia at diagnosis was observed in all four EBV+ and two of five patients with EBV– PBL with available results (Online Supplementary Table S2).

morphic PTLD or monomorphic PTLD (DLBCL) preceding PBL. Insufficient material precluded analysis of the PT-plasmacytoma.

Targeted genomic sequencing IGH gene rearrangement analysis All PT-PBL and both preceding monomorphic PTLD showed clonal IGH gene rearrangements. Clonal relatedness was established in all three recurrent PBL and between the PBL and prior monomorphic PTLD. The polymorphic PTLD demonstrated oligoclonal products.

Cytogenetic abnormalities Cytogenetic findings are listed in Table 3. All three PBL with informative results showed complex karyotypes. IGHMYC rearrangements were detected in five of 11 (45%) cases (3 at diagnosis, 2 at recurrence). Multiple copies of MYC due to polyploidy were detected in two cases (concurrent with MYC rearrangement in 1 case). Three of six cases with MYC abnormalities (2 with rearrangements, 1 with gain) showed ≥40% MYC positivity by immunohistochemistry. PBL of patients who died of PTLD, and those who did not, revealed MYC abnormalities in two of four (50%) versus four of seven (57%) cases, respectively (P=1.0). Two cases demonstrated IGH rearrangements with unknown partners. Chromosome 17/TP53 abnormalities were detected in six; subclonal (30-40% of cells) monosomy 17 (1 EBV+, 1 EBV–), multiple copies (polyploidy) of chromosome 17 (2 EBV–), and TP53 deletion (1 EBV+, 1 EBV–). Loss of TP53 was detected by targeted genomic sequencing in two cases, including one case with insufficient material for FISH analysis. No MYC or TP53 alterations were observed in the polyhaematologica | 2022; 107(1)

Pathogenic and likely pathogenic somatic non-synonymous single nucleotide variants (SNV) are listed in Table 3 and all, including variants of unknown significance (VUS), are listed in Online Supplementary Table S1. Excluding samples from two patients with high-level MSI, which exhibited up to 194 total SNV, the PBL harbored one to 16 pathogenic/likely pathogenic (median 7) and three to 26 total SNV (median 13) including VUS. MSI status was confirmed by a polymerase chain reaction-based method. All four PBL samples classified as having high-level MSI (from patients 6 and 7) demonstrated loss of expression of at least two mismatch repair (MMR) proteins (Table 2). After factoring out cases with high level MSI, the number of variants was still higher in EBV– than EBV+ PBL, but the difference was not statistically significant (4.7 mean pathogenic/likely pathogenic and 9.3 total SNV in EBV+ vs. 10.0 mean pathogenic/likely pathogenic and 19.3 total SNV in EBV– cases; P=0.15 and P=0.09, respectively). There was no a statistically significant difference in the number of variants between patients who died of PBL and those who did not (4.7 mean pathogenic/likely pathogenic and 10.0 total SNV vs. 7.3 mean pathogenic/likely pathogenic and 14.0 total SNV; P=0.51 and P=0.6, respectively). In our series, mutations were most frequent in epigenetic modifier genes, occurring in eight of 11 (73%) patients, and included recurrent mutations in the KMT2/MLL family of methyltransferases (KMT2C, n=5; KMT2D, n=3; and 203

R.J. Leeman-Neill et al.

Figure 1. Morphologic spectrum of post-transplant plasmablastic lymphomas. Representative hematoxylin and eosin (H&E)-stained sections of post-transplant plasmablastic lymphomas - from (A) case 4, showing a monotonous infiltrate of plasmablasts, with insets highlighting areas of plasmablastic morphology (top) and focal areas of plasmacytic differentiation (bottom), (B) case 5, showing numerous tingible body macrophages that impart a “starry sky” appearance, and (C) case 7, showing pleomorphic morphology, with insets displaying areas of plasmablastic morphology (top) and anaplastic-appearing multinucleated cells (bottom).

KMT2A, n=2), TET2 (n=3), ASXL1 (n=2), and KDM5C (n=2). DNA damage response and repair pathway genes were mutated in seven of 11 (64%) cases, including four of five (80%) EBV– and three of six (50%) EBV+ PBL. Three of seven (43%) cases, all EBV–, including both casese with high-level MSI, showed multiple TP53 mutations. One case with concomitant subclonal monosomy 17 showed 30% P53+ cells, as did two with concurrent chromosome 17 gains, suggesting bi/multiallelic inactivation (Tables 2 and 3). Chromosome 17/TP53 abnormalities were more frequent in PBL exhibiting plasmacytic differentiation (4/5, 80%) than in those lacking it (2/6, 33%), although the difference was not statistically significant (P=0.24). PBL with high-level MSI (cases 6 and 7) had mutations in the MMR genes: MLH1 and PMS2 or MSH6, respectively. Immunohistochemistry showed that these cases also had loss of MMR proteins (Table 2) as well as a high mutational burden. Other recurrently-mutated genes included BRCC3, ATRX, FANCA, and BRIP1. Mutations in the mitogen-activated protein kinase (MAPK) pathway genes were observed in six of 11 (55%) cases. Eight of nine mutations, occurring in KRAS, NRAS, HRAS, and BRAF, are well-known activating hotspot mutations (RAS codons 12, 13, and 61, and BRAF codons 469 and 601). Four cases harbored mutations in multiple genes, including one case with concomitant KRAS G13D, NRAS G12D, and BRAF G469V mutations. Mutations in NOTCH signaling pathway genes were detected in five of 11 (45%) cases. Putative gain- or loss-offunction mutations in genes encoding the NOTCH family of proteins were the most frequent. SPEN, a negative regulator of NOTCH signaling, was mutated in three cases. Many of the NOTCH pathway mutations were in PBL with MMR defects and one case exhibited mutations in multiple NOTCH pathway members at different time points. Janus kinase (JAK)/signal transducer and activator of transcription (STAT) signaling pathway genes, including STAT6 (n=3), STAT3 (n=2) and SOCS1 (n=1), were mutated in four of 11 (36%) cases. Immune surveillance pathway genes were mutated in four of 11 (36%) cases; FAS was mutated in all four cases, all EBV– PBL, and in the monomorphic PTLD (DLBCL) preceding one PBL. One PBL harbored concurrent FAS and CD58 mutations. The monomorphic PTLD (plasmacytoma), which harbored a NOTCH1 mutation, acquired a BRIP1 mutation upon transformation. In contrast, some of the mutations 204

observed in the monomorphic PTLD (DLBCL), including a STAT3 variant, were not detected in the clonally related transformed PBL, suggesting divergent evolution from a common ancestor. No variants were identified in the polymorphic PTLD preceding one PBL. The recurrent PBL showed both acquisition and loss of variants; the case with recurrent high-level MSI PBL (case 6) showed an increasing mutational burden over time.

Discussion Our understanding of PT-PBL pathogenesis is limited and its molecular underpinnings are largely inferred from those reported for HIV-related PBL. Transcriptional analyses have revealed upregulation of JAK-STAT pathway genes and similarities between PBL and plasma cell neoplasms (multiple myeloma [MM] and extra-osseous plasmacytoma) and differences between immunocompetent DLBCL and PBL, the latter displaying increased expression of MYC and MYB target genes and genes regulating plasma cell differentiation and decreased expression of B-cell receptor and NF-kB signaling pathway genes.7,9 However, microRNA expression analysis has suggested two different subclasses of PBL, resembling either Burkitt lymphoma or extra-osseous plasmacytoma.10 Chang et al. documented overlapping chromosomal aberrations between PBL and immunocompetent DLBCL as well as PBL-specific segmental gains at chromosomes 1p and 1q by array comparative genomic hybridization.5 Whole exome sequencing has uncovered recurrent gains at 1q21, 6p22 and 11p13, loci that contain several genes (histones, IL6R, MCL1, and CD44) mechanistically linked with B-cell lymphomagenesis.7 MYC deregulation, due to rearrangement, amplification, or activation of certain signaling pathways (e.g., STAT3) is considered important for the development of PBL.11,17-19 The frequency of MYC abnormalities (rearrangement and/or gain) in our series (55%) was higher than that reported previously for PT-PBL (38%), but within the range (36-69%) reported for PBL arising in other settings.2,3,7,11,12,17,19 In contrast to the study of mostly HIV-related PBL by Valera et al., MYC rearrangements were more common in EBV– PT-PBL,19 and in some cases, the rearrangement was detected upon PTLD transformation to PBL or at disease relapse. Recently, whole exome sequencing and targeted sequencing studies have unveiled the mutational landscape, primarily of HIV-related PBL,7,11 but until now the spectrum of genetic alterations underlying PT-PBL has not been explored. haematologica | 2022; 107(1)

PBL

Diagnosis

PBL

haematologica | 2022; 107(1)

ND ND

+ II ND

+ 0/I ND

+ II ND

20 No CD19-, CD20-, PAX5-, CD79a-, CD138+, BCL6-, CD10-, MUM1+, CD56+, EMA-/+ IgA <10% 20% <10% <5% 50% Rare/-

M-PTLD (Plasmacytoma)

5 PBL

PBL

5 5 5 <5 No No Yes Yes No CD19-, CD19+, CD19-, CD19+, CD19+, CD20-, CD20+, CD20-, CD20-/+, CD20-/+, PAX5-, PAX5+, PAX5+, PAX5+, PAX5+, CD79a-, CD79a+, CD79a-, CD79a-/+, CD79a+, CD138+, CD138+, CD138+/-, CD138-, CD138+, BCL6-, BCL6-, BCL6-, BCL6-, BCL6-, CD10+, CD10-, CD10-, CD10-, CD10-, MUM1+, MUM1+, MUM1+, MUM1+, MUM1+, CD56+, CD56CD56-, CD56CD56-, EMA+ EMA+ IgA IgG IgG IgG 20% 50% <10% 60% <10% 50% 30% 80% 30% 90% <10% <10% 50% 30% 90% 10% 70% 70% 60% 80% 60% 90% 50% >90% Mild/Mild/- Mild/10% Mild/Rare/with with macromacrophages+ phages+ + + + II II II NA NA ND ND ND MLH1-, PMS2-, MSH2+, MSH6+

PBL

4 P-PTLD

8 PBL

PBL

<5 10 10 20 Yes No Yes Yes Yes CD20-, CD19+, CD19+, CD19+, CD19-, PAX5+, CD20-/+ CD20+/-, CD20-, CD20-, CD79a+, PAX5+/-, AX5+/-, PAX5+/-, PAX5-, CD138-, CD79a+, PCD79a+, CD79a+, CD79a-/+, BCL6-, CD138-, CD138+, CD138+/-, CD138-, CD10-, BCL6-, BCL6-, BCL6-, CD10+, MUM1+, CD10-, CD10-, CD10-, BCL6-, CD56- MUM1+, MUM1+, MUM1+, MUM1+, CD56CD56CD56CD56-, EMA-/+ IgG IgM ND IgG IgG <10% 70% >90% 20% 30% ND 20% 20% 40% 30% 40% 30% <10% 10% 30% >90% 10% 20% 10% ND 70% 80% 90% 70% ND Mild/30% Mild/20% Mild/20% Mild/with with with macromacromacrophages+ phages+ phages+ + + NA NA III III NA ND MLH1-, ND ND ND PMS2-, MSH2+, MSH6-

PBL

NA ND

10% >90% 40% <10% <10% <10% 80% 20% Moderate/ ND with macrophages+

20 Yes CD19-, CD20-, PAX5-, CD79a+, CD138+, BCL6-, CD10+, MUM1+, CD56-, EMA+ IgA 50% 70% <10% 60% Mild/with macrophages+ NA ND

PBL

20 Yes No CD19+, CD19+, CD20+, CD20-, PAX5+, PAX5-, CD79a+, CD79a+, CD138-, D138+, BCL6+, BCL6-, CD10-, CD10-, MUM1-/+, MUM1+, CD56-, CD56-

M-PTLD (DLBCL)

80% 30% <10% 10% 90% Mild/with macrophages+ + II ND

<5 Yes CD19+, CD20-/+, PAX5-/+, CD79a+, CD138+, BCL6+/-, CD10+, MUM1+, CD56-

PBL

PBL: plasmablastic lymphoma; M-PTLD: monomorphic post-transplant lymphoproliferative disorder; P-PTLD: polymorphic post-transplant lymphoproliferative disorder; DLBCL: diffuse large B-cell lymphoma; PC: plasma cells; IG: immunoglobulin; IHC: immunohistochemistry; ND: not determined; NA: not applicable; EBER: Epstein-Barr virus encoded small RNA; MMR: mismatch repair. *sIg negative by flow cytometry.

EBER Latency type MMR IHC

Mature PC (%) <5 <5 <5 Necrosis No No No Immunophenotype CD19-, CD19+, CD19-, CD20-, CD20-, CD20-, PAX5-, PAX5+/-, PAX5-, CD79a-, CD79a+, CD79a-, CD138+, CD138-/+, CD138+, MUM1+ BCL6-, BCL6-, CD10+, CD10+, MUM1+, MUM1+, CD56+ CD56-, EMA+ IG heavy chain sIg-* IgM IgG BCL2 IHC ND >90% >90% CMYC IHC ND <10% 80% P53 IHC ND 10% 30% CD30 <5% Ki-67 ND 90% 90% PD1/PDL1 ND Mild/Mild/-

Table 2. Morphologic and immunophenotypic features of post-transplant plasmablastic lymphomas.

Patient

Phenogenomic features of post-transplant PBL

205

R.J. Leeman-Neill et al.

Figure 2. Immunophenotypic features of post-transplant plasmablastic lymphomas. (A) Hematoxylin and eosin (H&E)-stained section of tissue from case 2 showing plasmablasts, which are (B) partially positive for CD138 and (C) diffusely positive for MUM1 and display variable (D) PAX5 and (E) CD56 expression and evidence of (F) EBV infection by in situ hybridization for EBER. (G) H&E-stained section of case 9 showing plasmablasts, which are (H) partially positive for CD79a, (I) diffusely positive for MUM1 and display (J) aberrant CD10 expression, (K) P53 overexpression, and (L) moderate (30%) MYC expression.

While HIV-related PBL and PT-PBL appear to exhibit overlapping genomic changes, some differences are evident. In our series of PT-PBL, mutations in epigenetic modifiers were among the most frequent alterations (73%). Inactivating mutations of the KMT2/MLL family of histone H3 methyltransferases, which were most common, promote neoplasia via modification of global transcriptional activity.20 These mutations, particularly in KMT2D, have also been identified in monomorphic PTLD (DLBCL) (39%) and immunocompetent DLBCL (30-43%) and occur in 610% of cases of MM.21-26 Infrequent KMT2A mutations (6%), but no KMT2D mutations, have been reported in HIVrelated PBL.7 Recurrent loss-of-function mutations were also observed in the methylcytosine dioxygenase TET2 (27%), 206

which have been shown to alter gene transcription via widespread DNA hypermethylation, a process important in the pathogenesis of PTLD.27 TET2 mutations are frequent in immunocompetent DLBCL and MM22,23,25,26,28 and occur in 9% of HIV-related PBL.7 In contrast to their common occurrence in EBV+ DLBCL, in our cohort TET2 mutations were exclusively seen in EBV– PBL.29 Other epigenetic modifiers recurrently mutated in HIV-related PBL include EP300, which was mutated in one PT-PBL, as well as TRRAP and HDAC6, which were not in our sequencing panel.7,11 Alterations (mutations and copy number changes) in DNA damage response and repair pathway genes were detected in 82% of our cohort, with chromosome 17/TP53 abnormalities being the most common recurrent events haematologica | 2022; 107(1)

Phenogenomic features of post-transplant PBL

Table 3. Genetic alterations in post-transplant plasmablastic lymphomas.

PBL: plasmablastic lymphoma; M-PTLD: monomorphic post-transplant lymphoproliferative disorder; FFPE: formalin-fixed, paraffin-embedded; VUS: variant of unknown significance; SNV: single nucleotide variant; FISH: fluorescence in situ hybridization; MSI: microsatellite instability MSI-H: high-level microsatellite instability; NR: no result.

haematologica | 2022; 107(1)

207

R.J. Leeman-Neill et al.

(55%). The frequency of TP53 mutations in PT-PBL (27%) was comparable to that reported in monomorphic PTLD (DLBCL) (36-44%) and, as in the latter, mutations, were more common in EBV– cases.21,30 A lower frequency of TP53 mutations has been observed in HIV-related PBL (9%).7 Of interest, TP53 mutations are present in up to 23% of immunocompetent DLBCL; however, a high proportion of DLBCL with plasmablastic/plasmacytoid features (85%) harbor TP53 deletions.24-26,31 TP53 mutations are uncommon in MM at diagnosis,22,23,28 but can be detected at disease progression, concomitant with TP53 deletions, and are associated with poor prognosis.23,32 MSI, resulting from mutations in DNA MMR genes, was identified in two EBV– PBL that showed a high mutation burden. MMR defects and MSI are unusual in B-cell non-Hodgkin lymphomas of immunocompetent individuals, but not infrequent in MM33 and immunodeficiency-associated B cell neoplasms, including PTLD.34 Gain-of-function mutations in MAPK pathway genes also appear to be more frequent in PT-PBL (55%) compared to HIV-related PBL (28%).7 Moreover, concurrent mutations of multiple MAPK pathway members, noted in several PT-PBL, could reflect the presence of multiple subclones, as described in MM.35 Different members of the MAPK pathway are mutated in diverse hematologic and lymphoid malignancies. However, KRAS and NRAS mutations, which are common in MM,23 are infrequent in immunocompetent DLBCL. Knowles et al. reported an NRAS mutation in a post-transplant plasmacytoid immunoblastic lymphoma, which could have represented a PBL.36 Well-known, activating BRAF K601E and codon 469 mutations were observed in our study. The latter and the canonical V600E mutation have also been documented in PBL arising in other settings.7,11 Intriguingly, the BRAF V600E mutation was recently reported in an immunomodulatory therapy-associated EBV+ anaplastic large cell lymphoma.37 Further studies are required to determine whether this mutation is a recurrent, lineageindependent, phenomenon in immune dysregulation-related lymphomas. Recurrent mutations in members of the NOTCH signaling pathway, which controls B-cell fate determination,38 were observed in 45% of PT-PBL, mostly in EBV– cases (80% vs. 17% in EBV+ cases); it is unclear whether EBNA2 activates NOTCH signaling in a proportion of EBV+ PBL.39 Gain- and loss-of-function alterations in NOTCH pathway genes have been described in a variety of hemato-lymphoid neoplasms, including 24% of HIV-related PBL and some immunocompetent PBL.7,1 The pathogenesis of immunocompotent DLBCL of the N1 molecular subclass, which harbor NOTCH1 mutations and display a plasmacytic phenotype,26 is considered to be distinct from DLBCL of the BN2/Cluster 1 molecular subclass that have mutations in NOTCH2 and/or the NOTCH regulator, SPEN.24,26 However, we observed mutations in both, NOTCH1 and NOTCH2, as well as SPEN, at times concurrently, in PT-PBL. Deregulated activity of the NOTCH signaling pathway has also been implicated in the pathogenesis of MM, facilitating plasma cell growth and migration, but via different (non-mutational) mechanisms.40 JAK/STAT signaling, due to constitutive activation consequent to mutations, downstream effects of cytokine signaling, or EBV infection, contributes to the pathogenesis of several types of lymphoid neoplasms.41 Recurrent alterations in constituents of the JAK/STAT pathway, as observed in 36% of our cases and a higher proportion (62%) of HIV-related PBL, are known or predicted to enhance signaling. The 208

STAT3 D661Y mutation, also detected in 8% of HIV-related PBL7 and STAT6 E372K mutations, occur in the SH2 and DNA binding domains, respectively, resulting in nuclear localization and activation of the transcription factors.42,43 Recurrent STAT3 mutations were noted exclusively in EBV+ HIV-related PBL,11 but mutations in several JAK/STAT pathway members, including STAT3, were observed in both EBV+ and EBV– PT-PBL. Multiple concomitant mutations in SOCS1, a negative regulator of JAK family proteins, present in one PT-PBL, have not been functionally characterized, but are predicted to inactivate SOCS1.44 Abrogation of SOCS1 and SOCS3 function by epigenetic silencing or mutations has been described in other immunodeficiency-associated non-Hodgkin lymphomas, including monomorphic PTLD (DLBCL) and polymorphic PTLD.45 Mutations in immune surveillance-associated genes also occurred in PT-PBL. FAS, a member of the tumor necrosis factor receptor superfamily and an important mediator of Tcell cytotoxicity, was recurrently mutated in our series. FAS mutations have not been previously reported in PTLD or PBL, but have been documented in MM.46 They are also common in immunocompetent DLBCL, particularly in Cluster 1, which frequently also harbor NOTCH pathway mutations,24 and, as in our series, are almost exclusively seen in EBV– cases.29 In addition to FAS mutation, one PBL had a frameshift mutation in CD58, another immune surveillancerelated protein required for activation of natural killer cells, which is commonly mutated in immunocompetent DLBCL.47 Prior studies have reported variants in PRDM1, an inducer of terminal B-cell differentiation and regulator of MYC, in 20-50% of PBL.11,17 However, many of the variants were not expected to be deleterious.11,17 PRDM1 variants were detected in 4% of HIV-related PBL.7 None of the PT-PBL in our cohort had pathogenic PRDM1 mutations and only a single VUS (Q586H) was observed. Similar to our findings in PT-PBL, differences in the frequency or spectrum of genomic abnormalities between EBV+ and EBV– tumors have also been delineated in other types of B-cell PTLD.21,48-50 A lower mutation burden has been noted in EBV+ compared to EBV– monomorphic PTLD (DLBCL); this has been ascribed to the inherent oncogenic activity of the virus and, hence, a reduced requirement for proto-oncogene or tumor suppressor gene alterations. EBV+ cases constituted 55% of our PT-PBL, a frequency not significantly different from previous studies of PT-PBL (67-79%) and intermediate between that reported for HIVrelated PBL (75-90%) and immunocompetent PBL (3350%).2-4,7,12 Data regarding the EBV latency program in PBL have been conflicting. Ambrosio et al. reported a non-canonical EBV latency program i.e., partial expression of proteins characteristic of type II latency with simultaneous expression of lytic phase proteins in HIV+ and HIV– cases.10 Similarly, gene expression studies of HIV-related PBL have shown a much higher expression of the EBV lytic genes (BALF4 and BALF5) than canonical latency program genes, in most cases.7 Castillo et al., however, described latency I or III in most HIV-related EBV+ PBL, latency I in immunocompetent PBL, and predominantly latency III in PT-PBL.3 The vast majority of the EBV+ cases in our study displayed a latency II profile, similar to the observations of Morscio et al.,2 but different from those of Zimmerman et al., who reported mostly latency 0/I in PT-PBL.14 In our series, PBL exhibited morphologic and immunophenotypic heterogeneity, in line with prior obserhaematologica | 2022; 107(1)

Phenogenomic features of post-transplant PBL

vations.2,51 Almost half of the cases, including both EBV+ and EBV– cases, displayed minor foci of plasmacytic differentiation, concordant with the findings of other investigators.3,51 This feature was more common in PBL with chromosome 17/TP53 abnormalities. As reported for other B-cell PTLD,48 the majority (64%) of PT-PBL showed evidence of germinal center transit. Although the B-cell program is characteristically downregulated in PBL,2,3 a proportion of cases express B-cell antigens.3,52,53 Partial CD20 expression was observed in 27% of PT-PBL, a frequency similar to that reported for other types of PBL (23%).53 Over half of the PT-PBL showed variable PAX5 and/or CD79a positivity. Expression of CD79a has been documented in 45% of HIV-related and 68% of PT-PBL3 and PAX5 in 23-26% of mostly HIV-related PBL.52,53 In contrast to the findings of Montes-Moreno et al., who noted more frequent CD20 and/or PAX5 expression in EBV– PBL, a larger proportion of EBV+ PT-PBL (60%) was PAX5+.53 The two EBV– PAX5+ and CD79a+ PBL had highlevel MSI; otherwise, no differences in functional groups of mutations were apparent between cases expressing or lacking B-cell antigens. Since all patients with PTLD preceding PBL received rituximab, it is unclear whether the anti-CD20 antibody therapy was responsible for CD20 negativity of the PBL and/or promoted plasmablastic differentiation. Expression of CD56 and CD10, observed in a subset of PTPBL, has been reported more frequently in HIV-related and PT-PBL.2,3,53 Moreover, variability in the Ki-67 proliferation indices of our PT-PBL is in line with the findings of Morscio et al. (25-100% Ki-67 labeling in PT-PBL).2 PD-L1 expression, tumor infiltration by PD1+ T cells, and upregulation of genes related to immune escape, have been observed in EBV+ PBL7,8,11 and PTLD.54 PD-L1 expression by tumor cells was observed in subsets of EBV+ and EBV– PTPBL; the latter also harboring mutations in immune evasionrelated genes (FAS and CD58). PD-1 expression by tumor cells noted in one case has been previously reported in 5% of PBL.52 Thorough clinicopathological correlation is essential for resolving the differential diagnosis of PBL, which includes other neoplasms with plasmablastic features e.g., large B-cell lymphoma arising from HHV8-associated multicentric Castleman disease, primary effusion lymphoma, and ALK+ large B-cell lymphoma. Negative staining for HHV8 and ALK excluded these possibilities. Distinguishing between PBL and plasmablastic MM , however, can be difficult and a multimodal approach is required to make a correct diagnosis. Serum paraprotein analysis is not helpful as monoclonal proteins can be observed in some PBL patients,14 as was the case in our series. Importantly, none of the PT-PBL with bone marrow biopsies showed morphologic or immunophenotypic evidence of marrow involvement, all lacked bone (lytic) lesions on imaging as well as myelomarelated laboratory abnormalities, including hypercalcemia, and the vast majority occurred at mucosal sites, findings that

References 1. Swerdlow SH, Campo E, Harris NL, et al. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. Revised 4th edition, Volume 2. 2016. 2. Morscio J, Dierickx D, Nijs J, et al. Clinicopathologic comparison of plasmablastic lymphoma in HIV-positive, immunocompetent, and posttransplant

haematologica | 2022; 107(1)

do not support a diagnosis of MM.55 Furthermore, although some of the genetic abnormalities observed in PT-PBL overlapped with those of MM, they were detected in both EBV+ and EBV– PT-PBL and in HIV-related PBL.7,11 Furthermore, the overall complement of alterations differed from that of MM. PT-PBL usually occur late after transplantation (median 96 months; range, 2-360 months)2,14 and are more frequent in males and in recipients of heart and kidney allografts,2,3,12,14 which was also true in our series. However, in contrast to a predominance of skin and lymph node involvement described previously,2,3 we noted a high frequency of intestinal disease (55%), with primary skin involvement observed in only 18% of patients. In addition to the oral cavity, the gastrointestinal tract is also a frequent site for HIVrelated PBL.3 Age, stage, and nodal involvement influence the prognosis of PBL, which is typically poor,12 although long survival, as observed for some of our PT-PBL patients, has been reported previously.2 In contrast to Zimmermann et al., we did not observe that patients with EBV– PT-PBL or PBL harboring MYC and/or IGH rearrangements had a worse prognosis.14 We could not determine any correlation between patients’ outcome and particular functional groups of mutations or mutational burden, although most had stage IV disease. Most patients in our cohort received lymphoma-directed therapies using conventional chemotherapy and/or radiotherapy. However two patients, including one who is still alive, also received bortezomib, a proteasome inhibitor frequently used to treat MM, which, in combination with lymphoma regimens, has been shown to be effective in treating PBL.3 In summary, our study is the first to investigate the genetic landscape of PT-PBL, revealing recurrent mutations in epigenetic modifiers and DNA damage response and repair, MAPK, JAK/STAT, NOTCH, and immune surveillance pathway genes. The observed genomic alterations overlap those reported for HIV-related PBL as well as subtypes of immunocompetent DLBCL and MM. Our findings reiterate the phenotypic heterogeneity of this rare type of PTLD, provide novel insights into PT-PBL biology and identify pathways amenable to targeted therapies. Disclosures No conflicts of interest to disclose. Contributions RL and GB led the project, analyzed data and wrote the manuscript, PR performed research and analyzed data, CS performed research, analyzed data and critically reviewed the manuscript, MM supervised molecular analyses, VM performed cytogenetic analyses, SH performed molecular analyses and analyzed data, DR and FM performed research and obtained clinical data, BA and DP contributed to interpreting the data and critically reviewed the manuscript.

patients: single-center series of 25 cases and meta-analysis of 277 reported cases. Am J Surg Pathol. 2014;38(7):875-886. 3. Castillo JJ, Bibas M, Miranda RN. The biology and treatment of plasmablastic lymphoma. Blood. 2015;125(15):2323-2330. 4. Li YJ, Li JW, Chen KL, et al. HIV-negative plasmablastic lymphoma: report of 8 cases and a comprehensive review of 394 published cases. Blood Res. 2020;55(1):49-56.

5. Chang CC, Zhou X, Taylor JJ, et al. Genomic profiling of plasmablastic lymphoma using array comparative genomic hybridization (aCGH): revealing significant overlapping genomic lesions with diffuse large B-cell lymphoma. J Hematol Oncol. 2009;2:47. 6. Sarkozy C, Kaltenbach S, Faurie P, et al. Array-CGH predicts prognosis in plasma cell post-transplantation lymphoproliferative disorders. Genes Chromosomes Cancer.

209

R.J. Leeman-Neill et al. 2017;56(3):221-230. 7. Liu Z, Filip I, Gomez K, et al. Genomic characterization of HIV-associated plasmablastic lymphoma identifies pervasive mutations in the JAK–STAT pathway. Blood Cancer Discov. 2020;1(1):112-125. 8. Gravelle P, Pericart S, Tosolini M, et al. EBV infection determines the immune hallmarks of plasmablastic lymphoma. Oncoimmunology. 2018;7(10):e1486950. 9. Chapman J, Gentles AJ, Sujoy V, et al. Gene expression analysis of plasmablastic lymphoma identifies downregulation of B-cell receptor signaling and additional unique transcriptional programs. Leukemia. 2015;29 (11):2270-2273. 10. Ambrosio MR, Mundo L, Gazaneo S, et al. MicroRNAs sequencing unveils distinct molecular subgroups of plasmablastic lymphoma. Oncotarget. 2017;8(64):107356107373. 11. Garcia-Reyero J, Martinez Magunacelaya N, Gonzalez de Villambrosia S, et al. Genetic lesions in MYC and STAT3 drive oncogenic transcription factor overexpression in plasmablastic lymphoma. Haematologica. 2021;106(4):1120-1128. 12. Loghavi S, Alayed K, Aladily TN, et al. Stage, age, and EBV status impact outcomes of plasmablastic lymphoma patients: a clinicopathologic analysis of 61 patients. J Hematol Oncol. 2015;8:65. 13. Dierickx D, Tousseyn T, Sagaert X, et al. Single-center analysis of biopsy-confirmed posttransplant lymphoproliferative disorder: incidence, clinicopathological characteristics and prognostic factors. Leuk Lymphoma. 2013;54(11):2433-2440. 14. Zimmermann H, Oschlies I, Fink S, et al. Plasmablastic posttransplant lymphoma: cytogenetic aberrations and lack of EpsteinBarr virus association linked with poor outcome in the prospective German posttransplant lymphoproliferative disorder registry. Transplantation. 2012;93(5):543-550. 15. van Dongen JJ, Langerak AW, Bruggemann M, et al. Design and standardization of PCR primers and protocols for detection of clonal immunoglobulin and T-cell receptor gene recombinations in suspect lymphoproliferations: report of the BIOMED-2 Concerted Action BMH4-CT98-3936. Leukemia. 2003;17(12):2257-2317. 16. Pang J, Gindin T, Mansukhani M, Fernandes H, Hsiao S. Microsatellite instability detection using a large next-generation sequencing cancer panel across diverse tumour types. J Clin Pathol. 2020;73(2):83-89. 17. Montes-Moreno S, Martinez-Magunacelaya N, Zecchini-Barrese T, et al. Plasmablastic lymphoma phenotype is determined by genetic alterations in MYC and PRDM1. Mod Pathol. 2017;30(1):85-94. 18. Sarosiek KA, Malumbres R, Nechushtan H, Gentles AJ, Avisar E, Lossos IS. Novel IL-21 signaling pathway up-regulates c-Myc and induces apoptosis of diffuse large B-cell lymphomas. Blood. 2010;115(3):570-580. 19. Valera A, Balague O, Colomo L, et al. IG/MYC rearrangements are the main cytogenetic alteration in plasmablastic lymphomas. Am J Surg Pathol. 2010;34(11): 1686-1694. 20. Zhang J, Dominguez-Sola D, Hussein S, et al. Disruption of KMT2D perturbs germinal center B cell development and promotes lymphomagenesis. Nat Med. 2015;21(10):1190-1198. 21. Menter T, Juskevicius D, Alikian M, et al. Mutational landscape of B-cell post-trans-

210

plant lymphoproliferative disorders. Br J Haematol. 2017;178(1):48-56. 22. Chapman MA, Lawrence MS, Keats JJ, et al. Initial genome sequencing and analysis of multiple myeloma. Nature. 2011;471(7339): 467-472. 23. Maura F, Bolli N, Angelopoulos N, et al. Genomic landscape and chronological reconstruction of driver events in multiple myeloma. Nat Commun. 2019;10(1):3835. 24. Chapuy B, Stewart C, Dunford AJ, et al. Molecular subtypes of diffuse large B cell lymphoma are associated with distinct pathogenic mechanisms and outcomes. Nat Med. 2018;24(5):679-690. 25. Reddy A, Zhang J, Davis NS, et al. Genetic and functional drivers of diffuse large B cell lymphoma. Cell. 2017;171(2):481-494. 26. Schmitz R, Wright GW, Huang DW, et al. Genetics and pathogenesis of diffuse large Bcell lymphoma. N Engl J Med. 2018;378 (15):1396-1407. 27. Rossi D, Gaidano G, Gloghini A, et al. Frequent aberrant promoter hypermethylation of O6-methylguanine-DNA methyltransferase and death-associated protein kinase genes in immunodeficiency-related lymphomas. Br J Haematol. 2003;123(3): 475-478. 28. Walker BA, Mavrommatis K, Wardell CP, et al. Identification of novel mutational drivers reveals oncogene dependencies in multiple myeloma. Blood. 2018;132(6):587-597. 29. Kataoka K, Miyoshi H, Sakata S, et al. Frequent structural variations involving programmed death ligands in Epstein-Barr virusassociated lymphomas. Leukemia. 2019;33(7):1687-1699. 30. Courville EL, Yohe S, Chou D, et al. EBVnegative monomorphic B-cell post-transplant lymphoproliferative disorders are pathologically distinct from EBV-positive cases and frequently contain TP53 mutations. Mod Pathol. 2016;29(10):1200-1211. 31. Simonitsch-Klupp I, Hauser I, Ott G, et al. Diffuse large B-cell lymphomas with plasmablastic/plasmacytoid features are associated with TP53 deletions and poor clinical outcome. Leukemia. 2004;18(1):146-155. 32. Lode L, Eveillard M, Trichet V, et al. Mutations in TP53 are exclusively associated with del(17p) in multiple myeloma. Haematologica. 2010;95(11):1973-1976. 33. Velangi MR, Matheson EC, Morgan GJ, et al. DNA mismatch repair pathway defects in the pathogenesis and evolution of myeloma. Carcinogenesis. 2004;25(10):1795-1803. 34. Capello D, Rossi D, Gaidano G. Post-transplant lymphoproliferative disorders: molecular basis of disease histogenesis and pathogenesis. Hematol Oncol. 2005;23(2):61-67. 35. Lohr JG, Stojanov P, Carter SL, et al. Widespread genetic heterogeneity in multiple myeloma: implications for targeted therapy. Cancer Cell. 2014;25(1):91-101. 36. Knowles DM, Cesarman E, Chadburn A, et al. Correlative morphologic and molecular genetic analysis demonstrates three distinct categories of posttransplantation lymphoproliferative disorders. Blood. 1995;85(2): 552-565. 37. Gru AA, Williams E, Junkins-Hopkins JM. An immune suppression-associated EBVpositive anaplastic large cell lymphoma with a BRAF V600E mutation. Am J Surg Pathol. 2019;43(1):140-146. 38. Gu Y, Masiero M, Banham AH. Notch signaling: its roles and therapeutic potential in hematological malignancies. Oncotarget. 2016;7(20):29804-29823.

39. Kempkes B, Ling PD. EBNA2 and its coactivator EBNA-LP. Curr Top Microbiol Immunol. 2015;391:35-59. 40. Colombo M, Galletti S, Garavelli S, et al. Notch signaling deregulation in multiple myeloma: a rational molecular target. Oncotarget. 2015;6(29):26826-26840. 41. Vainchenker W, Constantinescu SN. JAK/STAT signaling in hematological malignancies. Oncogene. 2013;32(21):2601-2613. 42. Koskela HL, Eldfors S, Ellonen P, et al. Somatic STAT3 mutations in large granular lymphocytic leukemia. N Engl J Med. 2012;366(20):1905-1913. 43. Yildiz M, Li H, Bernard D, et al. Activating STAT6 mutations in follicular lymphoma. Blood. 2015;125(4):668-679. 44. Liau NPD, Laktyushin A, Lucet IS, et al. The molecular basis of JAK/STAT inhibition by SOCS1. Nat Commun. 2018;9(1):1558. 45. Capello D, Gloghini A, Baldanzi G, et al. Alterations of negative regulators of cytokine signalling in immunodeficiency-related nonHodgkin lymphoma. Hematol Oncol. 2013;31(1):22-28. 46. Landowski TH, Qu N, Buyuksal I, Painter JS, Dalton WS. Mutations in the Fas antigen in patients with multiple myeloma. Blood. 1997;90(11):4266-4270. 47. Challa-Malladi M, Lieu YK, Califano O, et al. Combined genetic inactivation of b2microglobulin and CD58 reveals frequent escape from immune recognition in diffuse large B cell lymphoma. Cancer Cell. 2011;20(6):728-740. 48. Vakiani E, Basso K, Klein U, et al. Genetic and phenotypic analysis of B-cell post-transplant lymphoproliferative disorders provides insights into disease biology. Hematol Oncol. 2008;26(4):199-211. 49. Rinaldi A, Capello D, Scandurra M, et al. Single nucleotide polymorphism-arrays provide new insights in the pathogenesis of post-transplant diffuse large B-cell lymphoma. Br J Haematol. 2010;149(4):569-577. 50. Rinaldi A, Kwee I, Poretti G, et al. Comparative genome-wide profiling of posttransplant lymphoproliferative disorders and diffuse large B-cell lymphomas. Br J Haematol. 2006;134(1):27-36. 51. Colomo L, Loong F, Rives S, et al. Diffuse large B-cell lymphomas with plasmablastic differentiation represent a heterogeneous group of disease entities. Am J Surg Pathol. 2004;28(6):736-747. 52. Laurent C, Fabiani B, Do C, et al. Immunecheckpoint expression in Epstein-Barr virus positive and negative plasmablastic lymphoma: a clinical and pathological study in 82 patients. Haematologica. 2016;101(8): 976-984. 53. Montes-Moreno S, Gonzalez-Medina AR, Rodriguez-Pinilla SM, et al. Aggressive large B-cell lymphoma with plasma cell differentiation: immunohistochemical characterization of plasmablastic lymphoma and diffuse large B-cell lymphoma with partial plasmablastic phenotype. Haematologica. 2010;95(8):1342-1349. 54. Green MR, Rodig S, Juszczynski P, et al. Constitutive AP-1 activity and EBV infection induce PD-L1 in Hodgkin lymphomas and posttransplant lymphoproliferative disorders: implications for targeted therapy. Clin Cancer Res. 2012;18(6):1611-1618. 55. Ofori K, Soderquist C, Murty VV, et al. The clinical and pathological features of plasma cell myeloma post solid organ transplantation. Am J Hematol. 2020;95(12):15311541.

haematologica | 2022; 107(1)

ARTICLE

Non Hodgkin-Lymphoma

Shallow-depth sequencing of cell-free DNA for Hodgkin and diffuse large B-cell lymphoma (differential) diagnosis: a standardized approach with underappreciated potential Lennart Raman,1,2,* Malaïka Van der Linden,1,2,* Ciel De Vriendt,3,* Bliede Van den Broeck,4 Kristoff Muylle,5 Dries Deeren,6 Franceska Dedeurwaerdere,7 Sofie Verbeke,1 Amélie Dendooven,1,8 Katrien De Grove,3 Saskia Baert,3 Kathleen Claes,2 Björn Menten,2 Fritz Offner3# and Jo Van Dorpe1# Department of Pathology, Ghent University, Ghent University Hospital, Ghent; 2Center for Medical Genetics, Department of Biomolecular Medicine, Ghent University, Ghent University Hospital, Ghent; 3Department of Clinical Hematology, Ghent University, Ghent University Hospital, Ghent; 4Department of Nuclear Medicine, Ghent University, Ghent University Hospital, Ghent; 5Department of Nuclear Medicine, AZ Delta, Roeselare; 6 Department of Hematology, AZ Delta, Roeselare; 7Deparment of Pathology, AZ Delta, Roeselare and 8Faculty of Medicine and Health Sciences, University of Antwerp, Wilrijk, Belgium 1

Ferrata Storti Foundation

Haematologica 2022 Volume 107(1):211-220

*LR, MVDL and CDV contributed equally as co-first authors. #

JVD and FO contributed equally as co-senior authors.

ABSTRACT

hallow-depth sequencing of cell-free DNA, an inexpensive and standardized approach to obtain molecular information on tumors non-invasively, has been insufficiently explored for the diagnosis of lymphoma and disease follow-up. This study collected 318 samples, including longitudinal liquid and paired solid biopsies, from a prospectively-recruited cohort of 38 Hodgkin lymphoma (HL) and 85 aggressive B-cell non-HL patients, represented by 81 diffuse large B-cell lymphoma (DLBCL) cases. Following sequencing, copy number alterations and viral read fractions were derived and analyzed. At diagnosis, liquid biopsies showed detectable copy number alterations in 84.2% of HL patients (88.6% for classical HL) and 74.1% of DLBCL patients. Of the DLBCL patients, copy number profiles between liquid-solid pairs were highly concordant (r=0.815±0.043); and, compared to tissue, HL liquid biopsies had abnormalities with higher amplitudes (P=0.010). This implies that tumor DNA is more abundant in plasma. Additionally, 39.5% of HL and 13.6% of DLBCL cases had a significantly elevated number of plasma Epstein-Barr virus DNA fragments, achieving a sensitivity of 100% compared to the current standard. A longitudinal analysis determined that, when detectable, copy number patterns were similar across (re)staging moments in refractory or relapsed patients. Further, the overall profile anomaly correlated highly with the total metabolic tumor volume (P<0.001). To conclude, as a proof of principle, we demonstrate that liquid biopsy-derived copy numbers can aid diagnosis: e.g., by differentiating HL from DLBCL, random forest modeling is represented by an area under the receiver operating characteristic curve of 0.967. This application is potentially useful when tissue is difficult to obtain or when biopsies are small and inconclusive.

Introduction B-cell-derived lymphomas can develop at different stages along the B-lymphocyte differentiation pathways. This generates a range of tumor entities1 including Hodgkin lymphoma (HL) as well as diverse non-HL (nHL) varieties, with diffuse large B-cell lymphoma (DLBCL) the most common nHL.

haematologica | 2022; 107(1)

Correspondence: JO VAN DORPE Jo.VanDorpe@uzgent.be Received: August 3, 2020. Accepted: November 25, 2020. Pre-published: December 10, 2020. https://doi.org/10.3324/haematol.2020.268813

211

L. Raman et al.

Both HL and DLBCL typically display distinct morphological, immunohistochemical and (epi)genetic features.2– 5 Their high cell turnover makes them excellent candidates for ‘liquid biopsy’-based research. Necrotic and apoptotic tumor cells have been shown to shed DNA6 into the peripheral circulation and this is informative with regard to diagnosis, prognosis and therapy.7 In theory, this would enable real-time (serial) sampling of the entire (often heterogeneous)8 genomic lymphoma architecture in a convenient and non-invasive manner through traditional blood sampling. Recent studies9–11 have focused on plasma cell-free DNA (cfDNA) analysis using ultra-deep targeted sequencing12 for single nucleotide variant (SNV) and translocation detection. Overall, these studies have provided two main insights: firstly, that variant allele frequencies often correlate with disease status; and, secondly, that surveilling circulating tumor DNA (ctDNA) might outperform 18F-fluoro-2-deoxyglucose positron emission tomography/computed tomography (PET/CT) scans in terms of sensitivity. This latter finding holds a great deal of potential for relapse risk assessment by quantifying minimal residual disease. Despite their potential, ultra-deep sequencing techniques are still expensive and require targeted panels. In contrast, shallow (coverage ~0.25x) whole-genome sequencing (sWGS) for primarily copy number detection is cheaper and fully operative in hospitals that offer noninvasive prenatal testing.13 One of the hallmarks of sWGS is its high specificity in identifying malignant cells: in comparison to SNV, large (i.e., >5 Mb) somatic copy number alterations are rarely detected in unaffected subjects14 whereas SNV accumulate over time. Worth mentioning is that this also occurs in hematopoietic stem cells, a phenomenon referred to as clonal hematopoiesis,15 potentially resulting in incorrect driver identifications. sWGS of cfDNA may therefore have clinical potential, especially for patients with lesions that are difficult to biopsy (e.g. those in brain; deep lymph nodes in the thoracic cavity or abdomen); or when dealing with ambiguous PET/CT scans, as both imaging and clinical symptoms of lymphoma are often non-specific. Nevertheless, this approach has only been superficially investigated.16 We, therefore, evaluated whether sWGS could serve as a molecular test in addition to established diagnostic methods, using a diverse set of 123 prospectively recruited lymphoma patients, comprising baseline and longitudinal blood samples, supplemented with paired formalin-fixed paraffin-embedded (FFPE) biopsies.

phocyte-depleted cHL; and 1 not otherwise specified [NOS]); 81 DLBCL (61 NOS; 4 Epstein-Barr virus [EBV]-positive; 10 primary mediastinal large B-cell lymphomas [PMBCL]; 2 high-grade Bcell lymphomas; 2 T-cell/histiocyte-rich large B-cell lymphomas; 1 intravascular large B-cell lymphoma; and 1 plasmablastic lymphoma) and four grey-zone lymphomas (GZL) (Online Supplementary File S2: Tables S1 and S2). Patients were included at the start, ‘baseline’, of a new line of therapy (i.e., baseline represents initial diagnosis for 115 cases; after ineffective treatment in 6 other cases; and at relapse for the 2 remaining cases). Although patient recruitment was random, the final cohort was enriched for HL patients.

Sample series Liquid biopsies were collected at baseline for all patients (n=123) and a random selection of paired FFPE tissue was made (n=33); this was obtained from 15 HL and 18 DLBCL patients. Longitudinal liquid biopsies were taken following milestones in treatment with 93 sequenced for 31 patients. These were intentionally enriched for refractory or relapsed disease: interim evaluations after two and four cycles of ABVD in HL; interim evaluations after four cycles of R-CHOP in nHL; at restaging after ineffective treatment or relapse; following successful treatment; and every 6 months for patients in maintained complete remission (CR). For HL (n=9), six patients had obtained CR within a year of inclusion whereas three were refractory. For nHL (n=22), 11 patients reached CR within a year; six were refractory; and five relapsed. Negative liquid biopsies were included from non-invasive prenatal assays (n=60), supplemented with benign FFPE tissue (n=9) as a control for the solid biopsies. In total, 318 samples were thus analyzed.

Clinical and pathological characteristics

Methods

Patient demographics and clinical variables were available through routine clinical practice. The total metabolic tumor volume (MTV) was measured by PET/CT using AW workstation semi-automated segmentation software (GE medical systems, Waukesha, WI, USA), with a standard uptake value cutoff of 2.5. Pathology diagnosis included standard histological examination of FFPE biopsy material in combination with immunohistochemistry (IHC). Double/triple hit (MYC, BCL2 and/or BCL6) and double expressor (MYC and BCL2) samples were detected using fluorescence in situ hybridization and IHC, respectively. The germinal center B-cell (GCB) cell of origin (COO) status was assigned through the Hans algorithm.17 When using chromogenic in situ hybridization (CISH), the presence of EBV-encoded RNA (EBER) was evaluated via INFORM EBER probes (Ventana Medical Systems, Tucson, AZ, USA). Selected liquid biopsies were additionally tested using the quantitative EBV-polymerase chain reaction (PCR) method, as established by Bordon et al.18

Ethics statement

Sequencing and bioinformatic analysis

This study was approved by the institutional ethics committee at Ghent University Hospital (EC/2016/0307). Written informed consent was obtained from all patients.

Each laboratory step and subsequent computational analysis can be found in Online Supplementary File S1: Supplemental methods. These include: copy number profiling;19 defining the tumor burden using the copy number profile abnormality (CPA) score20 and estimated tumor fraction;21 methods used to derive viral read fractions of HIV-1, HIV-2, EBV, JC polyomavirus, human Tlymphotropic virus 1, human herpesvirus 8 and hepatitis C virus; random forest modeling22 of copy number profiles to predict tumor subtype; detection of copy number driver peaks;23 and general statistical testing.

Study population Between January 2016 and November 2019, 123 lymphoma patients were recruited at Ghent University Hospital and AZ Delta Roeselare. The subtypes studied include 38 HL (3 nodular lymphocyte predominant HL; 23 nodular sclerosis classical HL [cHL]; 5 mixed cellularity cHL; 5 lymphocyte-rich cHL; 1 lym-

212

haematologica | 2022; 107(1)

Shallow-depth sequencing of lymphoma cell-free DNA

Results Shallow-depth sequencing of cell-free DNA enables accurate copy number profiling in lymphoma For healthy individuals, copy number profiles are assumed to be approximately ‘flat’, implying that large alterations are absent. In a previous study,20 in order to statistically distinguish cancerous from control profiles, we developed a score that increases with rising deviance from the naturally-occurring healthy background variance: the CPA score. Based on this variable, 95/123 baseline liquid biopsies were detected as cancerous (Figure 1A; Online Supplementary File S2: Table S3). As expected, the score reached significantly higher (P<0.001; Welch test) values in the lymphoma group compared to the control group. There was no significant difference (P=0.250; Fisher exact test) in the fraction of abnormal observations between HL (32/38; 84.2%) and DLBCL (60/81; 74.1%), nor when specifically considering cHL (31/35; 88.6%; P=0.091). At Ann Arbor stage I, only one of six liquid biopsies showed deviations whereas stage II to IV presented comparable detection rates (Online Supplementary File S1: Figure S1).

In order to confirm whether the copy number profiles derived from blood samples represent those from tissue biopsies, we executed a concordance analysis using 33 liquid-solid pairs (Online Supplementary File 1: Liquid-solid pairs). One representative DLBCL case, patient 59, demonstrated very similar deviations in both samples (Figure 1B), and the Pearson correlation between this patient’s smoothened profiles amounted to 0.916. We can note, from left to right, four clusters distinguished in a corresponding scatter plot (Figure 1C): deletions, copy neutral bins, duplications and amplifications. It should be noted that the amplitudes of alterations, a concept defined as the absolute value of a segment’s log ratio,19 are more extreme in liquid than in tissue biopsies (Figure 1B), caused by a greater ctDNA fraction in patient 59’s liquid biopsy. This is also indicated by a steeper identity line than least squares fit in the scatter plot (Figure 1C). Overall, when the CPA score in the liquid biopsy was abnormal, DLBCL pairs showed high concordance (r=0.815±0.043; 95% confidence interval). However, this was not the case for HL (r=0.260±0.055; 95% confidence interval) patients (Figure 1D). Next, the amplitudes were compared between paired 2

Figure 1. Feasibility of copy number profiling through sWGS of cfDNA. (A) CPA score comparison between common HL and DLBCL subtypes. Dots represent blood samples at baseline; default box plots indicate underlying distributions. The bottom grey box delineates the abnormality cutoff defined by the controls. According to this limit, the numbers at the top define which samples have abnormal copy number profiles. The overlapping histograms on the left display the overall CPA distribution. (B) Copy number profile comparison between the liquid and solid biopsy of patient 59. Dots are bins for which copy number is inferred whereas graph lines represent smoothened profiles. The ‘ratio’ equals the observed over the expected number of reads. (C) Scatter plot comparison of bins (grey dots) and smoothened values (solid black graph line) between the liquid and solid biopsy of patient 59. Identity line (black dotted) and least squares fit (red dotted), with corresponding Pearson correlation coefficient (r), are shown. (D) Scatter plot of all Pearson correlations between solid and liquid biopsy pairs, defined as in C, in relation to the CPA of the liquid biopsy. The grey box on the left matches the abnormality cutoff from A. Horizontal colored lines embedded in colored boxes represent means and their uncertainty (95% confidence interval), respectively. (E) Control-normalized amplitude comparison between the liquid and solid biopsies of HL and DLBCL patients. Samples are represented by dots; same-patient biopsies are connected; the underlying distribution is clarified by violin plots. P-values result from paired t-tests. sWGS: shallow whole-genome sequencing; cfDNA: cell-free DNA; CPA: copy number profile abnormality; HL: Hodgkin lymphoma; nHL: non-HL; DLBCL: diffuse large B-cell lymphoma; cHL: classical HL; ncHL: non-cHL; NLPHL: nodular lymphocyte predominant HL; NSCHL: nodular sclerosis cHL; MCCHL: mixed cellularity cHL; LRCHL: lymphocyte-rich cHL; GZL: grey zone lymphoma; NOS: not otherwise specified; GCB: germinal center B-cell; unc: unclassified; EBV: Epstein-Barr virus; PMBCL: primary mediastinal large B-cell lymphoma; HGBCL: high-grade B-cell lymphoma; THRLBCL: T-cell/histiocyte-rich large B-cell lymphoma; c, control.

haematologica | 2022; 107(1)

213

L. Raman et al.

liquid and solid biopsies (Figure 1E). These numbers were averaged per sample and additionally normalized by subtracting the mean across the control samples (0.005 for liquid; 0.026 for solid), which results in a fairer comparison, as FFPE material tends to produce profiles with higher levels of noise,24 naturally increasing the overall deviance. This variable was found to be higher in liquid than in solid biopsies for HL (P=0.010; paired ttest), but not for DLBCL (P=0.316; paired t-test). Worthy of note is that, within HL, the mean of the solid biopsies barely varied from zero, meaning that most of these profiles are hard to differentiate from normal controls, and, therefore, the low liquid-solid correlation observed in HL patients (Figure 1D), seems to be caused by a scarcity of tumor cells in the FFPE samples rather than a low ctDNA fraction. From a biological perspective, this is reasonable: neoplastic Hodgkin and Reed-Sternberg cells are always embedded in an inflammatory background of non-affected cells,25 causing resemblance to healthy tissue.

Shallow-depth sequencing of cell-free DNA enables sensitive Epstein-Barr virus detection in lymphoma Sequencing reads were mapped to seven different lymphoma-associated viruses. Of the lymphoma liquid biopsies, 39/216 had abnormally-elevated EBV read fractions compared to the negative controls (Online Supplementary File S2: Table S3); whereas one patient tested positive for JC polyomavirus (13 reads) at surveillance (patient 79). In accordance with clinical records, no other viruses were detected. Within control samples, no or exclusively low amounts of EBV reads were noted. The latter was the case for 15/60 liquid (read fraction range 0.032-0.059 parts per million [ppm]) and 6/9 solid controls (0.029-0.08 ppm). The EBV read fraction was evaluated for all baseline liquid biopsies, with 28/123 patients testing positive (Figure 2A). For GZL, 2/4 patients were positive, and, as expected, more HL samples were EBV-positive (15/38; 39.5%) than DLBCL samples (11/81; 13.6%) (P=0.004; Fisher exact test). All but one EBV-positive sample had an abnormal CPA

Figure 2. Feasibility of EBV detection through sWGS of cfDNA. (A) EBV read fraction comparison between common HL and DLBCL subtypes. Dots represent blood samples at baseline. The bottom grey box delineates the abnormality cutoff defined by the controls. According to this limit, the numbers at the top define which samples have abnormally-elevated EBV fractions. The overlapping histograms on the left display the overall EBV read fraction distribution. (B) Scatter plot between the CPA score and the EBV read fraction. Symbols represent blood samples at baseline; color shows whether CPA is abnormal; the + or – symbol is assigned in accordance to A, thereby specifying EBV positivity or negativity, respectively. Pie charts visualize contingency table; the P-value results from Fisher’s exact test. Swarm and box plots on the right-hand side indicate the EBV read fraction, grouped by CISH for EBER detection outcome. Dots represent blood samples; numbers at the top are assigned as in A. Colored dots are additionally tested by PCR in blood samples. (C) Control-normalized EBV comparison between the liquid and solid biopsies. Samples are represented by dots; same-patient biopsies are connected. Patients without elevated an EBV level in both their solid and liquid biopsy were omitted. The P-value results from a paired Wilcoxon rank-sum test. (D) ROC analyses executed using ranked viral-read fractions and three different gold standards: CISH, overridden by PCR; CISH only; solid biopsy EBV read fraction. EBV: Epstein-Barr virus; sWGS: shallow whole-genome sequencing; cfDNA: cell-free DNA; HL: Hodgkin lymphoma; nHL: non-HL; DLBCL: diffuse large B-cell lymphoma; CPA: copy number profile abnormality; CISH: chromogenic in situ hybridization; EBER: EBV-encoded RNA; PCR: polymerase chain reaction; cHL: classical HL; ncH:, non-cHL; ROC: receiver operating characteristic; AUC: area under the curve; NLPHL: nodular lymphocyte predominant HL; NSCHL: nodular sclerosis cHL; MCCHL: mixed cellularity cHL; LRCHL: lymphocyte-rich cHL; GZL: grey zone lymphoma; NOS: not otherwise specified; GCB: germinal center B-cell; unc: unclassified; PMBCL: primary mediastinal large B-cell lymphoma; HGBCL: high-grade B-cell lymphoma; THRLBCL: T-cell/histiocyte-rich large B-cell lymphoma; c: control.

214

haematologica | 2022; 107(1)

Shallow-depth sequencing of lymphoma cell-free DNA

score in addition (Figure 2B), which significantly associates the latter categorical variables (P=0.004; Fisher exact test). The sensitivity and specificity of EBV detection through sWGS of cfDNA was established by a comparison with routine CISH for EBER detection results, performed on FFPE tissue. All 18 samples that tested positive for EBER, as well as 8/46 EBER-negative samples, had abnormally-increased EBV levels in plasma, achieving a sensitivity and specificity of 100% and 82.6%, respectively (Figure 2B; Online Supplementary File S2: Tables S3 and S4). The liquid biopsies with the highest viral read fractions of the EBER-negative cases did, however, accommodate increased levels of EBV fragments in blood, as confirmed by a subsequent PCR analysis (Figure 2B). This indicates that routine CISH might have inadequate sensitivity or that viral fragments may have been absent in the tumor cells studied. When considering the latter five as true positives, specificity increases to 93.5%. As for the copy number alteration amplitudes, liquid biopsies had higher (P=0.039; paired Wilcoxon signed-rank test) EBV read fractions than solid biopsies (Figure 2C). To conclude, three receiver operating characteristic (ROC) analyses (Figure 2D) were included to further demonstrate

the performance of this EBV detection approach. Ranked plasma EBV viral read fractions and three different gold standards were used: CISH overridden by PCR (area under the curve [AUC]=0.994); CISH only (AUC=0.948); and solid biopsy reads (AUC=0.868).

Copy numbers derived from liquid biopsies can aid histological classification When visualizing fractions of aberrant samples across genomic loci per histological subtype, both general (e.g., gains at 2p, holding REL) and subtype-specific fingerprints can be distinguished (Figure 3). Unsupervised clustering26 applied to these summarizing profiles results in the anticipated histological hierarchy. Indeed, nHL and HL are separated as two definite entities, with the exception of GZL and PMBCL, which cluster alongside the HL group. Notwithstanding their non-Hodgkin histology, both have been described as being molecularly related to HL,27,28 hence the structure of the dendrogram. GZL in particular demonstrates overlapping features with cHL and PMBCL.28 To mimic how this translates as a potential clinical application for samples that are difficult to classify morphologically, or in cases in which obtaining tumor tissue is compli-

Figure 3. Overview of aberrations detected using liquid biopsies across lymphoma subtypes. Colored waves represent fractions of aberrant samples across genomic loci. Patterns above the y=0 lines indicate gains whilst opposite contours represent losses (e.g., when the top of the wave is at 50%, 50% of samples have a gain at the corresponding locus; when the bottom of the wave is at 10%, 10% of samples have a loss at the corresponding locus). The dendrogram (left) results from hierarchical clustering applied to the Pearson distances (i.e., d = [1 – r]/2) between the means of these waves. Graph lines represent smoothened mean log2 ratios. The number of used baseline samples (n) is indicated on the right-hand side. Subtypes represented by fewer than three patients were excluded. DLBCL: diffuse large Bcell lymphoma; GCB: germinal center B-cell; unc: unclassified; EBV: Epstein-Barr virus; GZL: grey zone lymphoma; PMBCL: primary mediastinal large B-cell lymphoma; HL: Hodgkin lymphoma; cHL: classical HL; ncHL: non-cHL; NSCHL: nodular sclerosis cHL; MCCHL: mixed cellularity cHL; LRCHL: lymphocyte-rich cHL; NLPHL: nodular lymphocyte predominant HL.

haematologica | 2022; 107(1)

215

L. Raman et al.

cated, random forest predictive modeling, in combination with cross-validation, was applied to four differential diagnoses (Figure 4A): HL versus DLBCL; cHL versus PMBCL; DLBCL, NOS versus PMBCL; and within DLBCL, NOS, GCB versus non-GCB. For these analyses, obviously only baseline liquid biopsies were employed. Two ROC AUC can be discussed here: one for all samples and one disregarding flat copy number profiles. The latter statistic is considered in the next paragraph as it represents the performance of a rational application: profiles without aberrations will never be subjected to machine learning algorithms since predictions would be random. The differential diagnosis of HL and DLBCL, using computationally-processed copy number profiles, was found to be accurate (AUC of 0.967). Out of bag principles (i.e., cross-validation for bootstrapping methods, to overcome train/test bias) were applied to a set of 119 samples, meaning that, at each iteration, sufficient data were available for model training; and, biologically, HL and DLBCL express divergent profiles (Figure 3). PMBCL has features between those of cHL and DLBCL.27 Nonetheless, copy number profiles can aid differential diagnoses, indicated by AUC values of 0.931 and 0.981, respectively. Finally, the GCB versus non-GCB COO status does not seem to trigger specific copy number alterations, as shown by the resulting AUC of 0.450. We could not test whether the activated B-cell versus GCB distinction by gene expression profiling29 would perform better. This should be explored in future work. During training, learning algorithms assign higher weights to features that improve classification. For the HL

versus DLBCL problem, the top three discriminative features - or loci, in our case - are located at 2p, 9p and 11q (Figure 4B, bottom). In addition, the GISTIC software aims to define driver genes in pattern peaks across a set of samples. Of note, there were three interesting peaks in these regions that were significantly enriched (Figure 4B, top), matching several anticipated genes: in both HL and DLBCL, 2p16.1 (REL) gains;30,31 9p24.1 (JAK2 and PD-L1) gains in HL;32 9p21.3 (CDKN2A) losses in DLBCL;33 and to conclude, 11q23.3 (CEP164) gains in DLBCL34 (Online Supplementary File S2: Table S5). Although 2p16.1 was found to be amplified, or duplicated, in both HL and DLBCL, the model gives importance to the 2p arm, as these aberrations are more frequently detected in HL compared to DLBCL (Figure 4B, top).35 Finally, locus importance was examined for the two other well-performing models: to a certain extent, 5p, 9q, and chromosomes 14 and 19 allow separation of cHL and PMBCL, while more specific 7p and, once more, 9p24.1 (JAK2 and PD-L1)32 cytobands are decisive in differentiating DLBCL from PMBCL (Figure 4C).

Longitudinal liquid biopsies correlate with overall disease status Serial blood samples were analyzed to observe copy number alteration dynamics over time and to study longitudinal changes in CPA and EBV read fractions. One example of an intra-patient evolution is demonstrated by patient 32, who had refractory DLBCL (Figure 5A). In this case, alterations disappeared during treatment (MTV on PET/CT

Figure 4. Feasibility of copy number profiles as a clinical lymphoma subtyping tool. (A) ROC analysis following out of bag random forest modeling to differentially diagnose HL versus DLBCL; cHL versus PMBCL; DLBCL, NOS versus PMBCL; and within DLBCL, NOS, GCB versus non-GCB. Exclusively, blood samples at baseline were included. ROC curves when omitting flat profiles (i.e., only CPA abnormal cases) are shown in addition. (B) The top plot is generated as in Figure 3. The black graph line shows the mean absolute difference between these profiles. The four horizontal grey bars are colored at significantly-enriched aberration peaks, assumed to hold driver genes, detected by GISTIC (gains on top, losses below; yellow for DLBCL, green for HL; emphasized by (inverted) ‘V’ symbols). The plot below visualizes the importance of regions for the differential diagnosis according to a HL vs. DLBCL random forest. (C) Similar to (B) but for the cHL versus PMBCL and DLBCL, NOS versus PMBCL classification problems. ROC: receiver operating characteristic; AUC: area under the curve; CPA: copy number profile abnormality; HL: Hodgkin lymphoma; cHL: classical HL; DLBCL: diffuse large B-cell lymphoma; PMBCL: primary mediastinal large B-cell lymphoma; NOS: not otherwise specified; GCB: germinal center B-cell.

216

haematologica | 2022; 107(1)

Shallow-depth sequencing of lymphoma cell-free DNA

decreased from 915 cm3 at baseline to 4 cm3 between both interim moments), and reappeared when progression was diagnosed after 212 days (MTV increased to 79 cm3). The smoothened profiles at initial staging and subsequent restaging at progression were very similar (r= 0.946; Figure 5B), notwithstanding the long time interval, suggesting that

a portion of ‘remnant’ tumor had never been completely eliminated and that minimal residual disease had caused recidivism. Despite the significant correlation, evidence of tumor evolution was present, as indicated by several changes along the copy number alteration pattern, e.g.,on chromosome arms 2q and 6p; and on chromosome 14

Figure 5. Copy number profile dynamics and EBV read fractions across serial liquid biopsies. (A) Four smoothened profiles corresponding to longitudinal samples of refractory patient 32, colored in - ordered chronologically - black (diagnosis), green (response), yellow (response) and brown (progression). Arrow on top indicates relative point in time of each sample collection moment. The grey background represents the 95% confidence interval of healthy controls. (B) Scatter plot of smoothened copy number profiles between sample one and four from (A). Scatter symbols are tiny characters, representing chromosomes; for perceptibility, each chromosome character is uniquely colored. (C) Dot plot of pairwise Pearson correlations between copy number profiles at longitudinal staging moment. Patients are represented by solid (refractory) or dotted lines (relapsed). Each dot shows the correlation between two staging moments. Horizontal lines indicate means. (D) Swarm plot CPA score comparison of the most recent follow-up liquid biopsy between patients in CR and patients that are not. The grey box shows the abnormality cutoff. (E) Longitudinal EBV read fraction evaluation. Exclusively EBV-positive patients, according to sWGS data, are included. Same-patient samples are connected with lines. The timescale is relative, where 0 corresponds to baseline. The grey box shows the abnormality cutoff. Colors represent a patient’s current disease status. Thicker trend lines result from loess fitting. (F) The same as (E) but evaluates CPA score rather than EBV read fraction. All HL patients with followup are included. Swarm and box plots on the righthand side additionally indicate the CPA score distribution, grouped according to whether the samples were drawn at staging, interim or surveillance. (G) Same as (F), but for nHL patients. EBV: Epstein-Barr virus; CPA: copy number profile abnormality; CR: complete remission; sWGS: shallow whole-genome sequencing; HL: Hodgkin lymphoma; nHL: non-HL.

haematologica | 2022; 107(1)

217

L. Raman et al.

(Figure 5A). These zones are represented by irregularities in the scatter plot trend between the profiles (Figure 5B). Overall, observations were consistent for refractory and relapsed patients: when detectable, copy number patterns were very similar across different longitudinal staging moments (Figure 5C). One exception was found in patient 73 (rightmost patient in Figure 5C), who appeared to relapse with a divergent copy number profile (Online Supplementary File S1: Longitudinal samples). This case concerned an aggressive EBV-positive plasmablastic lymphoma that is known to present complex karyotypic abnormalities and, presumably, an instable genome.36 The most recent follow-up of all 17 patients in CR showed that none had remaining copy number alterations; in contrast, this was the case in only 5/14 patients in whom CR was thought not to be reached (Figure 5D). In total, five EBV-positive patients, according to sWGS of cfDNA, were detected at diagnosis of the 31 patients with follow-up. Of these, three went into CR and, consequently, no longer had abnormally elevated EBV levels; one refractory patient (patient 6) was noted to have recurrently high EBV levels and one relapsed patient (patient 73) was detected with an unusually high EBV read fraction, following an earlier drop to 0 ppm at first response evaluation (Figure 5E). It should be noted that, for this patient, no tumor was detected under PET/CT at the fourth time point (Online Supplementary File 2: Table S4), however, the steadily increasing EBV fraction anticipated actual relapse at the fifth time point, after which the MTV increased rapidly along with the plasma concentration of viral fragments (Online Supplementary File S1: Longitudinal samples). A similar analysis, concerning the CPA score, was performed for all patients with HL (Figure 5F) and nHL (Figure 5G) with follow-up. Similar conclusions were made according to the trend lines, which largely summarize the longitudinal analyses: a prompt drop for patients in CR; varying values for refractory patients; and a drop, followed by a newly-triggered rise, for relapsed patients. It is significant how the CPA decreased rapidly when treatment was initiated: interim samples had normal CPA scores unless they represented a refractory patient. Finally, 6/31 patients never transcended the CPA abnormality cutoff, meaning that these, probably, have insufficient ctDNA. In two cases (patients 66 and 71), however, the MTV reached extreme levels (>1000 cm3) while the CPA score stayed low (Online Supplementary File S1: Longitudinal samples). For these patients, it is more likely that copy number alterations really were absent. Of the above cases, additional clinical information from four interesting patients further illustrates the potential of shallow-depth sequencing for disease monitoring - e.g., when dealing with ambiguous PET/CT images or tumor and viral entities beyond initial suspicion. These are discussed as short case-reports in Online Supplementary File S1: Case reports.

Copy number anomalies detected in liquid biopsies correlate with clinical parameters Five lymphoma-related clinical variables were included in an extensive concordance analysis of the baseline and longitudinal liquid biopsies: Ann Arbor stage; International Prognostic Index/Score; lactate dehydrogenase concentration; b2 microglobulin concentration; and MTV (Online Supplementary File S1: Figure S2). The estimated tumor fraction, according to copy number profiles, positively and sig218

nificantly correlates with all of these clinical variables, rendering the former estimate relevant. The CPA score is intrinsically correlated to the tumor fraction, so similar associations as for the tumor fraction arise. Since all these variables are naturally cross-related, the derived relations are obviously not claimed to be independently significant.

Discussion To date, molecular profiling of lymphoma has largely been focused on the characterization of actionable targets, such as point mutations and translocations. This profiling requires panel sequencing, currently using tissue biopsies as the DNA source and targeting a substantial number of genes. However, until now, the availability of this precise information has contributed little to actual progress in patient care, despite its tempting rationale.37 In addition, applying these concepts in a minimally-invasive manner requires ultra-deep sequencing (~2000x),38 of which the implementation in routine practice remains challenging. In this study, we evaluated sWGS of cfDNA for the diagnosis, differential diagnosis, and disease monitoring of HL and DLBCL. In contrast to targeted sequencing, this approach provides genomic insight by copy number profiling in an accessible and standardized manner. Indeed, noninvasive prenatal testing, which involves the same laboratory and computational steps, has evolved into an application performed daily in molecular diagnostic laboratories. The short turnaround time (approximately, 4 days) and lowprice tag (around $200, including processing costs) renders sWGS a fortiori feasible. Copy number profiling is not a surrogate for targeted sequencing, nor does it enable the characterization of phenotypic traits, such as CD20 expression. However, to a limited extent, it does enable the prediction of targetable genes, such as JAK2, PD-L1 and REL. Further, regarding tumor fraction and specificity, it has the advantage of easier and more stable interpretation: once a copy number alteration is visible, the presence of sufficient ctDNA is most likely ascertained given that large somatic copy number alterations are rarely detected in unaffected subjects.14 For DLBCL, the sensitivity of copy number profiling is a shortcoming: only 74.1% of baseline liquid biopsies were found to have detectable aberrations, which is lower than that reported from targeted ultra-deep duplex sequencing studies (e.g., 98% in Kurtz et al.39). On the other hand, the sensitivity was surprisingly high for HL (84.2% for HL overall and 88.6% for cHL in this study; 88.9% in a smaller proof-of-concept study16) and compares to that of SNVbased studies (e.g., 81.2% in Spina et al.11). It ought to be noted that the lower limit of detection for CNA is often considered to be 3% tumor fraction.21 It is worth mentioning that this limit could be lowered as paired-end sequencing at greater depths (1x-1.5x) is expected to become available for routine diagnosis at a similar cost. This would empower computational tumor-read enrichment by insertsize filtering, as shorter cfDNA fragments are more likely be tumor-derived, thus increasing sensitivity.20,40,41 The signs and symptoms of aggressive lymphoma are often non-specific, having much in common with a broad range of different disorders, including autoimmune, immunodeficiency and systemic diseases, infections and other malignancies. Therefore, currently, accurate lymphoma diagnosis is essentially only possible through invahaematologica | 2022; 107(1)

Shallow-depth sequencing of lymphoma cell-free DNA

sive tissue biopsy. However, affected sites are sometimes difficult to reach and tissue-based subtyping may be complicated when dealing with inconclusive or small biopsies, or partially-involved lymph nodes. Since several cases of differential diagnosis by computationally-processed copy number profiles have been found to be accurate, copy number profiling of cfDNA might enter clinical practice as a test for patients with: suspected lymphoma with a mediastinal mass (to guide the need for mediastinoscopy); deep-seated pathologic lymph nodes in the abdomen, especially following surgery; a remaining suspicion of lymphoma after a negative biopsy, caused by ambiguous PET/CT images; probable central nervous system lymphoma (to avoid radical surgery); probable intravascular localization;42 and, generally, lymphoma in children (to avoid unnecessary surgery). For these indications, the approach presented here could be easily integrated with other tests in order to decide whether invasive procedures are still necessary. sWGS of cfDNA was found to be 100% sensitive for the detection of EBV. However, one fact that could preclude its application in lymphoma diagnosis is that all possible causes of elevated EBV levels (e.g., ongoing infectious mononucleosis) are detected simultaneously through blood samples irrespectively of whether these are related to lymphoma. CISH does not have this shortcoming and can be used to assess a neoplastic association. Nonetheless, the combination of an increased EBV level and typical HL or DLBCL alterations strongly indicates an EBV-driven lymphoma. This is especially relevant for the diagnosis of post-transplant or immunodeficiency-associated lymphoma. sWGS could be employed, in part, to detect EBV prior to performing a CISH test (around $140), when required. The latter test should only be performed when EBV is positive. Future in-house research will focus on the inclusion of other (rarer) lymphoma subtypes, including T-cell lymphomas, to obtain a broader image of the alterations’ speci-

References 1. Shaffer AL, Young RM, Staudt LM. Pathogenesis of human B cell lymphomas. Annu Rev Immunol. 2012;30:565-610. 2. Re D, Küppers R, Diehl V. Molecular pathogenesis of Hodgkin’s lymphoma. J Clin Oncol. 2005;23(26):6379-6386. 3. Steidl C, Connors JM, Gascoyne RD. Molecular pathogenesis of Hodgkin’s lymphoma: increasing evidence of the importance of the microenvironment. J Clin Oncol. 2011;29(14):1812-1826. 4. Schmitz R, Wright GW, Huang DW, et al. Genetics and pathogenesis of diffuse large B-cell lymphoma. N Engl J Med. 2018;378 (15):1396-1407. 5. Steidl C, Gascoyne RD. The molecular pathogenesis of primary mediastinal large B-cell lymphoma. Blood. 2011;118(10): 2659-2669. 6. Schwarzenbach H, Hoon DSB, Pantel K. Cell-free nucleic acids as biomarkers in cancer patients. Nat Rev Cancer. 2011;11(6): 426-437. 7. Domínguez-Vigil IG, Moreno-Martínez AK, Wang JY, Roehrl MHA, BarreraSaldaña HA. The dawn of the liquid biopsy in the fight against cancer. Oncotarget. 2017;9(2):2912-2922. 8. Araf S, Wang J, Korfi K, et al. Genomic profiling reveals spatial intra-tumor hetero-

haematologica | 2022; 107(1)

ficity compared to subtype. Studies will also concentrate on novel methods to study cfDNA (epi)genetics, including cellfree reduced-representation bisulfite sequencing for methylation detection, which are likely to improve specificity. Since this study was concluded, new liquid biopsies have been analyzed by sWGS at our institutions, mostly in cases of diagnostic ambiguity. While negative results currently never exclude the presence of malignancies, positive results ascertain disease. Disclosures No conflicts of interest to disclose. Contributions Study supervision: FO and JVD; patient recruitment: CDV, DD, FO, KDG and SB; clinical/pathological data acquisition/generation: CDV, BVDB, KM, DD, FD, SV, AD, FO and JVD; data processing: MVDL; study design: LR, MVDL, CDV, BM, FO and JVD; data analysis: LR; manuscript writing: LR and JVD; all authors read, revised and approved the final manuscript. Acknowledgments The authors would like to thank all collaborators at AZ Delta Roeselare and Ghent University Hospital for the active cohort recruitment, sample collection and provision of individual patient data. Funding This study was supported by Bijzonder Onderzoeksfonds (BOF), Ghent University, in the form of a doctoral research grant (BOF.STA.2017.0002.01 to LR). Data-sharing statement Copy number profiles and viral reads counts of all sequenced samples are included in Online Supplementary file: Table S6 and S7, respectively.

geneity in follicular lymphoma. Leukemia. 2018;32(5):1261-1265. 9. Rossi D, Diop F, Spaccarotella E, et al. Diffuse large B-cell lymphoma genotyping on the liquid biopsy. Blood. 2017;129(14): 1947-1957. 10. Bohers E, Viailly P-J, Becker S, et al. Noninvasive monitoring of diffuse large B-cell lymphoma by cell-free DNA high-throughput targeted sequencing: analysis of a prospective cohort. Blood. Cancer J. 2018;8(8):74. 11. Spina V, Bruscaggin A, Cuccaro A, et al. Circulating tumor DNA reveals genetics, clonal evolution, and residual disease in classical Hodgkin lymphoma. Blood. 2018;131(22):2413-2425. 12. Newman AM, Bratman SV, To J, et al. An ultrasensitive method for quantitating circulating tumor DNA with broad patient coverage. Nat Med. 2014;20(5):548-554. 13. Raman L, Baetens M, De Smet M, Dheedene A, Van Dorpe J, Menten B. PREFACE: In silico pipeline for accurate cell-free fetal DNA fraction prediction. Prenat Diagn. 2019;39(10):925-933. 14. Redon R, Ishikawa S, Fitch KR, et al. Global variation in copy number in the human genome. Nature. 2006;444(7118):444-454. 15. Busque L, Patel JP, Figueroa M, et al. Recurrent Somatic TET2 Mutations in Normal Elderly Individuals With Clonal

Hematopoiesis. Nat Genet. 2012;44(11): 1179-1181. 16. Vandenberghe P, Wlodarska I, Tousseyn T, et al. Non-invasive detection of genomic imbalances in Hodgkin/Reed-Sternberg cells in early and advanced stage Hodgkin’s lymphoma by sequencing of circulating cell-free DNA: a technical proof-of-principle study. Lancet Haematol. 2015;2(2):e5565. 17. Hans CP, Weisenburger DD, Greiner TC, et al. Confirmation of the molecular classification of diffuse large B-cell lymphoma by immunohistochemistry using a tissue microarray. Blood. 2004;103(1):275-282. 18. Bordon V, Padalko E, Benoit Y, Dhooge C, Laureys G. Incidence, kinetics, and risk factors of Epstein-Barr virus viremia in pediatric patients after allogeneic stem cell transplantation. Pediatr Transplant. 2012; 16(2):144-150. 19. Raman L, Dheedene A, De Smet M, Van Dorpe J, Menten B. WisecondorX: improved copy number detection for routine shallow whole-genome sequencing. Nucleic Acids Res. 2019;47(4):1605-1614. 20. Raman L, Van der Linden M, Van der Eecken K, et al. Shallow whole-genome sequencing of plasma cell-free DNA accurately differentiates small from non-small cell lung carcinoma. Genome Med. 2020;12(1):35.

219

L. Raman et al. 21. Adalsteinsson VA, Ha G, Freeman SS, et al. Scalable whole-exome sequencing of cellfree DNA reveals high concordance with metastatic tumors. Nat Commun. 2017; 8(1):1324. 22. Liaw A, Wiener M. Classification and Regression by random Forest. R News. 2002;2(3):18-22. 23. Mermel CH, Schumacher SE, Hill B, Meyerson ML, Beroukhim R, Getz G. GISTIC2.0 facilitates sensitive and confident localization of the targets of focal somatic copy-number alteration in human cancers. Genome Biol. 2011;12(4):R41. 24. Van der Linden M, Raman L, Vander Trappen A, et al. Detection of copy number alterations by shallow whole-genome sequencing of formalin-fixed, paraffinembedded tumor tissue. Arch Pathol Lab Med. 2020;144(8):974-981. 25. Mathas S, Hartmann S, Küppers R. Hodgkin lymphoma: Pathology and biology. Semin Hematol. 2016;53(3):139-147. 26. Ward JH Jr. Hierarchical grouping to optimize an objective function. J Am Stat Assoc. 1963;58(301):236-244. 27. Rosenwald A, Wright G, Leroy K, et al. Molecular diagnosis of primary mediastinal B cell lymphoma identifies a clinically favorable subgroup of diffuse large B cell lymphoma related to Hodgkin lymphoma. J Exp Med. 2003;198(6):851-862. 28. Eberle FC, Rodriguez-Canales J, Wei L, et al. Methylation profiling of mediastinal gray zone lymphoma reveals a distinctive signature with elements shared by classical Hodgkin’s lymphoma and primary mediastinal large B-cell lymphoma.

220

Haematologica. 2011;96(4):558-566. 29. Alizadeh AA, Eisen MB, Davis RE, et al. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature. 2000;403(6769):503-511. 30. Joos S, Menz CK, Wrobel G, et al. Classical Hodgkin lymphoma is characterized by recurrent copy number gains of the short arm of chromosome 2. Blood. 2002;99(4): 1381-1387. 31. Houldsworth J, Mathew S, Rao PH, et al. REL proto-oncogene is frequently amplified in extranodal diffuse large cell lymphoma. Blood. 1996;87(1):25-29. 32. Green MR, Monti S, Rodig SJ, et al. Integrative analysis reveals selective 9p24.1 amplification, increased PD-1 ligand expression, and further induction via JAK2 in nodular sclerosing Hodgkin lymphoma and primary mediastinal large B-cell lymphoma. Blood. 2010;116(17):3268-3277. 33. Lenz G, Wright GW, Emre NCT, et al. Molecular subtypes of diffuse large B-cell lymphoma arise by distinct genetic pathways. Proc Natl Acad Sci U S A. 2008;105(36):13520-13525. 34. Pasqualucci L, Trifonov V, Fabbri G, et al. Analysis of the coding genome of diffuse large B-cell lymphoma. Nat Genet. 2011;43(9):830-837. 35. Kober-Hasslacher M, Schmidt-Supprian M. The unsolved puzzle of c-Rel in B cell lymphoma. Cancers. 2019;11(7):941. 36. Castillo JJ, Reagan JL. Plasmablastic lymphoma: a systematic review. Scientific World Journal. 2011;11:687-696. 37. Iacoboni G, Zucca E, Ghielmini M, Stathis A. Methodology of clinical trials evaluating

the incorporation of new drugs in the firstline treatment of patients with diffuse large B-cell lymphoma (DLBCL): a critical review. Ann Oncol. 2018;29(5):1120-1129. 38. Scherer F, Kurtz DM, Newman AM, et al. Distinct biological subtypes and patterns of genome evolution in lymphoma revealed by circulating tumor DNA. Sci Transl Med. 2016;8(364):364ra155. 39. Kurtz DM, Scherer F, Jin MC, et al. Circulating tumor DNA measurements as early outcome predictors in diffuse large Bcell lymphoma. J Clin Oncol. 2018;36(28): 2845-2853. 40. Mouliere F, Mair R, Chandrananda D, et al. Detection of cell-free DNA fragmentation and copy number alterations in cerebrospinal fluid from glioma patients. EMBO Mol Med. 2018;10(12):e9323. 41. Cristiano S, Leal A, Phallen J, et al. Genome-wide cell-free DNA fragmentation in patients with cancer. Nature. 2019;570(7761):385-389. 42. Deeren D, Van Der Linden M, Dedeurwaerdere F, et al. Circulating cellfree DNA for response evaluation of intravascular lymphoma. Ann Hematol. 2019;98(8):2021-2023. 43. Rinaldi A, Kwee I, Poretti G, et al. Comparative genome-wide profiling of post-transplant lymphoproliferative disorders and diffuse large B-cell lymphomas. Br J Haematol. 2006;134(1):27-36. 44. Morscio J, Dierickx D, Tousseyn T. Molecular pathogenesis of B-cell posttransplant lymphoproliferative disorder: what do we know so far? Clin Dev Immunol. 2013;2013:150835.

haematologica | 2022; 107(1)

ARTICLE

Non Hodgkin-Lymphoma

Baseline SUV is related to tumor cell proliferation and patient outcome in follicular lymphoma max

Cédric Rossi,1-5 Marie Tosolini,1,3,6,7 Pauline Gravelle,1-4,6 Sarah Pericart,1,6 Salim Kanoun,8 Solene Evrard,6 Julia Gilhodes,9 Don-Marc Franchini,1-4 Nadia Amara,6 Charlotte Syrykh,6, 10,11 Pierre Bories,10,11 Lucie Oberic,11 Loïc Ysebaert,1-4,11 Laurent Martin,12,13 Selim Ramla,12,13 Philippine Robert,5,13 Claire Tabouret-Viaud,14 René-Olivier Casasnovas,5,13 Jean-Jacques Fournié,1-4 Christine Bezombes1-4# and Camille Laurent1-4,6# Centre de Recherches en Cancérologie de Toulouse (CRCT), UMR1037 INSERM, Université Toulouse III Paul-Sabatier, ERL5294 CNRS, Toulouse; 2Laboratoire d’Excellence TOUCAN, Toulouse; 3Programme Hospitalo-Universitaire en Cancérologie CAPTOR, Université Toulouse, Toulouse; 4CALYM Carnot Institute, Pierre-Bénite; 5CHU Dijon, Hématologie Clinique, Hôpital François Mitterrand, Dijon; 6Département de Pathologie, Institut Universitaire du Cancer de Toulouse, Université Toulouse, Toulouse; 7Pôle Technologique du Centre de Recherches en Cancérologie de Toulouse, Toulouse; 8Médecine Nucléaire, Institut Universitaire du Cancer ToulouseOncopole, Université Toulouse, Toulouse; 9Bureau des Essais Cliniques, Institut Universitaire du Cancer Toulouse-Oncopole, Université Toulouse, Toulouse; 10Réseau Régional de Cancérologie, Onco-Occitanie, Institut Universitaire du Cancer, Université Toulouse, Toulouse-Oncopole; 11Service d’Hématologie, Institut Universitaire du Cancer de Toulouse, Université Toulouse Toulouse; 12Département de Pathologie, CHU Hôpital François Mitterrand, Dijon; 13INSERM UMR 1231 UFR, Bourgogne and 14 Département de Médecine Nucléaire, Centre Georges-François Leclerc, Dijon, France 1

Ferrata Storti Foundation

Haematologica 2022 Volume 107(1):221-230

CB and CL contributed equally as co-senior authors.

Correspondence:

ABSTRACT

ollicular lymphoma (FL) is the most common indolent lymphoma. Despite the clear benefit of CD20-based therapy, a subset of FL patients still progress to aggressive lymphoma. Thus, identifying early biomarkers that incorporate positron emission tomography metrics could be helpful to identify patients with a high risk of treatment failure with rituximab. We retrospectively included a total of 132 untreated FL patients separated into training and validation cohorts. Optimal threshold of baseline whole-body maximum standardized uptake (SUV ) was first determined in the training cohort (n=48) to predict progression-free survival (PFS). The PET results were investigated along with the tumor and immune microenvironment, which were determined by immunochemistry and transcriptome studies involving gene set enrichment analyses and immune cell deconvolution, together with the tumor mutation profile. We report that baseline SUV >14.5 was associated with poorer PFS than baseline SUV ≤14.5 (hazard ratio =0.28; P=0.00046). Neither immune T-cell infiltration nor immune checkpoint expression were associated with baseline PET metrics. By contrast, FL samples with Ki-67 staining ≥10% showed enrichment of cell cycle/DNA genes (P=0.013) and significantly higher SUV values (P=0.007). Despite similar oncogenic pathway alterations in both SUV groups of FL samples, four out of five cases harboring the infrequent FOXO1 transcription factor mutation were seen in FL patients with SUV >14.5. Thus, high baseline SUV reflects FL tumor proliferation and, together with Ki-67 proliferative index, can be used to identify patients at risk of early relapse with rituximab chemotherapy. max

max

CHRISTINE BEZOMBES christine.bezombes@inserm.fr CAMILLE LAURENT laurent.camille@iuct-oncopole.fr CÉDRIC ROSSI cedric.rossi@chu-dijon.fr Received: June 24, 2020. Accepted: December 11, 2020. Pre-published: December 17, 2020. https://doi.org/10.3324/haematol.2020.263194

max

haematologica | 2022; 107(1)

Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or internal use. Sharing published material for non-commercial purposes is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for commercial purposes is not allowed without permission in writing from the publisher.

221

C.Rossi et al.

Introduction Although some patients exhibit long-term remission after anti-CD20-based therapy, 20-30% patients with follicular lymphoma (FL) experience early progression,1 and FL transforms into aggressive lymphoma in 2-3% patients per year.2 Determining the prognostic factors that could be used in the early stages to identify patients with a high risk of treatment failure has become a central challenge in the management of FL.3 Follicular Lymphoma International Prognostic Index (FLIPI) and FLIPI2 scores, which include baseline clinical and standard biological parameters, are not able to accurately identify early relapse associated with an increased risk of death. Because FL is fluorodeoxyglucose (FDG) avid, FDG positron emission tomography (PET) is used in routine practice to stage pretreatment disease and to identify sites with high FDG uptake that are at risk of transformation.4 The predictive power of postinduction PET status on outcome appears to be much stronger than FLIPI or FLIPI2 scores and computed tomography (CT)-based response, and most patients who achieve PET negativity can expect their first remission to last several years.5 However, the critical challenge is to identify patients who have a high risk of standard treatment failure before initiating therapy. Among baseline PET metrics, total metabolic tumor volume (TMTV) is a strong predictor of outcome in FL, independent from FLIPI2.6 Baseline whole-body maximum standardized uptake (SUV ) has been used to predict outcomes in mantle cell lymphoma patients,7 but to our knowledge the value of pre-treatment SUV for prognosis in FL patients treated with rituximab (R)-chemotherapy followed by R maintenance (Rm) remains unclear. Very recently, a retrospective study8 reported that pre-treatment SUV >18 was associated with lower overall survival (OS) in a cohort of 346 advanced stage FL patients, but only a few these patients received heterogeneous induction treatments with maintenance therapy. The biological basis of high SUV at baseline remains unclear, and the cell populations, either in tumors or their microenvironment, that mediate a high uptake of FDG remain unknown. However, a pro-tumor microenvironment and its interactions with cancer cells could play a key role in promoting tumor cell growth and invasion.9 Recently, we developed a set of 33 genes involved in tumor immune escape (Immune Escape Gene Set [IEGS33)]. IEGS33 includes the genes encoding for immune checkpoints (ICP) (e.g., CTLA4, PDCD1, LAG3, HAVCR2), for their ligands (e.g., CD80, CD86, CD274, PDCD1LG2, LGALS9), for enzymes producing immunosuppressive metabolites (e.g., IDO1, ARG1, ENTPD1), and for immunosuppressive cytokines and chemokines (e.g., IL10, HGF, GDF15). We discovered that the whole IEGS33 gene set is significantly up-regulated in all non-Hodgkin lymphoma (NHL) samples.10 Although immune escape strategies in lymphoma may vary between individuals, our former analysis of 1446 B-NHL transcriptomes evidenced the consistent up-regulation of IEGS33 in B-cell lymphomas.10 Furthermore, because the activation of immune effectors represents the ‘substrate’ of immune escape (IE), our metaanalysis of both ‘T-cell activation’ (44 T-cell genes such as IL2, CD28, ZAP70, LCK) and IEGS33 outlined four different stages of IE. These stages are seen in NHL patients in whom immune activation and escape are not observed, those in whom both are observed, those with mostly max

immune activation, and those with mostly immune escape. Patients from these four classes displayed correspondingly different rates of overall survival (OS). Along the same lines, another study identified a signature of 23 genes involved in tumor cell cycle events (B-cell development, DNA damage response, cell migration and cell cycle) and immune regulation, which was predictive of progression-free survival (PFS) and progression of disease within 24 months (POD24) in FL patients with a high tumor burden treated with R-chemotherapy plus maintenance.11 The tumor cell mutation profile may also influence patient outcomes. For instance, in combination with the FLIPI score, the mutation profile of seven genes (m7-FLIPI) involved in B-cell lymphomagenesis has been used to define a clinicogenetic risk index in FL patients receiving R-chemotherapy combined with cyclophosphamide, doxorubicin, vincristine and prednisone (R-CHOP).12 So far, however, none of these expression and mutational signatures has been analyzed with regard to baseline PET metrics. PET has become an important tool for staging and response assessment in FL patients. For this reason, the relationships between the molecular and functional imaging parameters need to be explored in order to ascertain whether they could improve FL patient risk stratification in the initial staging. The objective of our study was therefore to determine the prognostic value of PET metrics at baseline in FL patients, and to establish links to the molecular signatures of FL tumor cells and their immune cell infiltration.

max

222

Methods Patients The training cohort consisted of 48 FL patients (diagnosed by CL, CS or SP according to the World Health Oragnization classification13) treated in the Department of Hematology of the IUCT Oncopole (Toulouse, France) between 2011 and 2016. The validation cohort included 84 additional FL patients (diagnosed by LM or SR) treated in the Department of Hematology of the Dijon Bourgogne University Hospital (France) between 2012 and 2016. Patient characteristics are detailed in the Online Supplementary Table S1. Tissue samples were collected and processed at the CRB Cancer des Hôpitaux de Toulouse and CRB Cancer du CHU de Dijon following ethics guidelines (Declaration of Helsinki), and written informed consent was obtained from all patients. CRB collections were declared to the Ministry of Research (DC-2009-989 for Toulouse, and DC-2008-508 for Dijon) and a transfer agreement (AC-2008-820) was obtained after approval from the appropriate ethics committees.

Positron emission tomography/computed tomography acquisition and analysis Baseline PET acquisition was performed before any treatment and detailed in the Online Supplementary Methods. PET images at baseline were centrally reviewed by one experienced reader (SK), who was blinded to any medical information, and analyzed using the free open-source software, Beth Israel Plugin for Fiji (http://petctviewer.org). PET and CT images (single modality and fused images) were displayed in three axes with multi-planar reconstruction along with maximum intensity projection. Pathological uptake was defined as an increased uptake of 18 FDG over the physiological background. For each PET, whole-

haematologica | 2022; 107(1)

Baseline SUV in FL and tumor proliferation max

body SUV was recorded. The whole-body SUV corresponded to the single hottest tumor voxel in the whole-body acquisition. In order to determine TMTV, we performed the calculation based on relative threshold (>41% SUV threshold as recommended by the European Association of Nuclear Medicine) using Beth Israel Plugin.14 max

max

Data mining and transcriptome analyses For gene expression analysis, RNA was extracted from the available frozen tumor samples (n=38 of 48). cDNA was prepared from minimum 500 pg RNA per sample and hybridized on GeneChip Human Gene HTA 2.0 Affymetrix microarrays (Affymetrix UK Ltd.), by the Lyon University genomic facility ProfileXpert-LCMT (Lyon, France), were done according to the manufacturer's protocol. Data are available on the NCBI Gene Expression Omnibus website (http://www.ncbi.nlm.nih.gov/geo/): GEO dataset GSE148070. Details of the methods used are presented in the Online Supplementary Methods.

Histopathology and immunohistochemistry studies Samples fixed in 10% buffered formalin were processed for routine histopathological and immunohistochemical (IHC) examination. The IHC slides were digitalized using Panoramic 250 Flash II digital microscopes (3DHISTECH, Budapest, Hungary). IHC staining was evaluated both via manual scoring and with an automated method using image analysis software (see details in the Online Supplementary Methods).15,16

Mutation profile analysis with next-generation sequencing After DNA extraction of 51 available FL FFPE (n=33 from training cohort and n=18 from validation cohort), samples were sequenced on an Illumina MiSeqDx using our lymphopanel of 43 genes involved in B-cell lymphomagenesis.17 Sequencing and data analysis was performed as previously described17,18 (see details in the Online Supplemental Methods).

48 patients in the training cohort and 84 in the validation cohorts. The median age of the whole population was 61.8 years old (range, 28-87 years) (Online Supplementary Table S1). Most patients had advanced clinical stages (84%); 54% had high risk and 34% had intermediate risk FLIPI. FL histological grade was grade 1-2 in 91% of cases. The treatments were comparable between groups, with mostly rituximab plus CHOP or CHOP-like and rituximab maintenance (89%). With a median of follow-up of 43.4 months (interquartile range [IQR] 25.4-65.3 month), ten patients died from progressive disease and three from a second malignancy.

SUV at baseline correlates with the risk of progression in follicular lymphoma patients max

The median baseline SUV was 9.15 (range, 2.5-34.6; IQR 8.3-13.8) in our series of 132 patients (Online Supplementary Table S2). Baseline SUV was related neither to baseline TMTV (Pearson index= 0.35) nor to the largest lymph node size (Pearson index= 0.06). Baseline TMTV was related to the size of the largest mass assessed in 119 patients (Pearson index=0.76). It is worth noting that the SUV median of 12 FL grade 3A was not significantly different from that of FL grade 1-2. We determined that a SUV 14.5 was an accurate threshold (sensitivity of 0.95, specificity of 0.16, false positive rate of 0.59, and precision of 0.85) that was able to distinguish patients with different outcomes. Only 14% of patients (n=19) had SUV >14.5. Overall PFS was significantly lower in patients with SUV >14.5 than in those with SUV ≤14.5 (HR= 0.28; P=0.00046), and 2-year PFS was 54% versus 86% (P=0.006). (Figure 1A, left and 1B). On univariate analysis, factors associated with PFS were SUV (P=0.0016), FLIPI (P=0.048), elevated lactate dehydrogenase (LDH) (P=0.022), and elevated B2 microglobulin (P=0.038). The type of treatment regimen (R-chemotherapy vs. antiCD20-lenalidomide) and baseline TMTV had no effect on PFS (Figure 1A, right). In a multivariate Cox model (SUV , FLIPI, LDH, B2 microglobulin), SUV was the only factor retaining an independent prognostic value for PFS (HR=0.25; P=0.0066). Twelve patients (25%) experienced disease progression within 24 months after the start of therapy. POD24 events were more common in the subset of patients with SUV >14.5 (55% of patients with SUV >14.5) than in those with SUV ≤14.5 (15% of patients with SUV ≤14.5). Median OS was 23.9 months for patients with SUV >14.5 and not reached for patients with low SUV (HR 0.37; P=0.06). max

max

Statistics analysis PFS was defined as the time from the first cycle of immunochemotherapy to progression or death (event) or last follow-up/change of treatment (censored data). POD24 was defined as primary-refractory disease (less than partial response), progression, transformation or relapse within 24 months after diagnosis. OS was defined as the time from the start of therapy to death or last follow-up. All survival rates were estimated using the KaplanMeier method, and log-rank test were assessed using R software. Median follow-up was calculated with the Kaplan-Meier reverse method.19 Optimal cutoff to predict PFS was determined using the R ‘Survival’ package20,21 and log-rank test-based P-values. Indeed, the multiple tests performed by this package were corrected using the Benjamini-Hochberg method22 to control the false discovery rate (FDR) and determine the optimal threshold. For comparisons between groups, a normality test followed by a Wilcoxon test or Student’s t-test were performed using R software. The compound linear and nonlinear relationship between Ki-67 percentage and SUV level was automatically examined by computing the maximal information coefficient (MIC) with an algorithm running the MINE method.23 max

Results Study population A total of 132 patients were included in the study, with haematologica | 2022; 107(1)

max

Baseline SUV does not correlate with immune infiltration max

In order to investigate whether the immune infiltration signature in baseline biopsies was associated with an increased 18FDG uptake, we compared patients with SUV >14.5 and SUV ≤14.5, and the immune profile determined by immunochemistry (IHC) and transcriptome approaches. We first measured the sample enrichment score (SES) of both ‘T-cell activation’ and ‘IEGS33’ gene sets in each transcriptome from our FL training cohort and from 1,446 NHL and normal lymphoid tissue downloaded from public cohorts available in GEO data set.8,19 The 38 available frozen FL samples from our training cohort and 148 of the 1,446 NHL samples had a high score max

max

223

C.Rossi et al.

Figure 1 Clinical impact of SUV and total metabolic tumor volume at positron emission tomography baseline in the follicular lymphoma patients. (A) Distribution of whole body maximum standardized uptake (SUV ) (left) and total metabolic tumor volume (TMTV) (right) in 132 follicular lymphoma (FL) patients without progression/relapse or with progression/relapse 24 months after starting treatment. (B) Progression-free survival (PFS) according to SUV threshold of 14.5 since the start of treatment. Kaplan-Meier estimates of PFS in the whole cohort (n=132). An optimal threshold was set to separate patients into high-risk (14% of FL patients; n=19 of 132) and low-risk (86%, n=113 of 132) groups for progression/relapse (P=0.00046). Hi: high SUV ; HR: hazard ratio, Lo: low SUV . max

max

for IEGS33, indicating up-regulation of immune escape genes. The score for T-cell activation signature in these FL samples was also among the highest. Finally, scatterplots of all samples for ‘IEGS33’ versus ‘T-cell activation’ showed that the FL samples clustered together, in contrast with most other NHL (Figure 2). These findings were confirmed upon datamining of another 160 FL samples from the PRIMA study21(Online Supplementary Figure S1). In order to further characterize the IE status of FL samples at the protein level, FFPE samples were stained for ICP markers and were then scored. PD-1 expression in tumorinfiltrating lymphocytes (TIL) showed perifollicular, intrafollicular and diffuse patterns in 46%, 20% and 34% of cases, respectively. PD-1+ cells and PD-L1+ cells represented, respectively, 16% (range, 2-30%) and 5% (range, 210%) of CD3+ immune cells (Figure 3A). LAG3 staining was seen in less than 5% and TIM3 staining was seen in 15% of total CD3+ immune cells (ranging from 2% to 5% and from 5% to 15%, respectively). These results were consistent with those from transcriptomic analyses. All ICP protein expressions scored by IHC in each sample were strongly correlated with their respective IEGS33 scores (Figure 3B). We then analyzed the correlation between baseline PET and molecular scores for IEGS and T-cell activation signa224

max

Figure 2. Functional immune status of follicular lymphoma samples from the training cohort and previously published lymphoma cohorts. Dot plot of sample enrichment scores (SES) for the immune escape gene set (IEGS33) versus T-cell activation gene set (defined in 24) in 38 frozen follicular lymphoma (FL) samples from our training cohort (purple dots) and 148 FL (red dots) among 1,446 non Hodgkin lymphoma (NHL) (grey dots) public microarrays datamining analysis.

haematologica | 2022; 107(1)

Baseline SUV in FL and tumor proliferation max

tures in biopsies. However, no significant correlation emerged between immune escape/T-cell activation scores and baseline SUV or PFS (Online Supplementary Figure S2). In order to further assess the patterns of leukocyte infiltrates in the tumor samples, we performed algorithmic deconvolution by CIBERSORT to infer the proportion of 14 leukocyte and other non-hematopoietic cell types from each sample. At the same time, we assessed the immune cell composition of FL samples by IHC. Deconvolution of max

bulk tumor transcriptomes unveiled a composition containing more CD8+ T lymphocytes and macrophages and fewer NK cells and gd T-cell cytolytic lymphocytes (Online Supplementary Figure S3A). Accordingly, the IHC analyses evidenced median percentages of CD3+T cells, CD8+ T cells and CD163+ monocytes of 35% (range, 5-60%), 14% (range, 1-40%), and 11% (range, 1-30%), respectively, of the total immune cell infiltrate (Figure 4A). The abundance of CD8+ T cells scored by IHC was high (>30% of immune

Figure 3. Immunohistochemical validation of immune escape gene set (IEGS33) overexpression in follicular lymphoma samples. (A) Upper panel shows a representative case of PD-1 staining with diffuse (left) intrafollicular (right) patterns (magnification: 100x and 50x, respectively and inserts: 200x). Medium panel shows representative cases of PD-L1 (left) and LAG3 (right) staining (magnification: 100x and inserts: 200x). Lower panel shows a representative case of TIM3 staining (left) (magnification: 100x and inserts: 200x) and immunohistochemical (IHC) quantification of immune checkpoint (ICP): PD-1, PD-L1, LAG3 and TIM3 staining (right). (B) The heat map (left) represents IHC scoring of the four ICP markers and the graph (right) shows the correlation between the percentage of ICP-positive immune cells scored by IHC and the sample enrichment scores (SES) for immune escape gene sets (IEGS) in each FL sample. Each sample is shown by a dot. FL: follicular lymphoma.

haematologica | 2022; 107(1)

225

C.Rossi et al.

cells) in 20% of cases, low (<10%) in 40% cases, and intermediate (10-30%) in 40% of cases as illustrated in Figure 4B. The abundance of CD8+ T cells in FL correlated in the deconvolution and IHC results (Figure 4C). They also matched with T-cell activation scores, suggesting that transcriptomic evidence of T-cell activation in FL samples was consistent with and related to the cytotoxic CD8+ T cells observed by IHC (Figure 4C). By contrast however, the deconvoluted immune cell composition did not differ significantly according to SUV (Online Supplementary Figure S3B). max

DNA repair/tumor proliferation signatures and SUV

max

We compared the tumor cell DNA repair/proliferation by transcriptomic and IHC approaches to 18FDG uptake signatures obtained. In the 38 available frozen FL samples, using the Gene Ontology (GO)22 and Reactome,27 we scored the SES for gene signatures of tumor proliferation, such as GO cell cycle DNA, G2M DNA replication and base excision repair (BER), a pathway that increases with proliferative activity.24 We observed a relationship between

the level of SUV and DNA repair/proliferation signature scores (Figure 5A; Online Supplementary Figure S4). We then validated these findings at the protein level by scoring Ki-67 staining in FFPE samples. The Ki-67 staining was scored according to the percentage of Ki-67+ tumor cells determined by optical evaluation and quantified by automated image analysis solutions (Figure 5B). Using a cutoff value of 10%, Ki-67 immunostaining was found to be significantly associated with GO cell cycle DNA replication SES (P=0.013), BER SES (P=0.0086) and G2M checkpoint SES (P=0.0059) (Figure 5C). We then analyzed the correlation of PET-CT baseline and Ki-67 scoring. Using the MINE method23 we found a non-correlated association between the quantitative variable of the SUV value and Ki-67 percentage estimated by automated quantification for the training cohort (MIC=0.59) and for the validation cohort (MIC=0.35). In addition, using the optimal Ki-67 cut-off of 10%, we found that the SUV level was significantly increased in FL grade 1-2 and 3A samples with Ki-67 staining ≥10% (Wilcoxon P=0.007 in the training cohort and P=0.006 in the validation cohort) (Figure 5D). Finally,

max

Figure 4. Immunohistochemical validation of sample enrichment score for T-cell activation in follicular lymphoma samples. (A) Immunohistochemical (IHC) quantification of CD3, CD8 and CD163 staining in follicular lymphoma (FL) samples. (B) Representative cases of CD8 staining categorized according to the percentage of CD8+ T cells among the total immune cells (5-10%, 10-30%, and >30%). (C) Box plots of correlations between the percentage of CD8+ T cells scoring by IHC with CD8 abundance quantified by deconvolution algorithm24 (left) and with sample enrichment score (SES) for immune cytotoxic activity (right) .

226

haematologica | 2022; 107(1)

Baseline SUV in FL and tumor proliferation max

this SUV -to-Ki-67 index association suggests that 18FDG uptake in FL patients at baseline reflects tumor cell proliferation. max

Baseline SUV could be influenced by tumor cell mutation profiles max

In the 33 available FL samples from the training cohort, targeted next-generation sequencing (NGS) identified 243 non-synonymous alterations in 32 genes (Figure 6A; Online Supplementary Table S3). The most frequently mutated genes were KMT2D, CREBBP and BCL2 (22%, 14%, and 15% of total variants, respectively), which were detected in 82%, 73% and 58% of cases, respectively. A recurrent missense mutation in EZH2 tyrosine 646 (Y646) was also found in 21% of cases. Altogether, 94% of patients harbored mutations involved in genes from epigenetic regula-

tion pathways (48.5% of total variants), whereas mutations in apoptotic pathways (16% of total variants) or in genes involved in immune response (21% of total variants) were less frequent (64% and 70% of patients) (Figure 6B). The mutational frequency in these altered pathways did not vary with the level of SUV (Figure 6C). However, FOXO1 mutations (n=2 of 33) were exclusively detected in patients with high SUV (SUV >14.5), while MFHAS1 (n=2), MYC (n=1), and TP53 (n=6) mutations were only observed in patients with low SUV (SUV <6.5) (Online Supplementary Figure S5). The same analyses in the validation cohort (n=18 available FL samples) unveiled 132 non-synonymous alterations in 32 genes. The most frequently mutated genes were CREBBP, KMT2D and MEF2B (18%, 13%, and 6% of total variants, respectively, occurring in 78%, 66%, and 44% of max

max

Figure 5. Correlation between tumor proliferation signatures and SUV in follicular lymphoma samples. (A) The heat map represents quantification of whole body maximum standardized uptake (SUV ) and five sample enrichment score (SES) gene sets for proliferative index and DNA repair/tumor proliferation signatures (Gene Ontology [GO] cell cycle DNA replication, G2M DNA replication checkpoint, and base excision repair [BER]). Each column represents one patient. (B) Example of quantification of Ki-67 staining on two representative cases. Panels on the left show the original picture of Ki-67 staining (upper panel: Ki-67 <10%; original magnification 100x; scale bar =50 µm; lower panel: Ki-67 ≥10%; original magnification 100x; scale bar = 50mm) and panels on the right show the corresponding computerized image analysis of Ki67 staining (Ki-67 positive cells are purple and uncolored cells are Ki-67 negative). (C) Box plots of correlations between SES for GO cell cycle DNA replication (left), G2M DNA replication checkpoint (middle) or DNA repair gene sets (BER) (right) with the percentage of Ki-67 proliferative index by immunochemistry. (D) Correlation between SUV level and Ki-67 staining in follicular lymphoma samples from the training and validation cohorts. max

max

haematologica | 2022; 107(1)

227

C.Rossi et al.

patients). Here again, alterations of epigenetic modifiers were the most frequent (41.7% of total variants occurred in 89% of patients) (Figure 6; Online Supplementary Table S3). Altogether, FOXO1 mutations occurred in five FL patients (two from the training cohort and three from the validation cohort), four of which had SUV >14.5. max

Moreover, the lack of association between tumor immunological contexture and prognosis could be related to anti-CD20 maintenance therapy, as already suggested.32 While baseline SUV appears unrelated to immune T-cell infiltration, it remains related to tumor cell proliferation, as shown by high DNA repair and Ki-67 proliferative indexes. Thus, these results suggest that SUV >14.5 at baseline in FL patients with Ki-67 staining ≥10% could define patients with high risk of relapse or progression. While TMTV, a surrogate marker of tumor cell burden, was unrelated to baseline SUV or patient outcomes in our series, TMTV >510 cm3 has been associated with a higher risk of treatment failure in FL patients.6 However, this discrepancy could be related to the absence of maintenance treatment in the study by Meignan et al.6 Considering that the long-term results of the PRIMA study showed improved PFS in FL patients receiving rituximab maintenance (10.5 years compared to 4.1 years without maintenance treatment),33 and the GALLIUM study34 showed an additional benefit of obinutuzumab over rituximab for PFS, anti-CD20 maintenance could mitigate the prognostic value of TMTV at baseline. Accordingly, TMTV had no prognostic value in either arm (rituximab or obinutuzumab) of the GALLIUM study. This suggests that FL prognosis could be more related to the proportion of proliferating tumor cells than to the cell burden of the tumor. Ultimately, without evidence of transformation, high SUV in FL samples may be associated with more aggressive tumor cells. This has been observed in other low-grade lymphomas,35 suggesting that the ability of FL tumors to proliferate may affect their avidity to FDG and thus the PET signals observed in FL patients. The frequency of alterations in the oncogenic pathway did not vary with the SUV level, but we identified a FOXO1 mutation in a subset of patients with SUV >14.5. The prognostic value of FOXO1 mutations has already been shown in DLBCL36 and FL.12,37 For instance, FOXO1 mutations were more frequently observed in high-risk FL patients with POD24 when compared with the nonPOD24 subgroup (25% vs. 10%, P=0.04)12,37 and were associated with shorter failure-free survival (HR 2.74, range, 1.23-6.09; P=0.013) in the m7-FLIPI cohort.12 Moreover, the FOXO1 transcription factor appears to be implicated in the regulation of the abundance of CD20 on the surface of tumor cells, so alterations could influence the response to rituximab-based therapies as previously reported in an in vitro study of cell line assays.38 On the contrary, we found certain mutations such as TP53 and MYC more often in patients with SUV <14.5. However, these observations could be related either to the spatial genetic heterogeneity of FL39 or to possible areas of FL transformation.39 In this context, high SUV levels could be very helpful to identify those FL sites with suspected transformation. That is why we selected lymph node biopsies with the most FDG-avid sites to ensure lack of transformation. However, single biopsy specimens cannot rule out transformation/genomic alteration or variation in other sites. Finally, baseline SUV outperformed FLIPI to predict patient outcomes, in addition to multibiopsy site assessment of gene mutation profile; it could therefore possibly be used to identify patients who are at risk of poor outcomes. In conclusion, SUV >14.5 at baseline FDG-PET is associated with poor PFS and high risk of POD24 events in FL max

max

Discussion Our multi-parametric analysis of two independent cohorts of FL patients shows that baseline SUV values above 14.5 identified those with early unfavorable outcomes, defined here as poorer PFS and higher rate of POD24 events. Although this threshold is an absolute value which needs to be confirmed in larger cohorts, our study suggests that the overall high level of SUV at diagnosis could potentially reflect specific biological behaviors in FL tumors. Our results build on those recently obtained by Strati et al.8, who demonstrated the prognostic impact of baseline SUV >18 in FL patients. Their use of a higher threshold could be explained by i) the more favorable dissemination of disease in the FL patients from our study (17% vs. 0% of Ann Arbor stage I-II)8, ii) the use of R-CHOP or other R-chemotherapy as induction treatment in our cohorts, while 15% of patients in Strati et al.8 received rituximab alone; iii) the fact that all of our patients received maintenance therapy while only few patients received Rm in Strati et al.,8 and iv) PET acquisition and measures were performed on different PET/CT devices. These discrepancies suggest a better chance of disease control in our series, with a lower risk of relapse and fewer PFS events.25,28 Moreover, Strati et al.8 demonstrated a significant association between the largest lymph node ≥6 cm and SUV >18. In our study, we focused on TMTV to assess the tumor burden seeing as available computed tomography was not mandatory for patient enrollment in the study. Nevertheless, for 119 (91%) patients with available computed tomography, the size of the largest lymph node was strongly related to TMTV but not to SUV . We found no correlation between the tumor burden and SUV in our series. Pre-treatment PET staging is more sensitive than CT for identifying FL patients who have a high risk of transformation to aggressive lymphoma.29 In line with our results, several other studies have suggested that SUV threshold values ranging from 10 to 13 could be used to identify FL with a higher risk of transformation.4,30,31 However, only tumor biopsy possibly guided by PET can rule out FL transformation into high grade lymphoma, which is de facto associated with reduced PFS and OS. Here, SUV >14.5 did not necessarily indicate that FL would develop or transform into aggressive lymphoma. The biopsy sites used for FL diagnosis exhibited the higher SUV at PET baseline, while their histological analyses did not unveil any transformation. In our series, high avidity of FDG in FL samples was not associated with a specific infiltration pattern of major immune cell populations and was not correlated with increased of SES for IEGS33 and for T-cell activation gene set. Nevertheless, some FL samples lacked available frozen tissue for RNA sequencing and were unsuitable for further analyses. Our findings should thus be confirmed in a larger prospective cohort with sufficient biopsied material from the most FDG-avid FL tumor. max

max

228

max

haematologica | 2022; 107(1)

Baseline SUV in FL and tumor proliferation max

Figure 6. Molecular profiling of follicular lymphoma samples. (A) Heatmap of the most significantly mutated genes in follicular lymphoma (FL) samples from the training cohort (one column by sample, one line by gene). The colors refer to the type of mutation. (B) Circos plot illustrating the functional pathways involved in FL genomic abnormalities in the training (upper panel, n=33) and validation (lower panel, n=18) cohorts. The percentages refer to the frequency of alterations in the respective pathways. (C) Histogram showing the frequency of impaired functional pathways according to whole body maximum standardized uptake (SUV ) levels. max

patients. We also found an association between high SUV levels and tumor cell proliferation, but not with the cell content in the tumor microenvironment. Proliferative tumor cells are enriched in FOXO1 mutations, which could explain their resistance to anti-CD20 therapy and subsequent poorer patient outcomes. The integrative approach used here could be applied in a multi-site lymph node analysis to further clarify the biological heterogeneity of FL and its relationship with functional imaging patterns. max

Disclosures CR has received a research grant from Roche and personal fees as well as non-financial support from Janssen, Roche, Takeda; ROC has received research grant from Gilead and Takeda and personal fees and non-financial support from Janssen, Roche, Takeda, Merck/BMS, Abbvie and Amgen. The other authors declare no competing interests. Contributions CR, PG, CL and CB designed the experimental strategy,

haematologica | 2022; 107(1)

organized the experiments and collected and analyzed the data; CR, PG, PB, LY, LO and CG provided clinical data for the patients included in the series; SK performed analyses of PETFDG; TF and JG performed statistical analyses; MT generated SES and performed analysis by data mining with CR, CB, JJF and CL; CL selected FL biopsies, performed and analyzed IHC and t-NGS experiments helped by PG, SP, SE, SR, LM, and CS; ROC, NA and DMF provided experimental advice; CR, JJF, CL and CB wrote the manuscript with advice from AS, CK, CL, MT, LY, PG. All authors discussed and approved the manuscript. Acknowledgements The authors would like to thank Nathalie Van-Acker (immunochemistry staining, Dept. of Pathology, IUCT-O, Imag’in Plateform, Toulouse), Frédéric Escudié and David Grand (NGS analysis, Dept. of Pathology, IUCT-O, Toulouse, Karin Gordien (Registre des Cancers du Tarn, Albi), Sophie Péries (CRB CHU Toulouse). We also thank Suzanne Rankin from the Dijon-Bourgogne University Hospital for proofreading the manuscript.

229

C.Rossi et al.

Funding This work was supported in part by institutional grants from INSERM, Université Paul Sabatier and CNRS, Institut Claudius Regaud CLCC (contract R20027BB, CIEL) the Laboratoire

References 1. Sarkozy C, Maurer MJ, Link BK, et al. Cause of death in follicular lymphoma in the first decade of the rituximab era: a pooled analysis of French and US cohorts. J Clin Oncol. 2019;37(2):144-152. 2. Dada R. Diagnosis and management of follicular lymphoma: a comprehensive review. Eur J Haematol 2019;103(3):152-163. 3. Casulo C, Byrtek M, Dawson KL, et al. Early relapse of follicular lymphoma after rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone defines patients at high risk for death: an Analysis from the National LymphoCare Study. J Clin Oncol. 2015;33(23):2516-2522. 4. Bodet-Milin C, Kraeber-Bodéré F, Moreau P, Campion L, Dupas B, Le Gouill S. Investigation of FDG-PET/CT imaging to guide biopsies in the detection of histological transformation of indolent lymphoma. Haematologica. 2008;93(3):471-472. 5. Trotman J, Barrington SF, Belada D, et al. Prognostic value of end-of-induction PET response after first-line immunochemotherapy for follicular lymphoma (GALLIUM): secondary analysis of a randomised, phase 3 trial. Lancet Oncol. 2018;19(11):1530-1542. 6. Meignan M, Cottereau AS, Versari A, et al. Baseline metabolic tumor volume predicts outcome in high-tumor-burden follicular lymphoma: a pooled analysis of three multicenter studies. J Clin Oncol. 2016;34(30): 3618-3626. 7. Bailly C, Carlier T, Berriolo-Riedinger A, et al. Prognostic value of FDG-PET in patients with mantle cell lymphoma: results from the LyMa-PET Project. Haematologica. 2020;105 (1):e33-e36. 8. Strati P, Ahmed MAA, Fowler N, et al. Pretreatment maximum standardized uptake value predicts outcome after frontline therapy in patients with advanced stage follicular lymphoma. Haematologica. 2020;105(7): 1907-1913. 9. Dave SS, Wright G, Tan B, et al. Prediction of survival in follicular lymphoma based on molecular features of tumor-infiltrating immune cells. N Engl J Med. 2004;351 (21):2159-2169. 10. Laurent C, Charmpi K, Gravelle P, et al. Several immune escape patterns in nonHodgkin’s lymphomas. Oncoimmunology. 2015;4(8):e1026530. 11. Huet S, Tesson B, Jais J-P, et al. A geneexpression profiling score for prediction of outcome in patients with follicular lymphoma: a retrospective training and validation analysis in three international cohorts. Lancet Oncol. 2018;19(4):549-561. 12. Pastore A, Jurinovic V, Kridel R, et al. Integration of gene mutations in risk prognostication for patients receiving first-line immunochemotherapy for follicular lymphoma: a retrospective analysis of a prospective clinical trial and validation in a population-based registry. Lancet Oncol. 2015;16(9):1111-1122.

230

d'Excellence Toulouse Cancer (TOUCAN) (contract ANR11LABX) and the Programme Hospitalo-Universitaire en Cancérologie CAPTOR (contract ANR11-PHUC0001). CR obtained a grant from ITMO Cancer Plan Cancer 2014-2019.

13. Swerdlow SH, Campo E, Pileri SA, et al. The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood. 2016;127(20):2375-2390. 14. Kanoun S, Tal I, Berriolo-Riedinger A, et al. Influence of software tool and methodological aspects of total metabolic tumor volume calculation on baseline [18F]FDG PET to predict survival in Hodgkin lymphoma. PLoS One. 2015;10(10):e0140830. 15. Laurent C, Guérin M, Frenois F-X, et al. Whole-slide imaging is a robust alternative to traditional fluorescent microscopy for fluorescence in situ hybridization imaging using break-apart DNA probes. Hum Pathol. 2013;44(8):1544-1555. 16. Péricart S, Tosolini M, Gravelle P, et al. Profiling immune escape in Hodgkin’s and diffuse large B-cell lymphomas using the transcriptome and immunostaining. Cancers. 2018;10(11):415. 17. Evrard SM, Péricart S, Grand D, et al. Targeted next generation sequencing reveals high mutation frequency of CREBBP, BCL2 and KMT2D in high-grade B-cell lymphoma with MYC and BCL2 and/or BCL6 rearrangements. Haematologica. 2019;104 (4):e154-e157. 18. Vendrell JA, Grand D, Rouquette I, et al. High-throughput detection of clinically targetable alterations using next-generation sequencing. Oncotarget. 2017;8(25):4034540358. 19. Shuster JJ. Median follow-up in clinical trials. J Clin Oncol. 1991;9(1):191-192. 20. Therneau T. 2020. A package for survival analysis in R. R package version 3.2-3 https://CRAN.R-project.org/package=survival. 21. Terry M. Therneau, Patricia M. Grambsch. Modeling survival data: extending the Cox ModelA package for survival analysis in R. R package version 3.2-3, at https://CRAN.Rproject.org/package=survival. New York. Springer. 2000. 22. Benjamini, Y, Hochberg, Y. Controlling the false discovery rate - a practical and powerful approach to multiple testing. J R Stat Soc Ser B Methodol. 1995;57:289-300. 23. Reshef DN, Reshef YA, Finucane HK, et al. Detecting novel associations in large data sets. Science. 2011;334(6062):1518-1524. 24. Tosolini M, Algans C, Pont F, Ycart B, Fournié J-J. Large-scale microarray profiling reveals four stages of immune escape in nonHodgkin lymphomas. Oncoimmunology. 2016;5(7):e1188246. 25. Salles G, Seymour JF, Offner F, et al. Rituximab maintenance for 2 years in patients with high tumour burden follicular lymphoma responding to rituximab plus chemotherapy (PRIMA): a phase 3, randomised controlled trial. Lancet. 2011;377(9759):42-51. 26. Ashburner M, Ball CA, Blake JA, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet. 2000;25(1):25-29. 27. Fabregat A, Sidiropoulos K, Garapati P, et al.

The Reactome pathway Knowledgebase. Nucleic Acids Res. 2016;44(D1):D481-487. 28. Rummel MJ, Al-Batran SE, Kim S-Z, et al. Bendamustine plus rituximab is effective and has a favorable toxicity profile in the treatment of mantle cell and low-grade nonHodgkin’s lymphoma. J Clin Oncol. 2005;23(15):3383-3389. 29. Batlevi CL, Sha F, Alperovich A, et al. Positron-emission tomography-based staging reduces the prognostic impact of early disease progression in patients with follicular lymphoma. Eur J Cancer. 2020;126:78-90. 30. Alobthani G, Romanov V, Isohashi K, et al. Value of 18F-FDG PET/CT in discrimination between indolent and aggressive nonHodgkin’s lymphoma: A study of 328 patients. Hell J Nucl Med. 2018;21(1):7-14. 31. Schöder H, Noy A, Gönen M, et al. Intensity of 18 Fluorodeoxyglucose uptake in positron emission tomography distinguishes between indolent and aggressive nonHodgkin’s lymphoma. J Clin Oncol. 2005;23 (21):4643-4651. 32. Xerri L, Huet S, Venstrom JM, et al. Rituximab treatment circumvents the prognostic impact of tumor-infiltrating T-cells in follicular lymphoma patients. Hum Pathol. 2017;64:128-136. 33. Bachy E, Seymour JF, Feugier P, et al. Sustained progression-free survival benefit of rituximab maintenance in patients with follicular lymphoma: long-term results of the PRIMA Study. J Clin Oncol. 2019;37 (31):2815-2824. 34. Barrington SF, Trotman J, Sahin D, et al. Baseline PET-derived metabolic tumor volume metrics did not predict outcomes in follicular lymphoma patients treated with firstline immunochemotherapy and antibody maintenance in the phase III GALLIUM study. Blood. 2018;132(Suppl 1):S2882-2882. 35. Falchi L, Ferrajoli A, Jacobs I, Nava-Parada P. An evidence-based review of anti-CD20 antibody-containing regimens for the treatment of patients with relapsed or refractory chronic lymphocytic leukemia, diffuse large B-cell lymphoma, or follicular lymphoma. Clin Lymphoma Myeloma Leuk. 2018;18 (8):508-518. 36. Rushton CK, Arthur SE, Alcaide M, et al. Genetic and evolutionary patterns of treatment resistance in relapsed B-cell lymphoma. Blood Adv. 2020;4(13):2886-2898. 37. Jurinovic V, Kridel R, Staiger AM, et al. Clinicogenetic risk models predict early progression of follicular lymphoma after firstline immunochemotherapy. Blood. 2016;128(8):1112-1120. 38. Pyrzynska B, Dwojak M, Zerrouqi A, et al. FOXO1 promotes resistance of nonHodgkin lymphomas to anti-CD20-based therapy. Oncoimmunology. 2018;7(5): e1423183. 39. Araf S, Wang J, Korfi K, et al. Genomic profiling reveals spatial intra-tumor heterogeneity in follicular lymphoma. Leukemia. 2018;32(5):1261-1265.

haematologica | 2022; 107(1)

ARTICLE

Plasma Cell Disorders

Autologous stem cell transplantation is safe and effective for fit, older myeloma patients: exploratory results from the Myeloma XI trial

Ferrata Storti Foundation

Charlotte Pawlyn,1,2 David A. Cairns,3 Tom Menzies,3 John R. Jones,4 Matthew W. Jenner,5 Gordon Cook,3,6 Kevin D. Boyd,2 Mark T. Drayson,7 Martin F. Kaiser,1,2 Roger G. Owen,8 Walter Gregory,3 Gareth J. Morgan,9 Graham H. Jackson10 and Faith E. Davies9 on behalf of the UK NCRI Haemato-Oncology Clinical Studies Group. The Institute of Cancer Research, London, UK; 2The Royal Marsden Hospital, London, UK; Clinical Trials Research Unit, Leeds Institute of Clinical Trials Research, University of Leeds, Leeds, UK; 4King’s Hospital NHS Foundation Trust, London, UK; 5University Hospital Southampton NHS Foundation Trust, Southampton, UK; 6Leeds Cancer Centre, Leeds Teaching Hospitals Trust, Leeds, UK; 7Institute of Immunology and Immunotherapy, University of Birmingham, Birmingham, UK; 8HMDS, Leeds Cancer Centre, Leeds Teaching Hospitals Trust, Leeds, UK; 9Perlmutter Cancer Center, NYU Langone, New York, NY, USA and 10Department of Haematology, Newcastle University, Newcastle, UK 1 3

Haematologica 2022 Volume 107(1):231-242

ABSTRACT

utologous stem cell transplant (ASCT) remains the standard of care for consolidation after induction therapy for eligible patients with newly diagnosed myeloma. In recent clinical trials comparing ASCT to delayed ASCT, patients aged over 65 were excluded. In real-world practice stem cell transplants are not restricted to those aged under 65 and clinicians decide on transplant eligibility based on a patient’s fitness rather than a strict age cut-off. Data from the UK NCRI Myeloma XI trial, a large phase III randomized controlled trial with pathways for transplant-eligible and -ineligible patients, were used in an exploratory analysis to examine the efficacy and toxicity of ASCT in older patients including an analysis using an age-matched population to compare outcomes for patients receiving similar induction therapy with or without ASCT. Older patients within the transplant-eligible pathway were less likely to undergo stem cell harvest at the end of induction than younger patients and of those patients undergoing ASCT there was a reduction in progression-free survival associated with increasing age. ASCT in older patients was well tolerated with no difference in morbidity or mortality between patients aged <65, 65-69 and 70-75 years. In an age-matched population of patients including those in both the transplant-eligible and -ineligible pathways there was a significant advantage associated with undergoing ASCT with increases in progression-free survival (hazard ratio 0.41, P<0.0001) and overall survival (hazard ratio 0.51, P<0.0001), which persisted even after adjustment for baseline covariates including those related to frailty and response to induction. These findings support the use of ASCT for selected fit, older myeloma patients. EudraCT number, 2009-010956-93

Correspondence: CHARLOTTE PAWLYN charlotte.pawlyn@icr.ac.uk Received: June 7, 2020. Accepted: November 25, 2020. Pre-published: December 3, 2020. https://doi.org/10.3324/haematol.2020.262360

Introduction Autologous stem cell transplant (ASCT) is delivered as consolidation after induction therapy for eligible patients with newly diagnosed myeloma. The use of ASCT became standard of care based on several randomized controlled trials that demonstrated progression-free (PFS) and overall survival (OS) benefits.1-4 The ongoing use of ASCT in the context of current induction treatment regimens continues to be supported by data from two recent large phase III studies.5,6 Both these studies, however, excluded patients aged over 65 years old. In real-world practice stem cell transplants are not restricted to those aged under 65 and clinicians decide on transplant eligibility

haematologica | 2022; 107(1)

231

C. Pawlyn et al.

based on the individual patient’s fitness rather than a strict age cut-off. Standard-of-care conditioning for ASCT consists of melphalan given at a dose of 200 mg/m2 although a lower dose of 140 mg/m2 may be delivered in the case of renal impairment and is sometimes considered by clinicians for the treatment of older patients. The European Bone Marrow Transplant (EBMT) registry has documented an increase in the number of patients aged over 65 undergoing ASCT in recent years. Between 20012005, 2,478 patients aged 65-69 years underwent ASCT, constituting 14.1% of transplanted patients, and the number rose to 3,860 (15.8%) in the years 2006-2010. A similar pattern was seen for those aged 70 or over: 497 (2.8%) in 2001-2005 and 740 (3%) in 2006-2010.7 In this analysis there was no apparent difference in transplant-related mortality between those aged 60-64, 65-69 and ≥70 years, with the rates being 1.8%, 2.1% and 2.4%, respectively. This trend is mirrored in data from the USA-based Center for International Blood and Marrow Transplant Research (CIBMTR).8 Two randomized studies of ASCT in older patients were conducted using dose-reduced melphalan (100 mg/m2) tandem transplantation following conventional chemotherapy induction regimens. The first study randomized patients aged 50-70 years between melphalan prednisolone (MP) and vincristine, doxorubicin and dexamethasone for two cycles followed by dose-reduced melphalan and ASCT for two cycles (VAD+ASCT100).9 The use of ASCT was associated with improved event-free survival and OS. The second study randomized patients between MP, MP plus thalidomide and VAD+ASCT100. This study demonstrated an improvement in OS for VAD+ASCT100 versus MP but the use of MP plus thalidomide was superior to both approaches.10 These data supported the use of ASCT100 in the context of conventional chemotherapy induction in older patients but the combination of both immunomodulatory agent induction and ASCT was not examined. Several previous retrospective studies have examined outcomes following ASCT for patients over the age of 65 or 70 following immunomodulatory and/or proteasome inhibitor-based induction. A large retrospective study of patients treated at the Mayo Clinic (USA) compared 207 patients aged 70 and over to 1,765 patients aged less than 70.11 There was no significant difference in PFS, OS or transplant-related mortality between the groups. A similar analysis of patients treated in Heidelberg (Germany), found no difference between outcomes for those aged 60-64, 6569 or 70-75 years.12 Retrospective data analysis has also been used to compare ASCT and no-ASCT treatment strategies in small cohorts of patients over the age of 65.13,14 These studies support the use of ASCT in patients over the age of 65 thought to be fit, but did not address whether ASCT is preferred over conventional therapy for patients in this older age group. To our knowledge, no randomized comparison of ASCT to no-ASCT has been undertaken in patients over the age of 65 in the current treatment landscape. Data from the UK National Cancer Research Institute (NCRI) Myeloma XI trial, a large phase III randomized controlled trial with pathways for transplant-eligible (TE) and ineligible (TNE) patients, was used to explore the efficacy and toxicity of ASCT in older patients including an analysis using an age-matched population.15,16 Patients in the trial were randomized between induction treatment with thalidomide- or lenalidomide-based triplets, with the same 232

combinations being used in both the TE and TNE pathways. This gives the opportunity to examine outcomes for TE patients of different ages, but also to compare outcomes for similar patients receiving the same induction therapy with or without ASCT.

Methods Myeloma XI is a phase III, open-label, parallel-group, multi-arm, adaptive trial and recruited newly diagnosed patients of all ages. Eligible patients were aged ≥18 years. The trial was designed to reflect a population as close to real-world as was considered safe. Exclusion criteria were therefore limited, but included previous treatment for myeloma (excluding local radiotherapy, bisphosphonates, and corticosteroids), previous or concurrent malignancies (including myelodysplastic syndromes), grade ≥2 peripheral neuropathy, acute renal failure (unresponsive to up to 72 h of rehydration, characterized by creatinine >500 µmol/L or urine output <400 mL/day or requiring dialysis), and active or prior hepatitis C virus infection. There were separate pathways for TE and TNE patients. The trial was performed in accordance with the Declaration of Helsinki 1996, and the study was approved by the national ethics review board (National Research Ethics Service, London, UK), institutional review boards of the participating centers, and the competent regulatory authority (Medicines and Healthcare Products Regulatory Agency, London, UK). All patients provided written informed consent. The trial was registered with the EU Clinical Trials Register (EudraCT number, 2009-010956-93). The details of the trial therapy and most primary outcomes have been published already.15,16 In brief, patients in both pathways were randomized between a thalidomide-containing triplet (cyclophosphamide, thalidomide and dexamethasone) or a lenalidomide-containing triplet (cyclophosphamide, lenalidomide and dexamethasone). Induction treatment was given for a minimum of four cycles (in the TE pathway) or six cycles (TNE) and to maximum response, and there was an induction intensification question for those with a suboptimal response to initial induction. All TE patients were planned to undergo an ASCT. Patients in both pathways underwent maintenance randomization between lenalidomide (± vorinostat) and observation. The choice of pathway, TE or TNE, was left to the local investigator based on co-morbidities and the patient’s/clinician’s preference. There was no age limit for entry into the TE pathway. Induction therapy in the TNE pathway was administered with an attenuation of dexamethasone dosing (Online Supplementary Table S1) but was otherwise similar between pathways. All participants in the intensive pathway who responded (with at least a minimal response) to induction chemotherapy were planned to go on to receive high-dose melphalan and ASCT. Peripheral blood stem cell harvest was planned to commence after the participant had completed the induction and intensification (if applicable) treatment. Stem cell mobilization and stem cell harvest were performed according to local practice but with advice to aim for the collection of enough stem cells for at least two transplants. High-dose melphalan and ASCT were given according to local practice. Adjustment for renal insufficiency was advised. Participants with serum creatinine <200 mmol/L prior to transplantation were to receive the standard dose of 200 mg/m2 melphalan while those with serum creatinine >200 mmol/L were to receive 140 mg/m2. There was no recommendation to reduce the melphalan dose based on age in the protocol. This is a retrospective, exploratory analysis of data from the Myeloma XI trial. For the first set of analyses patients in the TE

haematologica | 2022; 107(1)

ASCT for fit, older patients in Myeloma XI

pathway were categorized by age group <65, 65-69 and 70-75 and their baseline characteristics, treatment and harvest data summarized. PFS and OS, measured from baseline trial randomization, were compared between age groups using the Kaplan-Meier method. Comparisons were made between the allocated groups using the Cox proportional hazards model stratified by the minimization stratification factors, excluding center, and hazard ratios (HR) and 95% confidence intervals (95% CI) were estimated. The frequencies of reported serious adverse events and patients’ deaths were examined to compare transplant-related morbidity and mortality between age groups. Relative survival estimates were obtained using flexible parametric survival models on the hazard scale with four degrees of freedom17 with the same covariates included in the model. Relative survival was defined as the observed survival divided by the expected survival where the expected survival was obtained from national life tables stratified by age at diagnosis, sex and calendar year. UK life-time risk was estimated from data available from the Office for National Statistics.18 In order to compare outcomes between patients undergoing ASCT or not, a second set of analyses was performed using an age-matched group of patients in the TE and TNE pathways. This was defined by overlap of the age distributions in each pathway and comprised the older patients within the TE pathway and younger patients within the TNE pathway but excluded the extremes in both pathways. The optimal overlap was of patients aged 64-70 within each pathway, which was chosen with the aim of maximizing the number of patients in the analysis while achieving a balance between the patients receiving ASCT (52.5%) and those who did not receive ASCT (47.5%). Three groups were considered: (i) TE patients who underwent ASCT (TE-ASCT); (ii) TE patients who did not undergo ASCT (TE-noASCT) and (iii) TNE patients (who did not undergo ASCT). The characteristics of the patients in the three groups at entry into the trial were summarized. Patients were scored according to the UK Myeloma Research Alliance Myeloma Risk Profile (MRP)19 and the proportion of patients in each of the groups compared. When analyzing time-to-event outcomes in order to avoid a survivor or immortal time bias that would be incurred by comparing all patients within these groups from baseline, patients were only included if they remained eligible to continue in the trial at the end of induction ± intensification and their outcomes were measured from the end of induction ± intensification therapy. This time-point was close to the ASCT date for the patients in the TE-ASCT group, but represents a common time-point that could be identified in all patients to enable comparison. To estimate the treatment effect of ASCT as compared to no ASCT in this subgroup, propensity score weighting using inverse probability of treatment weights was implemented. Propensity score weighting is a useful tool to account for imbalances in observed confounders between groups when estimating treatment effects from non-randomized data. A propensity score is a single score that represents the probability of receiving a treatment, conditional on a set of observed covariates. The goal of creating a propensity score is to balance covariates between individuals who did and did not receive a treatment, making it easier to isolate the effect of a treatment. The propensity score was based on a patient’s age, sex, World Health Organization (WHO) performance status, International Staging System (ISS) stage, induction treatment and response after completing induction treatments. The propensity score was applied using normalized inverse probability of treatment weighting (IPTW). In IPTW, each treated subject (ASCT) receives a weight equal to the inverse of the propensity score, and each control subject (no ASCT) receives a weight equal to the inverse of one minus the propensity score.

haematologica | 2022; 107(1)

For the IPTW analysis, probability weights were applied to the individual participants’ data for calculation of the survivor function estimate and partial likelihood in the Cox model. Statistical analysis was performed using SAS v9.4 (SAS Institute Inc., Cary, NC, USA) and Stata IC v16 (StataCorp. College Station, TX, USA). PFS was defined as the time from the point stated above to the date of confirmed disease progression or death from any cause. OS was defined as the time from the point stated above to the date of death from any cause. Cytogenetic profiling was performed using multiplex ligation-dependent probe amplification and quantitative real-time polymerase chain reaction analysis.20,21 Cytogenetic risk was defined as standard when there were no adverse lesions, high in the presence of t(4;14), t(14;16), t(14;20), del(17p), or gain(1q), and ultra-high when more than one adverse lesion was present.22,23 The data cut-off for inclusion in this analysis was May 31, 2019.

Results Outcomes for transplant-eligible pathway patients by age The 2,042 patients enrolled in the TE pathway had a median age of 61 (28-75) years; 546 (27%) were aged 65-69 and 101 (5%) were aged 70-75 (Table 1). Older patients were more frequently categorized as ISS stage III and had a lower performance status than younger patients. There was no significant difference in the proportion of patients in each of the cytogenetic risk groups or the number of patients in each arm of the induction treatment randomization between age groups. Response at the end of induction was similar across age groups. Older patients in the TE pathway were less likely than younger patients to undergo stem cell harvest at the end of induction. The percentage of patients undergoing stem cell harvest fell from 73.5% in those aged <65 years, to 62.2% among those aged 65-69 and only 57.4% in the group aged 70 or older (Table 1, Online Supplementary Figure S1). The reason given for not proceeding to stem cell harvest and subsequent transplantation was more likely to be due to the clinician/patient not considering that they were fit enough to proceed in the older age groups than in the younger patients (Table 1). Older patients had a lower median harvested CD34+ cell count but still had a high rate of achieving the standard cutoff of 2x106 CD34+ cells/kg needed for one ASCT. The percentage of patients achieving this target was 95.0% in those aged <65, 90.0% in those aged 65-69 and 88.7% among patients aged 70 or older. Conversely, fewer patients in the older age groups achieved the cut-off of 4x106 CD34+ cells/kg considered adequate for two ASCT, with reductions from 63.4% to 45.6% to 32.1% in the respective age groups. Of the 2,042 patients in the TE pathway, 1,370 (67%) received melphalan. Most patients (84.7%) received 200 mg/m2 with only 10.5% reported to have received 140 mg/m2. The proportion receiving the lower dose increased in the older age groups with the 140 mg/m2 dose being given to only 5.5% of patients aged <65 years, but 19.9% of those aged 65-69 and 45.5% of those aged >70 years. This appeared to be due to both an increased incidence of elevated serum creatinine (>200 mmol/L) in the older age groups and the systematic use of the lower dose for older patients in some centers. Response to transplant, PFS and OS outcomes for patients of different ages within the TE pathway who underwent 233

C. Pawlyn et al. Table 1. Baseline characteristics and treatment received for patients within the transplant-eligible pathway and by age group.

All n=2042 Baseline characteristics Age, median (range), years Sex, n (%) Male Female Performance status (World Health Organization), n (%) 0 1 2 ≥3 Not available International Staging System stage, n (%) I II III Not available Cytogenetic profile#, n (%) Standard risk High risk Ultrahigh risk Treatment within TE pathway Induction randomisation treatment arm, n (%) CTD CRD Patients’ response at end of Induction (+intensification when received), n (%) Complete response Very good partial response Partial response Minimal response No change Progressive disease Unable to assess No induction treatment or non- protocol treatment Death within 60 days of initial randomisation or within 60 of intensification Not available Patients who underwent stem cell harvest, n (%) Yes No Unknown Reasons stem cell harvest not performed, n (%) Patient decision Patient not fit/clinicians decision Disease progression Death Allogeneic transplant Other Median number of CD34+ cells harvested x106/kg (range) Patients achieving CD34+ stem cell harvest thresholds, n (%)*: ≥ 2x106/kg ≥ 4x106/kg Patients who received melphalan (% of all patients in age group) Dose of melphalan administered (% of all patients who received melphalan), n (%) 70 mg/m2 100 mg/m2 140 mg/m2 200 mg/m2 Unknown/other dose

<65 n=1396

Age group 65-69 n=545

70-75 n=101

61 (28-75)

57 (28-64)

67 (65-69)

71 (70-75)

1221 (59.8) 821 (40.2)

821 (58.8) 575 (41.2)

342 (62.8) 203 (37.2)

58 (57.4) 43 (42.6)

865 (42.4) 732 (35.8) 257 (12.6) 88 (4.3) 100 (4.9)

612 (43.8) 466 (33.4) 182 (13.0) 67 (4.8) 69 (4.9)

205 (37.6) 225 (41.3) 67 (12.3) 20 (3.7) 28 (5.1)

48 (47.5) 41 (40.6) 8 (7.9) 1 (1.0) 3 (3.0)

611 (29.9) 782 (38.3) 501 (24.5) 148 (7.2)

440 (31.5) 526 (37.7) 331 (23.7) 99 (7.1)

141 (25.9) 219 (40.2) 141 (25.9) 44 (8.1)

30 (29.7) 37 (36.6) 29 (28.7) 5 (5.0)

475 (55.2) 273 (31.7) 112 (13.0)

318 (54.8) 185 (31.9) 77 (13.3)

129 (55.4) 71 (30.5) 33 (14.2)

28 (59.6) 17 (36.2) 2 (4.3)

1021 (50.0) 1021 (50.0)

701 (50.2) 695(49.8)

273 (50.1) 272 (49.9)

47 (46.5) 54 (53.5)

149 (7.3) 1125 (55.1) 513 (25.1) 60 (2.9) 16 (0.8) 44 (2.2) 23 (1.1) 17 (0.8) 36 (1.8)

102 (7.3) 769 (55.1) 355 (25.4) 37 (2.7) 12 (0.9) 30 (2.2) 13 (0.9) 12 (0.9) 22 (1.6)

37 (6.8) 297 (54.5) 137 (25.1) 20 (3.7) 4 (0.7) 10 (1.8) 8 (1.5) 5 (0.9) 13 (2.4)

10 (9.9) 59 (58.4) 21 (20.8) 3 (3.0) 0 (0.0) 4 (4.0) 2 (2.0) 0 (0.0) 1 (1.0)

59 (2.9)

44 (3.2)

14(2.6)

1 (1.0)

1423 (69.7) 565 (27.7) 54 (2.6)

1026 (73.5) 332 (23.8) 38 (2.7)

339 (62.2) 192 (35.2) 14 (2.6)

58 (57.4) 41 (40.6) 2 (2.0)

129 (22.8) 200 (35.4) 70 (12.4) 58 (10.3) 8 (1.4) 100 (17.7) 4.4 (0.0-90.2)

70 (21.1) 102 (30.7) 52 (15.7) 33 (9.9) 8 (2.4) 67 (20.2) 4.6 (0.0-88.8)

48 (25.0) 81 (42.2) 13 (6.8) 23 (12.0) 0 (0.0) 27 (14.1) 3.8 (0.0-90.2)

11 (26.8) 17 (41.5) 5 (12.2) 2 (4.9) 0 (0.0) 6 (14.6) 3.1 (0.0-7.7)

1277 (93.6) 791 (58.0) 1370 (67.1)

941 (95.0) 628 (63.4) 993 (71.1)

289 (90.0) 146 (45.6) 322 (59.1)

47 (88.7) 17 (32.1) 55 (54.5)

1 (0.1) 14 (1.0) 144 (10.5) 1161 (84.7) 50 (3.6)

0 (0) 8 (0.8) 55 (5.5) 895 (90.1) 35 (3.5)

0 (0) 5 (1.6) 64 (19.9) 239 (74.2) 14 (4.3)

1 (1.8) 1 (1.8) 25 (45.5) 27 (49.1) 1 (1.8) continued on following page

234

haematologica | 2022; 107(1)

ASCT for fit, older patients in Myeloma XI

continued from previous page

Patients who received a melphalan dose and stem cell return (% of all patients in age group), n (%) Patients response after- ASCT (% of all patients who received a melphalan dose and stem cell return), n (%) Complete response Very good partial response Partial response Minimal response Progressive disease Unable to assess Death up to and including 100 days after ASCT Not available

1366 (66.9)

990 (70.9)

321 (58.9)

55 (54.5)

297 (21.7) 798 (58.4) 201 (14.7) 4 (0.3) 26 (1.9) 10 (0.7) 9 (0.7) 21 (1.5)

225 (22.7) 579 (58.5) 135 (13.6) 4 (0.4) 19 (1.9) 8 (0.8) 5 (0.5) 15 (1.5)

62 (19.3) 186 (57.9) 60 (18.7) 0 (0.0) 4 (1.2) 0 (0.0) 3 (0.9) 6 (1.9)

10 (18.2) 33 (60.0) 6 (10.9) 0 (0.0) 3 (5.5) 2 (3.6) 1 (1.8) 0 (0.0)

# data available for 860 of 2,042 (42.1%) patients (580 of 1,396 [41.5%] patients aged <65 years, 233 of 545 [42.8%] patients 65-69 years and 47 of 101 [46.5%] patients 70-75 years). The percentages given are of those with available data. *data available for 1,364 of the 1,423 patients who underwent stem cell harvest (991 of 1,026 patients aged <65 years, 320 of 339 patients aged 65-69 years, 53 of 55 patients aged 70-75 years). The percentages given are of those with data available. TE: transplant-eligible; CTD: cyclophosphamide, thalidomide and dexamethasone; CRD: cyclophosphamide, lenalidomide and dexamethasone; ASCT: autologous stem cell transplantation..

ASCT were compared. Patients of different ages achieved a similar depth of response at 100 days after ASCT (Table 1) with an improvement in response compared to the end of induction seen in 863 of 1,366 (63.2%) patients overall (62.7% of those aged <65, 64.5% of those aged 65-69 and 63.6% of those aged over 70). The median PFS was longest for patients aged under 65 and fell with increasing age (Figure 1). The median PFS was 50.8 months (95% CI: 46.3-54.9) for those aged <65 years, 40.0 months (95% CI: 36.3-46.0) for those aged 65-69 and 34.4 months (95% CI: 27.5-46.4) for those aged 70-75 years. The PFS for patients aged 65-69 years was shorter than that of patients aged under 65 (HR=1.26, P=0.003). Patients aged 70 years or over had a shorter PFS than the <65-year age group (HR=1.57, P=0.009), but not significantly shorter than that of the 65-69 age group (HR=1.24, P=0.229). The median OS was 95.5 months (95% CI: 89.8-not reached) for those aged <65 years, 91.9 months (95% CI: 82.3-not reached) for those aged 65-69 and 76.0 months (95% CI: 58.7-not reached) for those aged 70-75 years. There was no significant difference in the OS between any of the age groups (65-69 vs. <65: HR=1.09, P=0.484; >70 vs. <65: HR=1.59, P=0.051; 70-75 vs. 65-69, HR=1.47, P=0.127). The 5-year OS was 75.5% (95% CI: 72.6%78.3%) for those aged <65 years, 72.7% (95% CI: 67.3%78.1%) for those aged 65-69 and 65.0% (95% CI: 49.5%80.5%) for those aged 70-75 with some evidence of dissociation of the survival curve for the 70-75 age group after 3 years. There was no strong evidence of a difference in OS outcome when accounting for population-level mortality risk (excess mortality hazard-rate ratio [EHR] for OS 65-69 vs. <65: EHR=0.95, P=0.736; 70-75 vs. <65: EHR=1.33, P=0.368) (Online Supplementary Figures S2 and S3, Online Supplementary Table S2). There were also no differences in the percentages of patients who were reported as having commenced second-line therapy at the time of data cut-off (<65: 43.7%, 65-69: 49.5% and 70-75: 41.8%). However, there were notable differences in the proportions of patients whose second-line therapy included a second ASCT (23.8%, 9.4% and 8.7%, respectively). Survival outcomes were explored stratified by melphalan dose (Online Supplementary Figure S4). Taking the whole TE population, the 140 mg/m2 dose of melphalan appeared to be associated with a shorter PFS compared to the 200 mg/m2 dose (HR=1.31, P=0.012), with no difference in OS (HR=1.29, P=0.086) (Online Supplementary Figure S4A). However, this result was confounded by age as there were more patients who received 140 mg/m2 in the older age haematologica | 2022; 107(1)

group which was associated with inferior PFS (Figure 1). When examined within each of the age groups there was no difference between outcome for patients who received 140 mg/m2 in comparison to those who received 200 mg/m2: age <65: PFS HR=1.20, P=0.289; OS HR=1.35, P=0.189 (Online Supplementary Figure S4B); age 65-69: PFS HR=1.18, P=0.344; OS HR=0.97, P=0.905 (Online Supplementary Figure S4C); and age >70: PFS HR=1.35, P=0.442; OS HR=1.42, P=0.522 (Online Supplementary Figure S4D). Transplant-related morbidity and mortality were examined by comparing serious adverse events reported within 100 days of ASCT and deaths occurring within 100 days or 365 days in the different age groups. There were 230 serious adverse events reported within 100 days of ASCT: 172 events in 149 patients in the age group <65 years old (15.1% of patients), 54 events in 47 patients in the 65-69 age group (14.6%) and four events in four patients in the 70-75 age group (7.3%). Nine patients died within 100 days of ASCT: five aged <65 (0.5%), three aged 65-69 (0.9%) and one aged >70 (1.8%). Forty-eight patients died within 365 days of ASCT: 34 who were <65 years old (3.4%), 11 who were 65-69 (3.4%) and three who were 70-75 years old (5.5%). These data suggest that there is not a significant increase in mortality or morbidity after ASCT in older patients.

Outcomes in an age-matched group of older patients comparing transplantation to no transplantation Analysis of the patients in the TE pathway undergoing transplantation by age does not address the question of whether, at older ages, ASCT continues to be associated with better outcomes than those of patients of an equivalent age not undergoing ASCT. In order to answer this question, an age-matched group of patients was identified as described above and shown in Figure 2A. At baseline, patients in both the TNE and TE-noASCT groups had higher performance status and ISS stage than patients in the TEASCT group (Table 2). Response at the end of induction therapy was deeper in the patients in the TE-ASCT group than in the other two groups. Older patients undergoing ASCT (TE-ASCT) had a longer median PFS than that of age-matched patients not undergoing ASCT, whether TE-noASCT or TNE (Figure 2B). The median PFS for the TE-ASCT group was 39.4 months compared to 9.7 months for the TE-noASCT group and 16.5 months for the TNE group. Comparing those patients who underwent ASCT (TE-ASCT) to those who did not (TE235

C. Pawlyn et al.

Figure 1. Outcomes of patients of different ages undergoing autologous stem cell transplantation. (A, B) Progression-free survival (A) and overall survival (B) of patients in the groups aged <65 years (blue), 65-69 years (red) and 70-75 years (yellow). PFS: progression-free survival; OS: overall survival; HR: hazard ratio; 95% CI: 95% confidence interval.

noASCT or TNE), ASCT was associated with a significant improvement in PFS (ASCT vs. noASCT: HR=0.41, P<0.0001) (Figure 2D). The same benefit was seen in terms of OS: TE-ASCT median 84.1 months, TE-noASCT 50.9 months, TNE 60.2 months (Figure 2C) (ASCT vs. noASCT: HR=0.51, P<0.0001) (Figure 2E). The benefit of ASCT was independent of the subsequent use of mainte236

nance therapy, with longer PFS and OS seen in the TEASCT group than in the TNE group whether patients were randomized to observation or maintenance therapy (Online Supplementary Figure S5). Where possible a frailty-surrogate score was derived for patients in each of the age-matched groups using the UK Myeloma Research Alliance Risk Profile (MRP)17 (Table 2). haematologica | 2022; 107(1)

ASCT for fit, older patients in Myeloma XI

Figure 2. Outcomes of patients in agematched groups. (A) Histogram showing the age distribution of patients in the transplanteligible and transplant-ineligible pathways with the patients included in the age-matched groups highlighted. (B,C) Progression-free survival (B) and overall survival (C) of the agematched population (TE-ASCT [blue], patients in the TE pathway who underwent autologous stem cell transplant; TE-noASCT [red], patients in the TE pathway who did not undergo ASCT; TNE [yellow], patients in the transplant-ineligible pathway. (D,E) Progressionfree survival (D) and overall survival (E) comparing those patients who underwent ASCT with those who did not (ASCT [blue], no ASCT. TE: transplant eligible; TNE: transplant ineligible; HR: hazard ratio; 95%CI: 95% confidence interval.

The TE-ASCT group had most patients with a low-risk MRP score and least with a high-risk score compared to the other groups. The apparent differences in baseline variables and endof-induction responses may have confounded the comparisons between groups. To compensate for this propensity score, IPTW was used to adjust the estimate of the treatment effect of ASCT compared to noASCT in the age-matched group of patients. As expected, the adjustment had the effect of reducing the median PFS and OS for patients in the TE-ASCT group and increasing the median PFS and OS for the patients in the other two groups as compared to the values emerging from the unweighted intention-to-treat analysis. After adjustment,

the median PFS for the TE-ASCT group was 35.8 months compared to 10.4 months for the TE-noASCT group and 16.9 months for the TNE group (Figure 3A). Comparing the impact of ASCT versus no ASCT, the hazard ratio remained statistically significant (HR=0.44, P<0.001). The same benefit was seen in terms of OS, with the median OS for the TE-ASCT group being 79.8 months compared to 57.3 months for the TE-noASCT group and 59.5 months for the TNE group (Figure 3B) (ASCT vs. noASCT: HR=0.53, P<0.001). This analysis suggests that even when the measured baseline covariates were appropriately weighted, there remained a significant treatment benefit of ASCT as compared to no ASCT. Morbidity and mortality were examined by comparing

haematologica | 2022; 107(1)

237

C. Pawlyn et al.

Table 2. Baseline characteristics of patients in the age-matched groups.

All Age, years; median (range) Sex, n (%) Male Female Performance status, (WHO), n (%) 0 1 2 ≥3 Not available International Staging System, n (%) I II III Not available Cytogenetic profile#, n (%) Standard risk High risk Ultrahigh risk Induction Randomisation Treatment, n (%) CTD/CTDa CRD/CRDa MRP possible to define, n (%) Yes No MRP risk (% based on those patients with all MRP data available), n (%) Low Intermediate High Patients’ response after Induction (+intensification where received), n (%) Complete response Very good partial response Partial response Minimal response

n=770

TE-ASCT n=404

Age-matched groups TE-noASCT n=129

TNE n=237

67.0 (64.0-70.0)

66.0 (64.0-70.0)

67.0 (64.0-70.0)

68.0 (64.0-70.0)

471 (61.2) 299 (38.8)

264 (65.3) 140 (34.7)

73 (56.6) 56 (43.4)

134 (56.5) 103 (43.5)

269 (34.9) 321 (41.7) 109 (14.2) 34 (4.5) 37 (4.8)

160 (39.6) 173 (42.8) 41 (10.1) 7 (1.7) 23 (5.7)

44 (34.1) 51 (39.5) 20 (15.5) 10 (7.8) 4 (3.1)

65 (27.4) 97 (40.9) 48 (20.3) 17 (7.2) 10 (4.2)

199 (25.8) 318 (41.3) 195 (25.3) 58 (7.5)

125 (30.9) 164 (40.6) 79 (19.6) 36 (8.9)

29 (22.5) 52 (40.3) 43 (33.3) 5 (3.9)

45 (19.0) 102 (43.0) 73 (30.8) 17 (7.2)

184 (55.9) 105 (31.9) 40 (12.2)

86 (51.8) 55 (33.1) 25 (15.1)

37 (57.8) 19 (29.7) 8 (12.5)

61 (61.6) 31 (31.3) 7 (7.1)

345 (44.8) 425 (55.2)

194 (48.0) 210 (52.0)

62 (48.1) 67 (51.9)

89 (37.6) 148 (62.4)

646 (83.9) 124 (16.1)

303 (75.0) 101 (25.0)

106 (82.2) 23 (17.8)

237 (100) 0 (0)

430 (66.6) 152 (23.5) 64 (9.9)

243 (80.2) 47 (15.5) 13 (4.3)

60 (56.6) 26 (24.5) 20 (18.9)

127 (53.6) 79 (33.3) 31 (13.1)

94 (12.2) 554 (72.0) 114 (14.8) 8 (1.0)

51 (12.6) 286 (70.8) 65 (16.1) 2 (0.5)

10 (7.8) 97 (75.2) 19 (14.7) 3 (2.3)

33 (13.9) 171 (72.2) 30 (12.7) 3 (1.3)

# data available for 329 of 770 (42.7%) patients (166 of 404 [41.1%] TE-ASCT patients, 64 of 129 [49.6%] TE-noASCT patients and 99 of 237 [41.8%] TNE patients). Percentages given are of those with data available.TE: transplant eligible, ASCT: autologous stem cell transplant; TNE: not eligible for transplantation; WHO: World Health Organization; CTD: cyclophosphamide, thalidomide and dexamethasone; CRD: cyclophosphamide, lenalidomide and dexamethasone; CTDa: attenuated CTD; CRDa: attenuated CRD; MRP: UK Myeloma Research Alliance Myeloma Risk Profile.

serious adverse events and deaths reported within 100 days of the end of induction (± intensification). Two hundred and three serious adverse events were reported within 100 days: 132 events in 105 patients in the TEASCT group (26.0% of patients), 70 events in 49 patients in the TE-noASCT group (38.0%) and 65 events in 53 patients in the TNE group (22.4%). Five patients died within 100 days: none in the TE-ASCT group, two in the TE-noASCT group (2.7%) and three in the TNE group (2.18%). Thirty-seven patients died within 365 days: nine in the TE-ASCT group (2.2%), 14 in the TE-noASCT group (10.9%) and 14 in the TNE group (5.9%).

Discussion These results demonstrate that ASCT is safe and effective for selected, fit, myeloma patients up to the age of 75. In an age-matched population treated with similar induction therapy there was a significant benefit for PFS and OS 238

associated with the use of ASCT compared to no ASCT. The study showed that even in a group of patients initially felt to be eligible for transplantation by their treating clinician, there was a clear fall in the proportion of patients undergoing stem cell harvest and ASCT with increasing age. This likely reflects clinicians’ enthusiasm for giving patients the option of having an ASCT, by enrolling the patients in the TE pathway, but a subsequent realization that the patients were not fit enough. Commencing intensive induction therapy and using this as a therapeutic trial of fitness before making the final decision regarding ASCT may represent a valid approach to therapy especially in the intermediate age group of those aged 65-75 years. Although the median CD34+ harvest cell count was lower in older patients it is not certain whether this reflects a true difference in mobilization. The percentage of patients from whom enough stem cells were collected for one ASCT (2x106 CD34+ cells/kg) was very high across haematologica | 2022; 107(1)

ASCT for fit, older patients in Myeloma XI

Figure 3. Outcomes of patients in age-matched groups including inverse probability of treatment weighting. (A) Progression-free survival. (B) Overall survival. TE-ASCT (blue): patients in the transplanteligible pathway who underwent autologous stem cell transplantation; TE-noASCT (red): patients in the transplant-eligible pathway who did not undergo autologous stem cell transplantation; TNE (yellow): patients in the transplant-ineligible pathway; ITT: intention to treat; IPTW: inverse probability of treatment weighting adjustment.

the age groups. It is unknown what the local investigators set as the target harvest for each patient and it may be that the target for older patients was to collect enough stem cells for one transplant rather than to also save some for a possible subsequent transplant given that by the time of relapse the older patients would have achieved an even more advanced age. The PFS of patients in the TE pathway aged 65-69 and those aged 70-75 was shorter than that of patients aged under 65 years. This would be expected as outcomes are known to diminish with increasing age with all myeloma therapies. There was no significant difference in OS, although the survival curves appeared to dissociate for the 70-75 age group after 3 years. To further investigate this we performed OS analysis corrected for population-level mortality risk and found no evidence of a difference in survival. ASCT delivery to selected older patients was safe; there was no difference in survival at 3 months or 1 year after ASCT between age groups. Indeed, fewer serious adverse events occurred within 100 days of ASCT among haematologica | 2022; 107(1)

those in the oldest age group. This may be due to the small size of this age group or due to more stringent selection for fitness in these older patients. There was no significant difference in PFS or OS in this study when comparing patients of a similar age who received 140 mg/m2 melphalan (due to renal impairment or clinician choice) or 200 mg/m2 melphalan as conditioning for ASCT. This reinforces the approach of using a dose reduction of melphalan conditioning only in these selected subsets of patients, with no apparent detriment to outcomes. One previous study suggested that there was increased toxicity with the higher dose in those aged over 70 years24 but a much larger and more recent study of the EBMT Registry database did not reveal any significant difference in survival outcomes between the groups receiving different doses in the overall population, but a benefit from the use of 200 mg/m2 in those with a suboptimal response.25 As in our study, far fewer patients received the 140 mg/m2 dose than the 200 mg/m2 dose in the EBMT analysis, suggesting that the lower dose was only used in 239

C. Pawlyn et al.

very selected older patients or in those with renal failure. In the Myeloma XI trial the induction therapy for TE and TNE patients was the same triplet combination. This provided the opportunity to compare outcomes for those patients undergoing transplantation with patients of the same age who did not undergo transplantation but who had received the same induction therapy. We performed this analysis from the end of induction (± intensification if given) and only included patients who would have been eligible to continue in the study to avoid survivor and immortal time biases. This comparison showed a marked improvement in PFS and OS associated with ASCT. It is important to note that this comparison between ASCT and no ASCT was not randomized and therefore remains inherently subject to bias. Patients in the TE pathway of the trial were selected on the basis of clinician’s judgement and patient’s preference. We found that the older patients included in the TE pathway had a lower performance status than younger patients, whereas performance status usually increases with age. This suggests active selection of only the fittest older patients for entry into the TE pathway and consideration of ASCT. Consistent with this, in the age-matched population the performance status and ISS stage were higher in both the TE-noASCT and TNE groups than in the TE-ASCT patients. Unfortunately, data to calculate the International Myeloma Working Group frailty score,26 the Revised Myeloma Comorbidity Index27 or the Hematopoietic Cell Transplant Comorbidity Index28 were not collected within the study. We applied the UKMRP, an outcome score previously validated only in the TNE population, across the groups and found more patients in the TNE and TE-noASCT groups to have higher-risk MRP than in the TE-ASCT group. Additionally, the TE-ASCT group had achieved deeper responses at the end of induction. This may have affected their outcomes irrespective of transplantation. To address this, a matched analysis using IPTW was performed with factors including those associated with frailty and response to induction that were different between the age-matched groups. This analysis confirmed the markedly improved PFS and OS for the ASCT-TE group, suggesting that this finding was not confounded by fitness or prior response. It should be noted that propensity scores only balance measured covariates as confounders, and balance in measured covariates does not necessarily indicate balance in unmeasured confounders. If unmeasured covariates are confounders, they can bias treatment effect estimates. These results should therefore be interpreted with caution. Previous studies in younger, fitter patients under the age of 65 have demonstrated the efficacy of transplantation in the era of modern therapy including studies combining proteasome inhibitors and immunomodulatory agents for all patients as induction therapy. Such approaches may now be considered better than the largely immunomodulatory agent-based induction delivered in Myeloma XI. In the IFM/DFCI 2009 study patients received bortezomib, lenalidomide and dexamethasone (VRD) induction before randomization between ASCT and consolidation in the form of additional cycles of VRD, with ASCT deferred to first relapse.29 The trial demonstrated an association between improved PFS and early ASCT while follow-up for OS is ongoing. The EMN02 trial conducted a similar comparison but in the context of induction with cyclophosphamide, bortezomib and dexamethasone and compared the use of bortezomib, melphalan and pred240

nisone (VMP) consolidation to ASCT,6 also demonstrating a PFS advantage for first-line ASCT. Our data support the findings of these studies and extend them to older patients who were excluded from these trials. Sub-analysis of IFM/DFCI 2009 suggests that there is no benefit of transplantation in patients achieving very deep minimal residual disease-negative responses prior to ASCT. This is of great interest as it could lead to a response-adapted approach to ASCT. Minimal residual disease data were collected for a subset of the patients within the Myeloma XI trial, but there were too few patients in the agematched population to perform this analysis. Within the last 18 months two large phase III studies showed that the addition of daratumamb to standard induction regimens (lenalidomide-dexamethasone and VMP) dramatically improved the PFS and OS of older patients.30-32 TNE patients were randomized into these studies, although the reason for transplant ineligibility (age, co-morbidities) was not stated. Randomized clinical trials are therefore still warranted for older patients who are fit for transplant comparing an antibody-containing regimen ± transplantation. In summary, these findings support the use of ASCT for selected, fit older myeloma patients up to the age of 75. With effective clinician selection, older patients undergoing ASCT can experience long PFS and OS, comparable to those of younger patients, and without any significant increase in morbidity or mortality. Disclosures CP has provided consultancy services for and received travel support from Amgen and Takeda Oncology; honoraria and travel support from Janssen; and consultancy fees, honoraria, and travel support from Celgene Corporation. DAC has received research funding from Celgene Corporation, Amgen, and Merck Sharp and Dohme. TM has received research funding from Celgene Corporation, Amgen, and Merck Sharp and Dohme. JRJ has received honoraria and research funding from Celgene Corporation. MWJ has acted as a consultant for and received honoraria, travel support, and research funding from Janssen; has acted as a consultant for and received honoraria and travel support from Takeda and Amgen; has acted as a consultant for and received honoraria and research funding from Celgene Corporation; and has acted as a consultant for and received honoraria from Novartis. GC has provided consultancy and speakers bureau services for and received honoraria and research funding from Takeda, Celgene Corporation, Janssen and Amgen; has acted as a consultant for and received honoraria from Glycomimetics and Bristo-Myers Squibb; and has provided consultancy and speakers bureau services for and received honoraria from Sanofi. KDB has received honoraria from Celgene, Janssen, and Amgen; acted in a consulting or advisory role for Celgene, Janssen, Takeda, and Novartis; and received travel support from Celgene, Janssen, and Takeda. MTD has equity ownership and membership on the board of directors or advisory committees of Abingdon Health. MFK has acted as a consultant for and received travel support from Bristol-Myers Squibb and Takeda; has acted as a consultant for Chugai; has acted as a consultant for and received honoraria from Janssen and Amgen; and has acted as a consultant for and received honoraria and research funding from Celgene Corporation. RGO has received honoraria and travel support from Takeda; has provided consultancy services for and received travel support from Janssen; and has acted as a consultant for and received honoraria and research funding from Celgene Corporation. WG has provided consultancy services for and haematologica | 2022; 107(1)

ASCT for fit, older patients in Myeloma XI

received research funding from Celgene Corporation; has received research funding from Amgen and Merck Sharp and Dohme; and has received honoraria from Janssen. GJM has received research funding from Janssen; has provided consultancy services for and received honoraria from Bristol-Myers Squibb, Takeda, Roche, Amgen, GSK and Karyopharm; and has acted as a consultant for and received honoraria and research funding from Celgene Corporation. GHJ has provided consultancy and speakers bureau services for and received honoraria from Roche, Amgen, Janssen, and Merck Sharp and Dohme; and has provided consultancy and speakers bureau services for and received honoraria, travel support and research funding from Celgene Corporation and Takeda. FED has received honoraria from Adaptive; has provided consultancy services for and received honoraria and research funding from Celgene Corporation; and has provided consultancy services for and received honoraria from Janssen, Oncopeptide, Roche, Sanofi, and Takeda. Contributions CP, DAC and FED designed this analysis; GJM, GHJ and FED were Chief Investigators of the Myeloma XI trial; CP, KDB, JRJ, MWJ, GC, MFK, RGO, GJM, GHJ and FED participated in the recruitment and management of patients; MFK, MTD, RGO and GJM coordinated the central laboratory investigations; CP, DAC, TM and FED analyzed and interpreted the data for this analysis; CP, DAC and TM drafted the manuscript. All authors contributed to critically revising the manuscript and approved the final submitted version. Acknowledgments We thank all the patients at centers throughout the UK whose willingness to participate made this study possible. We are grateful to the UK National Cancer Research Institute Haematological Oncology Clinical Studies Group, UK Myeloma Research Alliance, and to all principal investigators, sub-investigators, and local center staff for their dedication and commitment to recruiting

References 1. Child JA, Morgan GJ, Davies FE, et al. Highdose chemotherapy with hematopoietic stem-cell rescue for multiple myeloma. N Engl J Med. 2003;348(19):1875-1883. 2. Attal M, Harousseau JL, Stoppa AM, et al. A prospective, randomized trial of autologous bone marrow transplantation and chemotherapy in multiple myeloma. Intergroupe Francais du Myelome. N Engl J Med. 1996;335(2):91-97. 3. Fermand JP, Katsahian S, Divine M, et al. High-dose therapy and autologous blood stem-cell transplantation compared with conventional treatment in myeloma patients aged 55 to 65 years: long-term results of a randomized control trial from the Group Myelome-Autogreffe. J Clin Oncol. 2005;23(36):9227-9233. 4. Bladé J, Rosiñol L, Sureda A, et al. High-dose therapy intensification compared with continued standard chemotherapy in multiple myeloma patients responding to the initial chemotherapy: long-term results from a prospective randomized trial from the Spanish cooperative group PETHEMA. Blood. 2005;106(12):3755-3759. 5. Attal M, Lauwers-Cances V, Hulin C, et al. Lenalidomide, bortezomib, and dexamethasone with transplantation for myeloma. N Engl J Med. 2017;376(14):1311-1320.

haematologica | 2022; 107(1)

patients to the study. We thank the members of the Myeloma XI Trial Steering Committee and Data Monitoring and Ethics Committee. The support of the Clinical Trials Research Unit at the University of Leeds was essential to the successful running of the study; we thank all their staff who have contributed, past and present. Central laboratory analysis was performed at the Institute of Immunology and Immunotherapy, University of Birmingham; the Institute of Cancer Research, London; and the Haematological Malignancy Diagnostic Service, St James’s University Hospital, Leeds. We are very grateful to the laboratory teams for their contribution to the study. We also acknowledge support from the National Institute of Health Biomedical Research Centre at the Royal Marsden Hospital and the Institute of Cancer Research. Funding Primary financial support was from Cancer Research UK (C1298/A10410). Unrestricted educational grants from Celgene Corporation, Amgen, and Merck Sharp and Dohme, and funding from Myeloma UK supported trial coordination and laboratory studies. The authors are solely responsible for the study design, data collection, data analysis and interpretation, writing, and decisions about publication submission; no funder had any role in these aspects of the trial. Trial data were accessible to all authors. Data-sharing statement De-identified participant data will be made available when all primary and secondary endpoints of the trial have been met. Any requests for trial data and supporting material (data dictionary, protocol, and statistical analysis plan) will be reviewed by the trial management group in the first instance. Only requests that have a methodologically sound basis and whose proposed use of the data has been approved by the independent trial steering committee will be considered. Proposals should be directed to the Chief Investigator responsible for trial governance, Prof. Graham Jackson in the first instance; to gain access, data requestors will need to sign a data access agreement.

6. Cavo M, Gay F, Beksac M, et al. Autologous haematopoietic stem-cell transplantation versus bortezomib-melphalan-prednisone, with or without bortezomib-lenalidomidedexamethasone consolidation therapy, and lenalidomide maintenance for newly diagnosed multiple myeloma (EMN02/HO95): a multicentre, randomised, open-label, phase 3 study. Lancet Haematol. 2020;7(6):e456e468. 7. Auner HW, Szydlo R, Hoek J, et al. Trends in autologous hematopoietic cell transplantation for multiple myeloma in Europe: increased use and improved outcomes in elderly patients in recent years. Bone Marrow Transplant. 2015;50(2):209-215. 8. Munshi PN, Vesole D, Jurczyszyn A, et al. Age no bar: a CIBMTR analysis of elderly patients undergoing autologous hematopoietic cell transplantation for multiple myeloma. Cancer 2020;126(23): 5077-87. 9. Palumbo A, Bringhen S, Petrucci MT, et al. Intermediate-dose melphalan improves survival of myeloma patients aged 50 to 70: results of a randomized controlled trial. Blood. 2004;104(10):3052-3057. 10. Facon T, Mary JY, Hulin C, et al. Melphalan and prednisone plus thalidomide versus melphalan and prednisone alone or reducedintensity autologous stem cell transplantation in elderly patients with multiple myeloma (IFM 99-06): a randomised trial. Lancet. 2007;370(9594):1209-1218.

11. Muchtar E, Dingli D, Kumar S, et al. Autologous stem cell transplant for multiple myeloma patients 70 years or older. Bone Marrow Transplant. 2016;51(11):1449-1455. 12. Merz M, Neben K, Raab MS, et al. Autologous stem cell transplantation for elderly patients with newly diagnosed multiple myeloma in the era of novel agents. Ann Oncol. 2014;25(1):189-195. 13. Biran N, Jacobus S, Vesole DH, et al. Outcome with lenalidomide plus dexamethasone followed by early autologous stem cell transplantation in patients with newly diagnosed multiple myeloma on the ECOGACRIN E4A03 randomized clinical trial: long-term follow-up. Blood Cancer J. 2016;6(9):e466. 14. Wildes TM, Finney JD, Fiala M, et al. Highdose therapy and autologous stem cell transplant in older adults with multiple myeloma. Bone Marrow Transplant. 2015;50(8): 1075-1082. 15. Jackson GH, Davies FE, Pawlyn C, et al. Lenalidomide maintenance versus observation for patients with newly diagnosed multiple myeloma (Myeloma XI): a multicentre, open-label, randomised, phase 3 trial. Lancet Oncol. 2019;20(1):57-73. 16. Jackson GH, Davies FE, Pawlyn C, et al. Response-adapted intensification with cyclophosphamide, bortezomib, and dexamethasone versus no intensification in patients with newly diagnosed multiple

241

C. Pawlyn et al. myeloma (Myeloma XI): a multicentre, open-label, randomised, phase 3 trial. Lancet Haematol. 2019;6(12):e616-e629. 17. Royston P, Lambert PC. Flexible parametric survival analysis using stata: beyond the Cox model. College Station, Tex. Stata, 2011. 18. https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/lifeexpectancies/datasets/nationallifet ablesunitedkingdomreferencetables. (Last accessed Feb 2017). 19. Cook G, Royle KL, Pawlyn C, et al. A clinical prediction model for outcome and therapy delivery in transplant-ineligible patients with myeloma (UK Myeloma Research Alliance Risk Profile): a development and validation study. Lancet Haematol. 2019;6(3):e154-e166. 20. Boyle EM, Proszek PZ, Kaiser MF, et al. A molecular diagnostic approach able to detect the recurrent genetic prognostic factors typical of presenting myeloma. Genes Chromosomes Cancer. 2015;54(2):91-98. 21. Kaiser MF, Walker BA, Hockley SL, et al. A TC classification-based predictor for multiple myeloma using multiplexed real-time quantitative PCR. Leukemia. 2013;27(8): 1754-1757. 22. Boyd KD, Ross FM, Chiecchio L, et al. A

242

novel prognostic model in myeloma based on co-segregating adverse FISH lesions and the ISS: analysis of patients treated in the MRC Myeloma IX trial. Leukemia. 2012;26(2):349-355. 23. Sonneveld P, Avet-Loiseau H, Lonial S, et al. Treatment of multiple myeloma with highrisk cytogenetics: a consensus of the International Myeloma Working Group. Blood. 2016;127(24):2955-2962. 24. Badros A, Barlogie B, Siegel E, et al. Autologous stem cell transplantation in elderly multiple myeloma patients over the age of 70 years. Br J Haematol. 2001;114(3):600607. 25. Auner HW, Iacobelli S, Sbianchi G, et al. Melphalan 140 mg/m(2) or 200 mg/m(2) for autologous transplantation in myeloma: results from the Collaboration to Collect Autologous Transplant Outcomes in Lymphoma and Myeloma (CALM) study. A report by the EBMT Chronic Malignancies Working Party. Haematologica. 2018;103(3): 514-521. 26. Palumbo A, Bringhen S, Mateos MV, et al. Geriatric assessment predicts survival and toxicities in elderly myeloma patients: an International Myeloma Working Group report. Blood. 2015;125(13):2068-2074. 27. Engelhardt M, Domm AS, Dold SM, et al. A

concise revised Myeloma Comorbidity Index as a valid prognostic instrument in a large cohort of 801 multiple myeloma patients. Haematologica. 2017;102(5):910921. 28. Saad A, Mahindra A, Zhang MJ, et al. Hematopoietic cell transplant comorbidity index is predictive of survival after autologous hematopoietic cell transplantation in multiple myeloma. Biol Blood Marrow Transplant. 2014;20(3):402-408. 29. Attal M, Lauwers-Cances V, Hulin C, et al. Lenalidomide, bortezomib, and dexamethasone with transplantation for myeloma. N Engl J Med. 2017;376(14):1311-1320. 30. Facon T, Kumar S, Plesner T, et al. Daratumumab plus lenalidomide and dexamethasone for untreated myeloma. N Engl J Med. 2019;380(22):2104-2115. 31. Mateos MV, Dimopoulos MA, Cavo M, et al. Daratumumab plus bortezomib, melphalan, and prednisone for untreated myeloma. N Engl J Med. 2018;378(6):518-528. 32. Mateos MV, Cavo M, Blade J, et al. Overall survival with daratumumab, bortezomib, melphalan, and prednisone in newly diagnosed multiple myeloma (ALCYONE): a randomised, open-label, phase 3 trial. Lancet. 2020;395(10218):132-141.

haematologica | 2022; 107(1)

ARTICLE

Platelet Biology & its Disorders

Post-translational polymodification of b1-tubulin regulates motor protein localization in platelet production and function

Ferrata Storti Foundation

Abdullah O. Khan,1 Alexandre Slater,1 Annabel Maclachlan,1 Phillip L.R. Nicolson,1 Jeremy A. Pike,1,2 Jasmeet S. Reyat,1 Jack Yule,1,2 Rachel Stapley,1 Julie Rayes,1 Steven G. Thomas,1,2 and Neil V. Morgan1 Institute of Cardiovascular Sciences, College of Medical and Dental Sciences, University of Birmingham, Edgbaston, Birmingham and 2The Centre of Membrane Proteins and Receptors (COMPARE), University of Birmingham and University of Nottingham, Midlands, UK 1

Haematologica 2022 Volume 107(1):243-259

ABSTRACT

n specialized cells, the expression of specific tubulin isoforms and their subsequent post-translational modifications drive and coordinate unique morphologies and behaviors. The mechanisms by which b1-tubulin, the platelet and megakaryocyte (MK) lineage restricted tubulin isoform, drives platelet production and function remains poorly understood. We investigated the roles of two key post-translational tubulin polymodifications (polyglutamylation and polyglycylation) on these processes using a cohort of thrombocytopenic patients, human induced pluripotent stem cell derived MK, and healthy human donor platelets. We find distinct patterns of polymodification in MK and platelets, mediated by the antagonistic activities of the cell specific expression of tubulin tyrosine ligase like enzymes and cytosolic carboxypeptidase enzymes. The resulting microtubule patterning spatially regulates motor proteins to drive proplatelet formation in megakaryocytes, and the cytoskeletal reorganization required for thrombus formation. This work is the first to show a reversible system of polymodification by which different cell specific functions are achieved.

Correspondence: ABDULLAH O. KHAN a.khan.4@bham.ac.uk

Introduction Microtubules are large, cytoskeletal filaments vital to a host of critical functions including cell division, signaling, cargo transport, motility, and function.1,2 The question of how ubiquitously expressed filaments can facilitate complex and highly unique behaviors (such as neurotransmitter release and retinal organization) has been addressed by the concept of the tubulin code.3,4 This paradigm accounts for the specialization of microtubules and their organization by describing a mechanism in which particular cells express lineage restricted isoforms of tubulin. These cell specific isoforms are then subject to a series of post-translational modifications (PTM) which alter the mechanical properties of microtubules, and their capacity to recruit accessory proteins (e.g., motor proteins).1,2,4,5 A host of PTM have been reported in a range of cell types, including tyrosination, acetylation, glutamylation, glycylation, and phosphorylation. The loss of specific tubulin PTM has been linked to disease and dysfunction in motile and non-motile cilia (including respiratory cilia, retinal cells), spermatogenesis, muscular disorders, and neurological development.1,4,6–11 Despite an increasing understanding of the importance of tubulin PTM in disease, the role of the tubulin code in the generation of blood platelets from their progenitors, megakaryocytes (MK) remains poorly understood. Platelets are the smallest component of peripheral blood, and circulate as anucleate cells with an archetypal discoid shape maintained by a microtubule marginal band.12,13 Antagonistic motor proteins maintain the resting state of the marginal band, and during platelet activation a motor dependent mechanism results in sliding which extends the marginal band and causes the transition to a spherical shape.13–15 Conversely MK are the largest and rarest hematopoeitic cells of the bone marrow.

haematologica | 2022; 107(1)

NEIL V. MORGAN N.V.Morgan@bham.ac.uk. Received: August 28, 2020. Accepted: December 11, 2020. Pre-published: December 17, 2020. https://doi.org/10.3324/haematol.2020.270793

243

A.O. Khan et al.

These cells are characteristically large, polyploid cells with unique morphological structures (e.g., the invaginated membrane system [IMS]) required to facilitate the production of thousands of blood platelets and packaged within them the required pro-thrombotic factors.15,16 MK form long, beaded extensions into the lumen of bone marrow sinusoids - where these proplatelet extensions then experience fission under the flow of sinusoidal blood vessels which results in the release of barbell shaped pre-platelets and platelets into the blood stream.16 Both MK and platelets express a lineage-restricted isoform of b1-tubulin encoded by the gene TUBB1.17 In humans, TUBB1 mutations have been shown to result in impaired platelet production, with a resulting macrothrombocytopenia.18,19 More recently, a C-terminal truncation of b1-tubulin has been shown to cause a macrothrombocytopenia, suggesting that C-terminal modifications may be drivers of protein function and causative of the disease phenotype observed.20 While the loss of b1-tubulin is known to result in macrothrombocytopenia, the mechanisms by which this isoform of tubulin effects the dramatically different cytoskeletal behaviors of platelets and MK remains poorly understood. In the context of the tubulin code, MK and platelets present a particularly interesting model. Both cells express b1-tubulin, but undergo markedly different cytoskeletal changes. To date, acetylation and tyrosination have been the PTM primarily reported in MK and platelets, however neither modification is specific to the C-terminal tail encoded by TUBB1. Fiore et al. showed that a C-terminal truncation of b1-tubulin phenocopies the complete loss of the protein.20 We, therefore, hypothesize that PTM specific to the C-terminus of TUBB1 are required for the complex morphological rearrangements required for both MK and platelet function. The C-terminal tail of b1 tubulin is particularly rich in glutamate residues which are often targeted for two key PTM implicated in human disease. Polyglutamylation and polyglycylation are PTM which target glutamate residues on both tubulin subunits (a and b) and result in the addition of glutamate or glycine residues respectively.1,2,21 As both PTM target the same residue they are together referred to as polymodification. Polymodification has been observed in microtubules in centrioles, axenomes, neuronal outgrowths, and mitotic spindles.1,2 To date polyglycylation has not been reported in MK or platelets. Recently Van Dijk et al. reported on the polyglutamylation of b1 tubulin downstream of the integrin a2 b in a CHO cell line engineered to express TUBB1, murine MK, and platelets spread on fibrinogen.22 In our work, we report the effects of the loss of the C-terminus of b1-tubulin in patients with rare TUBB1 variants linked to the low platelet counts. Using induced pluripotent stem cell (iPSC) derived MK and CRISPR knockout (KO), we report a mechanism of polymodification which distinctly patterns b1tubulin in MK and platelets to spatially coordinate key motors. We describe the graded, cell specific expression of enzymes which mediate this, and finally report a novel gene, TTLL10, which is associated with excessive bleeding. b

Whole exome sequencing In order to identify the possible causative variants in these families we sequenced the whole exome of the affected individuals with the SureSelect human All Exon 50 Mb kit (Agilent Technologies) and sequenced on the HiSeq 2500 (Illumina) with 100 bp paired-end reads.

Stem cell culture and induced pluripotent stem cell megakayocyte differentiation Gibco human episomal iPSC line was purchased from Thermo Scientific and cultured on Geltrex basement membrane in StemFlex medium (Thermo Scientific). iPSC differentiation to mature, proplatelet forming MK was performed using a protocol based on work published by Feng et al.23 The IDT Alt-R®RNP system was used to target and knock out TUBB1. iPSC transfection was performed using Lipofectamine Stem (Life Technologies) according to the manufacturer instructions.

TUBB1 homology modeling Homology models of TUBB1 wild-type (WT) and variants were made using SWISS MODEL software,24–27 using the solved TUBB3 heterodimer as a template PDB: 5IJ028. TUBB1 and TUBB3 share approximately 80% sequence identity, and the model created corresponds to residues 1-425 of TUBB1. (Note that the Cterminal tail of TUBB1 is not presented in this model as there is no known or homologous structure available for this highly divergent sequence. The C-terminal portion of this model is used only as a means to visualize the effect of genetic variations on the C-terminal tail of TUBB1).

Statistical analysis Statistical analysis was performed using GraphPad PRISM 7. Specifics of each test are detailed in the figure legends of the relevant figures. P-values below 0.05 were considered significant. Detailed Materials and Methods can be seen in the Online Supplementary Appendix.

Methods Study approval Whole blood was obtained for each experiment from healthy

244

volunteers under the University of Birmingham’s ERN 11-0175 license ’The regulation of activation of platelets’. The Genotyping and Phenotyping of Platelets (GAPP) study was approved by the National Research Ethics Service Committee West Midlands – Edgbaston (REC reference 06/MRE07/36). Participants gave written informed consent in compliance with the Declaration of Helsinki. The GAPP study is included in the National Institute for Health Research Non-Malignant Hematology study portfolio (ID 9858), and registered at ISRCTN (http://www.isrctn.org) as ISRCTN77951167.

Results Identification and initial characterization of TUBB1 variants in patients with inherited thrombocytopenia and platelet dysfunction Using whole exome sequencing and Congenica Clinical genetic variant interpretation software29 of patients recruited to the GAPP study, two C-terminal TUBB1 variants were identified in unrelated families presenting with macrothrombocytopenia (Figure 1A; Online Supplementary Figure S1). Affected individuals in family A were found to be heterozygous for an arginine to tryptophan amino acid substitution (c.1075C>T, p.R359W) in TUBB1. Individuals in this family also carry a GFI1B variant (p.Cys168Phe). Variants in both genes have been linked to thrombocytopenia, however only individuals A:1 and A:3 (carrying haematologica | 2022; 107(1)

b1-tubulin role in platelet function

TUBB1 variants), present with a macrothrombocytopenia (107x109/L and 85x109/L respectively). Individual A:2 carries the GFI1B variant but is WT for TUBB1 and presents with a normal platelet count (221x109/L). Individuals A:1 and A:3 also present with significantly higher immature platelet fractions (IPF) and mean platelet volumes (MPV) when compared to their TUBB1 WT relatives (53.5% and 55.1% compared to 24.5%, MPV for A:1 and A:3 too large for measurement). This variation in count, IPF and platelet morphology in individuals with the TUBB1 R359W variant may suggest that the TUBB1 variant is linked to the macrothrombocytopenia observed. Family/patient B was an elderly gentleman (now deceased) with a G insertion and subsequent frameshift truncation of the b1-tubulin protein 19 amino acids from the site of insertion (c.1080insG, p.L361Afs*19).30,31 This patient had a severe thrombocytopenia with a platelet count of 11x109/L and an MPV above 13.4 (Figure 1B). At the time of study, IPF measurement was unavailable. p.L361Afs*19 is absent in the latest version of the gnomad database (accessed October 2020), while p.R359W is a rare variant with a frequency of 6.78x10-3 (gnomad accessed October 2020) with a significant pathogenicity prediction score of 25.4 using combined annotation dependent depletion (CADD), where a score of greater or equal to 20 indicates the 1% most deleterious sequence variants in the genome. Both variants were analyzed using in silico bioinformatic tools and were scored as ‘uncertain significance’ based on the current ACMG guidelines32 (Online Supplementary Figure S1). Both TUBB1 variants are positioned towards the C-terminal region of b1-tubulin as indicated in Figure 1C and D. This region is positioned away from the dimer interface with a-tubulin, and the glutamate rich C-terminal tail, not present in the model, is an established site for PTM (Figure 1C and D).1 Both affected TUBB1 sequence variants are highly conserved in mammals (Figure 1E). R359 is found within a loop region towards the C-terminus of tubulin, and we predict mutating this residue would not cause dramatic misfolding within the protein secondary structure. R359 does form direct polar contacts with the N-terminal helix and removing this contact through the sequence variant to the hydrophobic tryptophan may result in minor structural changes to the connected C-terminal regions, potentially affecting PTM or interactions with critical microtubule accessory proteins (MAP) (Figure 1C and D). Similarly, the G insertion and subsequent frameshift effectively deletes the C-terminus of the protein and would therefore, if the protein folds correctly, result in effects similar, or more extreme, to the substitution of R359. Patient B demonstrated a significant reduction in surface P-selectin expression and fibrinogen binding by flow cytometry in response to all agonists tested (Online Supplementary Figure S1). Family A showed no change in the levels of surface receptor expression, but weak Pselectin and fibrinogen responses when activated with a low concentration ADP, CRP, and PAR-1, suggesting a mild functional defect (Online Supplementary Figure S1C). Patient B also showed marked reduced expression of Pselectin and fibrinogen uptake in response to the same activation agonists (Online Supplementary Figure S1D and E). Patients with C-terminal variants in this study and others previously reported by Fiore et al. present with a macrothrombocytopenia also found in individuals with a complete loss of the b1-tubulin, suggesting that the C-terhaematologica | 2022; 107(1)

minal tail is likely critical to the function of TUBB1 in the roles of microtubules in MK and platelets.18–20 As this C-terminal tail is rich in glutamate residues which are often targeted for polymodification, we began to investigate the polymodification of b1-tubulin.

C-terminal variants of b1-tubulin fold correctly and demonstrate a reduction in polymodification First, we investigated the effects of the two patient variants (p.R359W and p.L361Afs*19) on the folding and potential polymodification of b1-tubulin. We designed, generated, and validated a b1-tubulin-mApple fusion construct (Online Supplementary Figure S2). This plasmid was further mutated to harbor each of the patient variants (p.R359W and p.L361Afs*19), and an artificial C-terminal truncation which specifically deletes the glutamate rich Cterminal tail (Figure 2A; Online Supplementary Figure S2). The WT b1-tubulin construct was transfected into Hek293T cells and co-stained for polyglutamylated and polyglycylated tubulin specifically. Transfected cells are exclusively positive for both residues indicating that b1tubulin is indeed polymodified (Figure 2B). Expression of each of the mutated constructs show that both patient variants and the C-terminal truncation fold correctly (Figure 2C). Furthermore each of them results in a consistent and significant reduction in polymodification (Figure 2D). This data indicates that both patient variants result in a correctly folded b1-tubulin which has a similar effect to a truncation of the C-terminus, and is further supported by western blotting of mutant constructs (Online Supplementary Figure S3) which show a significant reduction in polymodification.

Induced pluripotent stem cell-derived proplatelet forming megakaryocytes are both polyglycylated and polyglutamylated The tubulin code posits that a highly lineage and species specific expression of modifying enzymes mediates cell specific PTM. While our transfection data successfully indicates that b1-tubulin is indeed polymodified, and that each of our patient genetic variants result in the expression of a functional b1-tubulin with a C-terminal truncation, data from both human MK and platelets is needed to dissect the role of polymodification in these cells. To date, polyglycylation has not been reported in either platelets or MK. Polyglutamylation has recently been reported in a modified CHO cell line and human platelets. In order to investigate polymodification in human MK, we adapted a directed differentiation protocol previously reported by Feng et al. to generate large populations of mature, proplatelet forming cells (Online Supplementary Figure S4). iPSC-MK were stained for CD42b as a marker for mature, and hence TUBB1 expressing, MK, and both polyglutamylated and polyglycylated tubulin. CD42b+ cells were found to be positive for both polyglutamylated tubulin and polyglycylated tubulin (Figure 3A), while neighboring cells in the sample negative for CD42b did not demonstrate these polymodifications (Figure 3B). Across multiple differentiations we consistently yielded a purity of approximately 50-60% CD42b+ cells (Figure 3C), which on analysis are positive for both polyglutamylated and polyglycylated tubulin (Figure 3D). Finally, 100% of proplatelet forming cells observed across replicates were positive for both polymodifications (Figure 3E), and this was further confirmed by western blotting (Figure 3F). 245

A.O. Khan et al.

Figure 1. Candidate TUBB1 variants and their hypothesised effect on the C-terminus of b1-tubulin. (A and B) Two unrelated families were identified as carrying genetic variations in the TUBB1 gene within six base pairs of one another. The first, family A, is comprised of three individuals, two of whom carry an arginine to tryptophan (p.R359W) coding variant. Interestingly, all three individuals in family A harbor an additional GFI1B variant. However, the individuals with the reported TUBB1 variant (A:1 and A:3) present with a macrothrombocytopenia and high immature platelet fractions (IPF) , while the patient without the R359W TUBB1 variant presented with a normal platelet count. The second family is comprised of a single individual, recently deceased, with a frameshift variant six base pairs from the missense change reported in family A. In this individual’s case, the insertion of a guanine nucleotide results in a frameshift with a premature stop codon 19 amino acids from the leucine to alanine change. Platelet count normal range 147-327x109/L, (n=40), mean platelet volumes (MPV) normal range 7.8-12.69 fL (n=40), immature platelet fraction (IPF) normal range 1.3-10.8% (n=40). (C) The C-terminal tail is downstream of both genetic variants in these families, and projects away from the dimer:dimer interface. The C-terminal sequence of TUBB1 is rich in glutamate residues which can be targeted for polymodification. (D) Based on homology modeling of TUBB1, we predict that the missense variant reported in family A is likely to affect the fold of the C-terminal tail, while the frameshift causes a truncation of this region. (Black arrows indicate the position of the R359 residue). (E) The arginine residue substitution in family A is highly conserved across species, as are sequences adjacent to the frameshift in patient B.

246

haematologica | 2022; 107(1)

b1-tubulin role in platelet function

CRISPR knockout of TUBB1 results in a complete loss of proplatelet formation To date, the loss of TUBB1 has not been studied in human MK. We generated an iPSC line with a CRISPR mediated bi-allelic loss of function mutation in the N-terminus of the coding region of TUBB1 (Figure 3G; Online Supplementary Figure S5). The mutation of the TUBB1 start codon on both alleles results in a complete loss of expres-

sion (absence of T2 band in Figure 3H) and proplatelet formation in vitro (Figure 3I and J, right panel). Unfortunately, our attempts to generate C-terminal truncations through CRISPR in iPSC-MK were unsuccessful as multiple guides targeting the 3’ end of the TUBB1 gene failed to cleave the genomic sequence. Interestingly while TUBB1 KO clones stain positively for polyglutamylated and polyglycylated tubulin, the distribution of these residues is disturbed

Figure 2. Continued on following page.

haematologica | 2022; 107(1)

247

A.O. Khan et al.

Figure 2. Cells expressing wild-type and mutated b1-tubulin demonstrate C-terminal polymodification. (A) Constructs carrying wild-type (WT), patient variants (p.R359W and p.L361Afs*19), and a C-terminal tail truncation of b1-tubulin fused to the fluorescent reporter mApple at the N-terminus were designed and cloned. (B) The WT construct was first transfected into Hek293T cells, which were then fixed and immunostained for polyglutamylated and polyglycylated tubulin residues specifically. (C and D) A comparison of the WT and mutated constructs shows a reduction in polyglutamylation and polyglycylation in each mutant compared to the WT. (n=3 independent differentiations, standard deviation plotted on graphs. Two-way ANOVA with multiple comparisons performed to establish significance.)

when compared to WT platelet forming iPSC-MK (Figure 3K). While polyglycylated and polyglutamylated residues form a distinct peripheral band around WT MK as shown in Figure 3K, the KO cells demonstrate a diffuse tubulin staining (evidenced by line profiles in Online Supplementary Figure S7). Figure 3L shows quantification of this staining whereby WT iPSC-MK display higher mean intensity than KO constructs in polyglycylated tubulin, however no significant quantitative change in polyglutamylation was observed. This is the first data demonstrating the effects of the loss of TUBB1 expression in human MK.

Platelets demonstrate a different pattern of polymodification We hypothesized that the b1-tubulin polymodifications 248

evident in MK might be differently regulated between resting and activated platelets. We therefore compared immunofluorescence staining of polyglutamylated and polyglycylated tubulin between resting platelets and cells spread on fibrinogen and collagen. Resting platelets demonstrate polyglutamylated tubulin which partially co-localizes with the b1-tubulin marginal band unlike MK which demonstrate extensive polyglycylation in proplatelet forming cells (Figure 4A). On fibrinogen and collagen spreading, polyglutamylation is evident, notably at the marginal band of spreading cells on fibrinogen (Figure 4B). A lack of polyglycylation is consistent in both resting and activated platelets. Western blotting of resting platelets and cells activated through stimulation by CRP over time does not show an increase in the total haematologica | 2022; 107(1)

b1-tubulin role in platelet function

Figure 3. Continued on following page.

haematologica | 2022; 107(1)

249

A.O. Khan et al.

Figure 3. Mature, proplatelet forming induced pluripotent stem cell megacarycytes are both polyglutamylated and polyglycylated. TUBB1 knockout (KO) iPSC-MK do not form proplatelets and demonstrate disordered polymodified tubulin. (A) iPSC-MK co-stained for CD42 and polyglycylated or polyglutamylated tubulin show that these cells are positive for both polymodifications. Both polyglutamylated and polyglycylated tubulin are evident in proplatelet extensions, including nascent platelet swellings on the proplatelet shaft. (B) Neighboring CD42b- cells are negative for both polymodifications (indicated by red arrows). (C and D) Approximately 50-60% of cells in multiple differentiations are CD42b+, and these cells are 100% double positive for polymodification and CD42b+. (E) All proplatelet extensions observed are positive for polymodification. (F) Polyglutamylation and polyglycylation are evident by western blotting in mature iPSC-MK derived from three separate differentiations. (G) iPSC were transfected with a TUBB1 targeting guide RNA, after which indel positive cells were isolated and sequenced to positively identify a bi-allelic insertion-deletion mutant (clone T2). (H) This clone was further analyzed and loss of b1-tubulin expression was confirmed by quantitative real-time polymerase chain reaction (qRT-PCR). (I and J) A comparison of proplatelet production in wild-type (WT) vs. TUBB1 KO cells revealed a complete loss of proplatelet formation in mutant CD41/42b+ cells. (K) TUBB1 KO clones show a disordered arrangement of polymodified residues when compared to WT cells. KO cells do not demonstrate the reorganization of tubulin to the periphery of the cell evident in WT iPSC-MK (indicated by red arrows). (L) A measure of the polyglutamylation and polyglycylation in WT vs. KO clones reveals a significant increase in polyglycylation in mutant cells consistent with an aberrant accumulation of these residues. (n=3 independent differentiations, standard deviation plotted on graphs).

250

haematologica | 2022; 107(1)

b1-tubulin role in platelet function

amount of polyglutamylated tubulin (Figure 4C). An increase in the co-localization between b1-tubulin and polyglutamylated residues suggests that while the total polyglutamylation of platelets remains unchanged through activation, the distribution of these residues is altered (Figure 4D). We confirmed the polyglutamylation (and loss of polyglycylation) in microthrombi generated by CRP stimulation (Figure 4E to G). This data shows that while polyglutamylation and polyglycylation are evident in platelet producing iPSC-MK, polyglycylation is completely lost in mature platelets. This is the first evidence of a system whereby these competitive polymodifications are dramatically altered between the ’parent’ and terminal cell, suggesting markedly different roles for these residues in mediating the function and activity of b1-tubulin. Acetylation and tyrosination have been previously reported in platelets, however, their role in maintaining the marginal band and/or driving morphological change on

platelet activation remains unclear.13 In order to determine whether the polyglutamylation of the marginal band we observe thus far coincides with these PTM, we performed a time course of spreading on fibrinogen to determine whether there is an equivalent increase in either acetylation or tyrosination of the marginal band. Interestingly, we found a significant decrease in acetylation and tyrosination over time (between 0 and 10 minutes), while a notable polyglutamylation of the marginal band is evident from the earliest time point (10 minutes spreading on fibrinogen) and does not decrease (Figure 4H and I).

Platelet and megakaryocyte polymodifications regulate motor protein localization to drive both proplatelet formation and platelet shape change on activation Polyglutamylation has been reported as a means by which motor protein processivity is regulated, and like in neuronal cells, MK proplatelet formation is known to be

Figure 4 Continued on following page.

haematologica | 2022; 107(1)

251

A.O. Khan et al.

Figure 4. Platelet activation results in polyglutamylation of the marginal band. (A) Resting platelets show a partial polyglutamylation of the marginal band and a loss of the polyglycylation evident in induced pluripotent stem cell megacaryocytes (iPSC-MK). (B) Platelet spreading on fibrinogen and collagen shows an accumulation of polyglutamylation at the marginal band as platelets spread, and no evidence of polyglycylation. (C) Western blotting of resting and CRP activated platelets confirms the presence of polyglutamylated tubulin and a loss of polyglyclation. No increase in polyglutamylation is evident over time. (D, E) A measurement of colocalization (corrected Manders coefficient) between polyglutamylated tubulin and b1-tubulin in resting and spread platelets shows a significant increase in the colocalization of these modified tubulin residues on platelet activation and spreading. (F, G) Platelets activated in vitro using CRP were fixed and co-stained for b1-tubulin and polyglutamylated residues, and then imaged in 3D using AiryScan confocal (stacks colorized in Z as indicated by the color chart in this figure). In these microthrombi, polyglutamylation of the marginal band is evident, while polyglycylation is not observed. (H) A time course was performed to compare polyglutamylated tubulin with two other previously reported post-translational modifications (PTM) in platelets (acetylation and tyrosination). Polyglutamylation is maintained over time, while acetylation and tyrosination decrease significantly as platelets spread. (I) The mean fluorescence intensity of polyglutamylated tubulin is markedly higher than either acetylated or tyrosinated tubulin. (n=3, standard deviation. Two-Way ANOVA with multiple comparisons. 10 mm scale bar.)

driven by a mechanism of dynein mediated proplatelet sliding.33 Similarly, the antagonistic movement of dynein and kinesin are known to maintain the marginal band in resting platelets.13 In order to test if polymodification affects the spatial distribution of motors, we performed a time course of platelet spreading on fibrinogen and measured co-localization between polyglutamylated tubulin and dynein (DNAL1 - axonemal light chain 1). We observe a loss of co-localization between dynein and polyglutamylated residues upon platelet spreading (Figure 5A and B). Interestingly, axonemal dynein is also localized towards the leading edge of spread platelets (Figure 5A). This data suggests that the increased polyglutamylation of the marginal band observed on platelet spreading drives an outward movement of axonemal dynein. In order to investigate the role of axonemal dynein specifically in this process, we also stained spreading platelets for cytoplasmic dynein and observe a central distribution, suggesting an alternative role for cytoplasmic dynein in platelets (Online Supplementary Figure S6). The loss of co-localization between polyglutamylated tubulin and dynein was confirmed on both collagen and 252

fibrinogen spread platelets (Figure 5C and E). This spatial relationship was also observed between polyglutamylated tubulin and kinesin-1, a motor protein recently reported to be important in platelet secretion on platelet spreading (Figure 5D and F).34 This data suggests that polyglutamylated tubulin is involved in localizing motor proteins during platelet activation and spreading. The actions of kinesin and dynein are known to be critical to driving platelet shape change on activation, and this work is consistent with previous reports of polyglutamylated tubulin altering the processivity of motors in axons. We then investigated the role of these polymodifications on the distribution of motors in iPSC-MK. In proplatelet extensions axonemal dynein and kinesin-1 are both evident along the length of the proplatelet shaft (Figure 5G). b1-tubulin KO cells show no proplatelet formation, and a significant reduction in the co-localization of dynein with polyglutamyulated residues when compared to WT iPSC-MK (Figure 5H and I). No significant change in the co-localization of these residues with kinesin-1 is observed between WT and KO iPSC-MK (Figure 5J). haematologica | 2022; 107(1)

b1-tubulin role in platelet function

Megakaryocyte and platelet polymodification is regulated through the expression of both modifying and reversing enzymes We hypothesized that the expression of cell specific subsets of effecting (tubulin tyrosine ligase like [TTLL]) and reversing (cytosolic carboxypeptidase [CCP]) enzymes are required to achieve the distinctive polymodification we observe in MK and platelets. We designed a quantitative real-time polymerase chain reaction (qRT-PCR) panel to interrogate the expression of the 13 known mammalian TTLL and six CCP. We generated RNA from iPSC-MK at different stages of maturation (Figures 6A; Online Supplementary Figure S4A). Day 1 (d1) cells are representative of a pool of hematopoeitic stem cells (HSC) and MK progenitors, while day 5 (d5) cells are comprised of 60% CD41/42b+ cells (Figures 6A; Online Supplementary Figure

S4E). Finally, d5 cells treated with heparin to induce proplatelet formation (d5+Hep) were used to interrogate whether there is any specific upregulation of TTLL and/or CCP on proplatelet formation (Figures 6A; Online Supplementary Figure S4D). GAPDH housekeeping controls for each of the three samples (d1, d5, d5+Hep) show equivalent amplification of the housekeeping control, while results for TTLL family proteins show a number of enzymes expressed at different levels across the maturation of these cells (Figure 6B and C). Candidate TTLL observed in the initial endpoint PCR were taken forward for quantification across replicates generated from multiple differentiations (Figure 6C and D). We find a significantly increased expression of TTLL1, TTLL2, TTLL4, and TTLL10 on proplatelet formation in cells treated with heparin (**P=0.0081, *P=0.0105, *P=0.0260, ***P=

Figure 5. Continued on following page.

haematologica | 2022; 107(1)

253

A.O. Khan et al.

Figure 5. Polyglutamylation regulates spatial distribution of motor proteins in platelets and megakarycytes. (A and B) Quantification of the co-localization of dynein, as measured by the corrected Manders coefficient, with polyglutamylated tubulin over time shows a marked decrease in co-localization between these polymodified residues and the motor over time. (C) In fibrinogen and collagen spread cells, dynein is observed on the periphery of the spread cells, (D) while kinesin-1 is evident as a diffuse punctate stain. (E) Dynein co-localization with polyglutamylated residues decreases dramatically on platelet spreading in both fibrinogen and collagen (**P=0.0082 and ***P=0.0004 respectively). (F) A loss of co-localization with polyglutamylated tubulin is evident in fibrinogen and collagen spread cells (**P=0.0017, ****P<0.0001 respectively). (G) Immunofluorescence staining of induced pluripotent stem cell megacaryocytes (iPSC-MK) for polymodified tubulin, dynein, and kinesin-1 show the distribution of both motors along the length of the proplatelet shaft in wild-type (WT) cells. (H and I) In b1-tubulin knockout (KO) iPSCMK, no proplatelet extensions are formed and a significant reduction in the co-localization of dynein to polyglutamlated residues is observed (*P=0.0166 and *P=0.0293 respectively), (J) with no significant change in the co-localization of kinesin-1 with polyglycylated tubulin. (n=3, standard deviation. Two-Way ANOVA with multiple comparisons.)

0.0004 respectively) (Figure 6D). CCP family enzymes were also found to be expressed in maturing MK, notably CCP1, 3, 4, 5, and 6 (Figure 6E), with CCP4 and CCP6 upregulated on proplatelet production (*P=0.0130, ***P=0.0009) (Figure 6F to H). In order to investigate TTLL and CCP expression in platelets, we repeated this panel on RNA extracted from resting and CRP stimulated donor platelets. We found that 254

none of the TTLL and CCP observed in iPSC MK were consistently expressed across donors with the exception of TTLL7, a known polyglutamylase (Figure 6I, see complete gel in the Online Supplementary Figure S10).35 No differences between resting and activated platelets were observed (Figure 6I). This data shows a markedly different pattern of TTLL and CCP expression in both MK and platelets, correlating with the observed differences in polymodification. haematologica | 2022; 107(1)

b1-tubulin role in platelet function

Figure 6. Continued on following page.

haematologica | 2022; 107(1)

255

A.O. Khan et al.

256

Figure 6. Induced pluripotent stem cell megacaryocytes and platelets differentially express TTLL and CCP enzymes to regulate polymodifications during megakaryocyte maturation and platelet production. The loss of tubulin tyrosine ligase like (TTL) TTLL10 is possibly linked to a bleeding phenotype in unrelated human patients. (A) RNA was generated from induced pluripotent stem cell megacaryocytes (iPSC-MK) at different stages of terminal differentiation. Day 1 (d1) day 5 (d5) day 5 cells treated with heparin to induce proplatelet formation (d5+Hep) were used to determine whether an upregulation of these enzymes is evident on platelet production. (B) Samples were amplified with housekeeping GAPDH primers (C) A number of TTLL family enzymes were observed, including TTLL1, 2, 3, 4, 5, 6, 7, 10, and 13. Of these, a number appeared to be upregulated in mature and proplatelet forming cells. (D and E) These samples were taken forward and expression was quantified over multiple differentiations using the ΔΔCT method TTLL1, 2, 4, and 10 expression was found to be significantly upregulated on treatment with heparin (**P= 0.0081, *P= 0.0105, *P= 0.0260, ***P=0.0004 respectively). (F to H) A similar panel was performed on cytosolic carboxypeptidase (CCP) enzymes which reverse polymodifications, with expression of CCP1, 3, 4, 5, and 6 was observed. Statistically significant upregulation of CCP4, and CCP6 were observed on proplatelet production (*P=0.0130, ***P=0.0009). (I) In resting (-) and Creactive protein (CRP) activated (+) platelets from three healthy donors TTLL7 was the only modifying enzyme found to be consistently expressed. (J) Three unrelated families were identified within the Genotyping and Phenotyping of Platelets Study (GAPP) cohort, two with frameshift truncations and one with a missense substitution (p.Pro15Argfs*38, p.Val249Glyfs*57, and p.Arg340Trp respectively, and as shown on the protein schematic of TTLL10). (K) All three families present with normal platelet counts and function, but demonstrate an elevated mean platelet volume (MPV) and consistent histories of bleeding including cutaneous bruising and menorrhagia. (L) Patient A:1:1 was recalled and demonstrated abnormally large platelets on spreading on fibrinogen when imaged by widefield and single molecule localisation microscopy (tubulin staining, 10 µm scale bar, 5 µm in cropped image). (n=3, standard deviation. Two-Way ANOVA with multiple comparisons performed. Complete unedited gels found in the Online Supplementary Figures S9 and S10).

haematologica | 2022; 107(1)

b1-tubulin role in platelet function

Figure 7. A system of competitive polymodification of TUBB1 driven by the expression of TTLL and CCP enzymes is required for platelet production and function. We observe a system by which as induced pluripotent stem cell megacaryocytes (iPSC-MK) mature and express TUBB1, they acquire both polyglutamylated and polyglycylated tubulin which co-incides with an increase in the expression of glutamylating and glycylating tubulin tyrosine ligase like (TTL) enzymes and reversing cytosolic carboxypeptidase (CCP) enzymes. A resting platelet is partially polyglutamylated, and on activation the marginal band is further polyglutamylated to drive shape change and spreading through the action of TTLL7. We show that this polymodification spatially affects the position of key motor proteins.

TTLL10 variants maybe associated with moderate to severe bleeding in three unrelated families Our qRT-PCR screen revealed that a number of TTLL and CCP are upregulated during the process of platelet production, including the polyglycylase TTLL10. Whole exome sequencing data from patients recruited to the GAPP study identified three unrelated families with rare and novel variants in the TTLL10 gene (Figure 6J; Online Supplementary Figures S12 and S13). Extensive in silico analysis was undertaken on the TTLL10 variants which were predicted as ‘uncertain significance’ using current ACMG guidelines (Online Supplementary Figures S12 and S13).32 Two of the three variants result in frameshifts towards the N-terminus of the protein, preceding the ATP binding region (p.Pro15Argfs*38 [novel] and p.Val249Glyfs*57 [frequency 2.15x10-3]) (Figure 6J). The final family has a missense p.Arg340Trp variant (3.34x10≠5 frequency). All three families report normal platelet counts, aggregation and secretion, and present with an established history of moderate to severe bleeding, including cutaneous bruising and menorrhagia (Figure 6K). Family A demonstrated a consistently high MPV (normal ranges MPF [fL] (7.83-10.5), while families B and C do not. Interestingly, one of the patients (A 1:1) was re-recruited, and on platelet spreading on fibrinogen coated coverslips we observed a marked increase in platelet area compared to controls when imaged using widefield and single molecule localisation microscopy (SMLM) (Figure 6L).

Discussion We hypothesized that a system of polymodification (polyglutamylation and polyglycylation) targeting the glutamate rich C-terminus of b1-tubulin isoform and analogous to similar PTM demonstrated in cilia and neuronal cells, is a likely mechanism by which the interactions of b1-tubulin with key motors are regulated in both MK and platelets. We report that mature and proplatelet forming CD42b+ iPSC-MK demonstrate both polyglutamylation and polyglycylation (polymodification). We observe a markedly difhaematologica | 2022; 107(1)

ferent distribution of these PTM in the resting platelet, where polyglycylation is lost and polyglutamylation is partially co-localized to the marginal band. On platelet activation, we observe a marked change in the localization of polyglutamylation specific to the marginal band. In iPSCMK with a CRISPR KO of TUBB1, we see a complete loss of proplatelet formation and lose the distinct reorganization of polyglutamylated and polyglycylated tubulin around the periphery of MK as seen in WT cells. MK proplatelet extensions are known to be driven by a system of dynein mediated microtubule sliding, while the marginal band in a resting platelet has been shown to be maintained by the antagonistic movement of dynein and kinesin.13,14,36 Interestingly, polyglutamylation has been reported as a mechanism of altering motor protein processivity, with in vitro assays suggesting that polyglutamylation of b1 tubulin isoforms like TUBB1 and TUBB3 accelerates these motors.28 We show a significant effect of polyglutamylation on the spatial localization of dynein and kinesin, supporting in vitro assays which suggest that polyglutamylation is an accelerator of motor proteins. We report two unrelated patient families with rare TUBB1 C-terminal variants (an R359W missense and a L361Afs*19 frameshift linked to macrothrombocytopenia). Interestingly within family A, a second genetic variant in the GFI1B gene, which directly affects the TUBB1 promoter is observed. Individuals within family A positive for the GFI1B variant but WT for TUBB1 have normal platelet counts and a milder increase in MPV and IPF, indicating that the macrothrombocytopenia within the family is possibly linked to the TUBB1 variant, but suggesting a potential additive role for the GFI1B variant which is a focus for future study. Both the clinical phenotype and laboratory investigations points to variable expressivity of the two genetic variants contributing to the overall phenotypes observed. We go on to show through the expression of these variants in Hek293T cells that each variant results in a dysfunctional b1-tubulin protein which is most likely to mirror the effects of a C-terminal truncation. The system of polymodification evidenced in iPSC-MK 257

A.O. Khan et al.

is analogous to the PTM of ciliated cells, and so we reasoned that axonemal dynein, an isoform of the motor exclusive to axonemes, may play a role in both platelet formation and activation.6,37,38 We find evidence of axonemal dynein on both proplatelet extensions and at the leading edge of spreading platelets. To our knowledge this is the first evidence of a functional role of axonemal dynein outside of classical ciliated structures. In our TUBB1 KO MK, we observe a decrease in the co-localization of dynein to polyglutamyulated tubulin, suggesting that the loss of proplatelet formation observed in these cells is due to a dysregulation of the dynein-mediated microtubule sliding known to drive the elongation of the proplatelet shaft.39 The similarities between flagella and MK proplatelet extensions have been discussed previously, Italiano et al. observe structures similar to flagella in taxol treated MK, an idea which led to the microtubule telescoping experiment reported by Patel et al.39,40 Our data suggests a tightly regulated, reversible system of polymodification which must be mediated by the cell specific expression of TTLL and CCP. Our expression profiles show a number of TTLL and CCP are expressed by MK, while only the polyglutamylase TTLL7 is expressed by platelets consistent with the different patterns of polymodification oberved in these cells. In MK we find expression of two TTLLs known to be involved in glycylation - the initiase TTLL3 and the elongase TTLL10, with a significant increase in the expression of TTLL10 on platelet production. Our findings support a role of TTLL10 as a polyglycylase on co-expression with TTLL3 as reported by Ikegami et al. in cell lines through co-transfection experiments (Figure 7). Finally, we report a novel gene, TTLL10, in three unrelated families with excessive bleeding. We identify three unrelated families with TTLL10 variants which result in an increase in an established history of bleeding which provides an invaluable insight to the potential role of polyglycylation in the context of platelet production and function. Our data shows that both TTLL3 and TTLL10 are expressed in platelet producing MK. Our patient cohort do not lose TTLL3 function, and as such the action of TTLL3 as an initiase will occupy glutamate residues which would otherwise be polyglutamylated. In these patients we likely see a loss of polyglycylation, but no coincident increase in polyglutamylation due to the normal function of the initiase (TTLL3). As polyglutamylation and monoglycylation are unaffected, platelet counts (and production) are normal, however, affected individuals appear to have an increased platelet volume and bleeding, suggesting a role for the extended glycine tail in regulating platelet size, with a downstream effect on the ability of platelets to prevent bleeding.

References 1. Wade RH. On and around microtubules: an overview. Mol Biotechnol. 2009;43(2):177191. 2. Gadadhar S, Dadi H, Bodakuntla S, et al. Tubulin glycylation controls primary cilia length. J Cell Biol. 2017; 216(9):2701-2713. 3. Ludueña RF. A hypothesis on the origin and evolution of tubulin. Int Rev Cell Mol Biol. 2013;302:41-185. 4. Magiera Maria M, Singh Puja, Janke Carsten.

258

Patel et al. describe a system by which continuous polymerization is key to the dynein-mediated sliding required for platelet production.39 This work, alongside reports of tyrosination, acetylation, and polyglutamylation in different MK models, suggests that tubulin PTM are likely part of a highly dynamic system in which polymerizing microtubules are subject to modifications which coordinate platelet production and packaging. Future work will interrogate the interplay between different PTM, and how they fit into a highly dynamic landscape of MT polymerization and function. Indeed, in our KO TUBB1 cells we do not observe a complete loss of polymodification, suggesting that compensatory mechanisms which target other isoforms of tubulin may come into play. There is likely a more extensive effect on the expression of tubulins and their interplay with other cytoskeletal proteins which should be a focus of future work. Similarly, the role of these PTMs will need to be further validated in primary human CD34-derived MK. This work supports the paradigm of a ’tubulin code’ and the importance of microtubule patterning in healthy, and thus diseased, MK and platelets. We provide novel insights into the mechanisms by which b1-tubulin functions in these unique cells. Disclosures No conflicts of interest to disclose. Contributions AOK and NVM devised and performed experiments; AOK and NVM wrote the manuscript; AS performed homology modeling, cloning, transfection, and platelet spreading experiments; AM performed platelet spreading and analysis of patient sequencing data; PLRN performed platelet preparations and reviewed the manuscript; JAP developed and applied image analysis workflows; JSR assisted with western blotting; JY contributed to platelet preparations; all authors reviewed the manuscript. Acknowledgments We thank the families for providing samples and our clinical and laboratory colleagues for their help. The authors would like to thank the TechHub and COMPARE Core facilities at the University of Birmingham. AOK is a Wellcome funded Sir Henry Wellcome Fellow (218649/Z/19/Z). We thank Professor Steve Watson for his ongoing support and invaluable mentorship. The authors do not have any conflict of interest. Funding This work was supported by the British Heart Foundation (PG/13/36/30275; FS/13/70/30521; FS/15/18/31317; PG/16/103/32650; IG/18/2/33544).This research was funded in whole or in part by the Wellcome Trust [218649/Z/19/Z].

SnapShot: functions of tubulin posttranslational modifications. Cell. 2018;173(6):15521552.e1. 5. Gadadhar S, Bodakuntla S, Natarajan K, Janke C. The tubulin code at a glance. J Cell Sci. 2017;130(8):1347-1353. 6. Wloga D, Joachimiak E, Louka P, Gaertig J. Posttranslational modifications of tubulin and cilia. Cold Spring Harb Perspect Biol. 2017;9(6):a028159. 7. Reiter JF, Leroux MR. Genes and molecular pathways underpinning ciliopathies. Nat Rev Mol Cell Biol. 2017;18(9):533-547.

8. Bosch Grau Montserrat, Masson Christel, Gadadhar Sudarshan, et al. Alterations in the balance of tubulin glycylation and glutamylation in photoreceptors leads to retinal degeneration. J Cell Sci. 2017;130(5):938949. 9. Ikegami K, Sato S, Nakamura K, Ostrowski LE, Setou M. Tubulin polyglutamylation is essential for airway ciliary function through the regulation of beating asymmetry. Proc Natl Acad Sci U S A. 2010;107(23):1049010495. 10. Wu HY, Wei P, Morgan JI. Role of cytosolic

haematologica | 2022; 107(1)

b1-tubulin role in platelet function

carboxypeptidase 5 in neuronal survival and spermatogenesis. Sci Rep. 2017;7:41428. 11. Vogel P, Hansen G, Fontenot G, Read R. Tubulin tyrosine ligase-like 1 deficiency results in chronic rhinosinusitis and abnormal development of spermatid flagella in mice. Vet Pathol. 2010;47(4):703-712. 12. Dmitrieff S, Alsina A, Mathur A, Nédélec FJ. Balance of microtubule stiffness and cortical tension determines the size of blood cells with marginal band across species. Proc Natl Acad Sci U S A. 2017;114(17):4418-4423. 13. Diagouraga B, Grichine A, Fertin A, Wang J, Khochbin S, Sadoul K. Motor-driven marginal band coiling promotes cell shape change during platelet activation. J Cell Biol. 2014;204(2):177-185. 14. Sadoul K. New explanations for old observations: marginal band coiling during platelet activation. J Thromb Haemost. 2015;13 (3):333-346. 15. Poulter Natalie S, Thomas SG. Cytoskeletal regulation of platelet formation: coordination of F-actin and microtubules. Int J Biochem Cell Biol. 2015;66:69-74. 16. Machlus KR, Italiano JE. The incredible journey: from megakaryocyte development to platelet formation. J Cell Biol. 2013;201(6): 785-796. 17. Schwer HD, Lecine P, Tiwari S, Italiano JE, Hartwig JH, Shivdasani RA. A lineagerestricted and divergent beta-tubulin isoform is essential for the biogenesis, structure and function of blood platelets. Curr Biol. 2001;11(8):579-586. 18. Kunishima S, Kobayashi R, Itoh Tomohiko J, Hamaguchi Mo, Saito H. Mutation of the beta1-tubulin gene associated with congenital macrothrombocytopenia affecting microtubule assembly. Blood. 2009;113(2):458461. 19. Kunishima S, Nishimura S, Suzuki H, Imaizumi M, Saito H. TUBB1 mutation disrupting microtubule assembly impairs proplatelet formation and results in congenital macrothrombocytopenia. Eur J Haematol. 2014;92(4):276-282. 20. Fiore M, Goulas C, Pillois X. A new mutation in TUBB1 associated with thrombocytopenia confirms that C-terminal part of b1tubulin plays a role in microtubule assembly. Clin Genet. 2017;91(6):924-926. 21. Natarajan K, Gadadhar S, Souphron J, Magiera MM, Janke Carsten. Molecular interactions between tubulin tails and glutamylases reveal determinants of glutamylation patterns. EMBO Rep. 2017;18(6):1013-1026. 22. Dijk J, Bompard G, Cau J, et al. Microtubule polyglutamylation and acetylation drive microtubule dynamics critical for platelet formation. BMC Biol. 2018;16(1):116. 23. Feng Q, Shabrani N, Thon JN, et al. Scalable generation of universal platelets from

haematologica | 2022; 107(1)

human induced pluripotent stem cells. Stem Cell Rep. 2014;3(5):817-831. 24. Waterhouse Andrew, Bertoni Martino, Bienert Stefan, et al. SWISS-MODEL: homology modelling of protein structures and complexes. Nucleic Acids Res. 2018;46(W1):W296-W303. 25. Bienert S, Waterhouse A, Beer TAP, et al. The SWISS-MODEL Repository—new features and functionality. Nucleic Acids Res. 2016;45(D1):D313-D319. 26. Guex N, Peitsch MC, Schwede T. Automated comparative protein structure modeling with SWISS-MODEL and SwissPdbViewer: a historical perspective. Electrophoresis. 2009;30:S162-S173. 27. Benkert P, Biasini M, Schwede T. Toward the estimation of the absolute quality of individual protein structure models. Bioinformatics. 2010;27(3):343-350. 28. Ti SC, Pamula MC, Howes SC, et al. Mutations in human tubulin proximal to the kinesin-binding site alter dynamic instability at microtubule plus- and minus-ends. Dev Cell. 2016;37(1):72-84. 29. Almazni I, Stapley RJ, Khan AO, Morgan NV. A comprehensive bioinformatic analysis of 126 patients with an inherited platelet disorder to identify both sequence and copy number genetic variants. Hum Mutat. 2020;41(11):1848-1865. 30. Johnson B, Lowe GC, Futterer J, et al. Whole exome sequencing identifies genetic variants in inherited thrombocytopenia with secondary qualitative function defects. Haematologica. 2016;101(10):1170-1179. 31. Khan AO, Stapley R, Pike JA, et al. Novel gene variants in patients with platelet-based bleeding using combined exome sequencing and RNAseq murine expression data. J Thromb Haemost. 2021;19(1):262-268. 32. Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med. 2015;17(5):405-424. 33. Bender M, Thon JN, Ehrlicher AJ, et al. Microtubule sliding drives proplatelet elongation and is dependent on cytoplasmic dynein. Blood. 2015;125(5):860-868. 34. Adam F, Kauskot A, Kurowska M, et al. Kinesin-1 Is a new actor involved in platelet secretion and thrombus stability. Arterioscler Thromb Vasc Biol. 2018;38(5): 1037-1051. 35. Ikegami K, Mukai M, Tsuchida J, Heier RL, Macgregor GR, Setou M. TTLL7 is a mammalian beta-tubulin polyglutamylase required for growth of MAP2-positive neurites. J Biol Chem. 2006;281(41):30707-30716. 36. Sirajuddin M, Rice LM, Vale RD. Regulation

of microtubule motors by tubulin isotypes and post-translational modifications. Nat Cell Biol. 2014;16(4):335-344. 37. O’Hagan R, Silva M, Nguyen Ken CQ, et al. Glutamylation regulates transport, specializes function, and sculpts the structure of cilia. Curr Biol. 2017;27(22):3430-3441. 38. Wloga D, Dave D, Meagley J, Rogowski K, Jerka-Dziadosz M, Gaertig J. Hyperglutamylation of tubulin can either stabilize or destabilize microtubules in the same cell. Eukaryot Cell. 2010;9(1):184-193. 39. Patel R, Richardson Jennifer L, Schulze Harald, et al. Differential roles of microtubule assembly and sliding in proplatelet formation by megakaryocytes. Blood. 2005;106(13):4076-4085. 40. Italiano JE, Lecine P, Shivdasani R A, Hartwig J H. Blood platelets are assembled principally at the ends of proplatelet processes produced by differentiated megakaryocytes. J Cell Biol. 1999;147(6): 1299-1312. 41. Kremers G, Hazelwood KL, Murphy CS, Davidson MW, Piston DW. Photoconversion in orange and red fluorescent proteins. Nat Methods. 2009;6(5):355358. 42. Khan AO, Simms VA, Pike JA, Thomas SG, Morgan NV. CRISPR-Cas9 mediated labelling allows for single molecule imaging and resolution. Sci Rep. 2017;7(1):8450. 43. Khan AO, White CW, Pike JA, et al. Optimised insert design for improved singlemolecule imaging and quantification through CRISPR-Cas9 mediated knock-in. Sci Rep. 2019;9(1):14219. 44. Smith CW, Raslan Z, Parfitt L, et al. TREMlike transcript 1: a more sensitive marker of platelet activation than P-selectin in humans and mice. Blood Adv. 2018;2(16):2072-2078. 45. Pike JA, Simms VA, Smith CW, et al. An adaptable analysis workflow for characterization of platelet spreading and morphology. Platelets. 2020:1-5. 46. Berthold MR, Cebron N, Dill F, et al. KNIME-the konstanz information miner: version 2.0 and beyond. AcM SIGKDD explorations Newsletter. 2009;11:26-31. 47. Sommer C, Strahle C, Kothe F, A. Hamptrecht F. ilastik: Interactive Learning and Segmentation Toolkit. In: Eighth IEEE International Symposium on Biomedical Imaging (ISBI). Proceedings. 2011;230-233. 48. Khan AO, Maclachlan A, Lowe GC, et al. High-throughput platelet spreading analysis: a tool for the diagnosis of platelet-based bleeding disorders. Haematologica. 2020;105(3):e124-e128. 49. McDonald JH, Dunn KW. Statistical tests for measures of colocalization in biological microscopy. J Microsc. 2013;252(3):295302.

259

ARTICLE Ferrata Storti Foundation

Platelet Biology & its Disorders

Dysregulation of oncogenic factors by GFI1B p32: investigation of a novel GFI1B germline mutation Michela Faleschini,1 Nicole Papa,1 Marie-Christine Morel-Kopp,2 Caterina Marconi,3 Tania Giangregorio,1 Federica Melazzini,4 Valeria Bozzi,4 Marco Seri,3 Patrizia Noris,4 Alessandro Pecci,4 Anna Savoia1,5 and Roberta Bottega1 Institute for Maternal and Child Health – IRCCS Burlo Garofolo, Trieste, Italy; Department of Hematology and Transfusion Medicine, Royal North Shore Hospital and Northern Blood Research Center, Kolling Institute, University of Sydney, Sydney, New South Wales, Australia; 3Department of Medical and Surgical Sciences, University of Bologna, Bologna, Italy; 4Biotechnology Research Laboratories, IRCCS Policlinico San Matteo Foundation, Pavia, Italy and 5Department of Medical Sciences, University of Trieste, Trieste, Italy 1

Haematologica 2022 Volume 107(1):260-267

ABSTRACT

FI1B is a transcription factor essential for the regulation of erythropoiesis and megakaryopoiesis, and pathogenic variants have been associated with thrombocytopenia and bleeding. Analysing thrombocytopenic families by whole exome sequencing, we identified a novel GFI1B variant (c.648+5G>A), which causes exon 9 skipping and overexpression of a shorter p32 isoform. We report the clinical data of our patients and critically review the phenotype observed in individuals with different GFI1B variants leading to the same effect on the p32 expression. Since p32 is increased in acute and chronic leukemia cells, we tested the expression level of genes playing a role in various type of cancers, including hematological tumors and found that they are significantly dysregulated, suggesting a potential role for GFI1B in carcinogenesis regulation. Increasing the detection of individuals with GFI1B variants will allow us to better characterize this rare disease and determine whether it is associated with an increased risk of developing malignancies.

Correspondence: ANNA SAVOIA anna.savoia@burlo.trieste.it Received: July 31, 2020. Accepted: December 21, 2020. Pre-published: January 21, 2021. https://doi.org/10.3324/haematol.2020.267328

260

Introduction Inherited thrombocytopenias represent a group of heterogeneous rare disorders characterized by a reduced platelet count that may associate with other defects.1 Mutations in at least 40 different genes are responsible for these disorders, including a recently discovered autosomal dominant condition reported as GFI1B-associated bleeding disorder or platelet-type bleeding disorder 17 (OMIM 187900), which is caused by pathogenic variants of the GFI1B (growth factor-independent 1B) gene.2–4 GFI1B is a transcription repressor of the GFI zinc-finger family, consisting of three domains: a highly conserved N-terminal repressor domain called Snail/Gfi1, which recruits chromatin regulatory proteins; an intermediary domain of unknown function; and a C-terminal cluster of six zinc-finger domains (ZNF), of which ZNF 3, 4, and 5 bind to DNA, while ZNF 1,2, and 6 have yet to be characterized.5,6 GFI1B plays a key role in hematopoiesis through different isoforms. Indeed, in addition to generating a long isoform consisting of 330 amino acids (p37), GFI1B mRNA undergoes an alternative splicing process producing a short isoform of 284 residues (p32) characterized by the skipping of exon 9 (NG_034227) and consequent removal of ZNF 1 and 2. The long p37 form is reported to have a pivotal role in megakaryopoiesis and platelet production, whereas the short p32 form is essential for erythroid lineage.7 GFI1B regulates the differentiation of the hematopoietic stem cells through the repression of different promoters, including its self-regulatory regions via an autoregulatory feedback mechanism.7–9 Among other targets is the hematopoietic stem cell marker CD34, whose downregulation during hematopoiesis is not quenched when GFI1B is mutated. Therefore, one feature of the GFI1B-associated bleeding disorder is the aberrant expression of CD34 on patients' platelets, which is due to the dysreg-

haematologica | 2022; 107(1)

Novel GFI1B variant dysregulates oncogenic factors

ulation of the respective gene at the transcriptional level.10 Since its first description,2 individuals with GFI1B mutations have been examined in order to characterize this novel rare platelet disorder. To our knowledge, these studies have identified at least 15 different mutations in approximately 20 unrelated families, allowing better characterization of the disease, whose features are increased bleeding tendency, thrombocytopenia with enlarged platelets and a reduced a-granule content with abnormal distribution.2,3,10–17 We report a novel heterozygous GFI1B variant (c.648+5G>A) in a family with mild thrombocytopenia and no other significant features. The substitution causes skipping of exon 9 and consequent overexpression of the short p32 isoform and dysregulation of GFI1B target genes, including CD34 and other genes that are involved in neoplastic transformation, suggesting a potential role for GFI1B in carcinogenesis regulation.

Methods

Foundation” of Pavia approved the study. All subjects provided written informed consent for the study, which was conducted in accordance with the Declaration of Helsinki.

Mutation screening and reverse transcriptase polymerase chain reaction analysis Mutational screening was performed by whole exome sequencing (WES) in the proband (II-2), as previously reported.20 Variants were confirmed by Sanger sequencing in her daughters (III-1 and III-2), sisters (II-3 and II-4) and brother (II5) (Table 1). Total RNA was extracted from peripheral blood cells of patients II-2 and III-1, as well as two healthy controls and cDNA amplified as previously reported,21 using primers enlisted in the Online SupplementaryTable S1.

Bioinformatic analysis The effect of the splice-site mutation was predicted by in silico analyses using two dedicated bioinformatic tools: splice site prediction by Neural Network (NNSplice; http://www.fruitfly.org/seq_tools/splice.html) and Human Splicing Finder Version 2.4.1 (http://www.umd.be/HSF/).

Family study and clinical features The propositus (Figure 1), a 65-year-old female with low platelet number, and her family members were studied to determine the molecular basis of thrombocytopenia. Their clinical features, as well as the methods used for blood cell analyses,18,19 are described in more detail in the Online Supplementary Appendix. The Institutional Review Board of the IRCCS “Policlinico San Matteo

Expression vectors and dual-luciferase reporter assay Wild-type and mutant (c.511_648del due to skipping of exon 9) GFI1B cDNA were cloned into the tagged (myc) expression vector pcDNA3.1 (Invitrogen, Carlsbad, CA, USA). The GFI1B, GFI, MEIS1, and CD34 promoters were cloned into a reporter firefly luciferase vector (pGL4/luc2, Promega). Plasmids were transiently

Figure 1 Genetic analyses of the family. (A) Sanger sequencing showing exon 9 (e9) and intron 9 (i9) boundary of GFI1B gene. The heterozygous c.648+5G>A substitution is indicated by an arrow. Nucleotide A of the ATG translation initiation start site of the GFI1B cDNA in GenBank sequence NM_004188.6 is indicated as nucleotide +1. (B) Real-time polymerase chain reaction showing different transcription pattern between patients (II-2; III-1) and healthy control (HC). The expected product of 246 basepairs (bp) was detected in the HC sample; an additional band of 108 bp corresponding to skipping of exon 9 was found in the affected individuals. M: molecular weight. (C) Family pedigree showing an autosomal dominant pattern of inheritance. Arrow indicates the proband whereas black filled symbols represent affected individuals carrying the mutation. The grey filled symbol indicates a thrombocytopenic brother without the mutation (platelet count in this patient may be influenced by a liver cirrhosis). wt: wild-type.

haematologica | 2022; 107(1)

261

M. Faleschini et al.

co-transfected in Meg01, as previously described.2 The luciferase activity was detected using the Dual Luciferase reporter assay system kit (Promega), according to the manufacturer’s instructions. Details of the experiments are included in the Online Supplementary Appendix.

Quantitative real-time polymerase chain reaction Quantitative real-time polymerase chain reaction (qRT-PCR) was performed to evaluate the absolute expression of GFI1B isoforms and relative expression of GFI1B major target, using specific primers (Online Supplementary Table S2). FastStart Universal SYBR Green Master Mix (Roche) and ABI PRISM 7900 detection (Applied Biosystem, Foster City, CA, USA) were used. All experiments were performed three times in triplicate, as previously reported.22

Western blot analyses HEK293T cells were transfected with myc-tagged wild-type and mutant GFI1B cDNA using calcium phosphate method and maintained in DMEM medium (Euroclone, Pero, Italy) with 10% fetal bovine serum. Protein fractionated cell extracts were analyzed by western blot using primary antibodies anti-myc (9E10; sc-40; Santa Cruz Biotechnology, Dallas, TX, USA), anti-HSP90 (sc-7947; Santa Cruz Biotechnology), anti-ORC2 (ab68348; Abcam, Cambridge, UK); as previously described.23

Immunofluorescence assay GFI1B transfected in HeLa cells was detected using 9E10 antibody against c-myc (Santa Cruz Biotechnology). Peripheral blood smears were double-labeled with antibodies against CD41 (Santa Cruz Biotechnology) and CD34 (Miltenyi Biotec, Bergisch Gladbach, Germany). General staining and image acquisition have previously been described.24,25 Detailed methods for both immunofluorescence analyses are included in the Online Supplementary Appendix.

Results Identification of the novel c.648+5G>A mutation in GFI1B WES of proband II-2 revealed a single nucleotide heterozygous substitution (c.648+5 G>A) in the GFI1B gene, affecting the splice donor site of intron 9 (Figure 1A). Sanger sequencing confirmed the variant in the proband (II-2), her sister (II-3) and her daughters (III-1 and III-2) but not in her

healthy sister (II-5) (Figure 1B). The proband’s brother (II-4) was homozygous for the wild-type allele and the thrombocytopenia was likely due to liver cirrhosis. The c.648+5G>A substitution is reported in GnomAD with a microcytic anemia factor (MAF) value of 0.000008025 and identified in two European (non-Finnish) individuals in heterozygous status. Based on NNSplice software, the c.648+5G>A variant is predicted to drop the score of the splice donor site from 0.89 to 0.24, suggesting potential alternative splicing processes. Therefore, RT-PCR analysis was performed using RNA extracted from affected individuals’ peripheral blood cells (II-2 and III-1) (Figure 1C). In contrast with the healthy control showing the expected product of 246 basepairs (bp), the patients’ samples amplified an additional band of 108 bp. Sequencing analysis revealed that the lower band corresponded to a transcript characterized by skipping of exon 9, resulting in an in-frame mRNA deletion of 138 bp (p.Val171_Gln216del). The alternative splicing is expected to produce a shorter protein of 32kDa, lacking zinc fingers 1 and 2, corresponding to what was previously described as p32.7,26

Blood cell studies in family members The major blood parameters of the family members are reported in Table 1. The propositus had moderate thrombocytopenia while the other individuals carrying the heterozygous c.648+5G>A GFI1B mutation had a platelet count only slightly lower or higher than the lower limit of the normal range. Platelet size calculated as mean platelet volume (MPV) by the automated counter was normal in all subjects. The study of myeloproliferative disorders (MPD) by image analysis of blood smears revealed slightly increased values, indicating mild platelet macrocytosis. The remaining parameters, including hemoglobin concentration, mean red cell volume (MCV) and red blood cell count were within the normal ranges in all investigated subjects. Examination of peripheral blood smears revealed a mild reduction in platelet a-granules and red blood cell anisocytosis in all affected family members. Flow cytometry revealed normal expression of the platelet surface GP complexes Ia-IIa, IIb-IIIa, and Ib-IX-V in individuals II-2 and III1 (data not shown). Platelet aggregation after stimulation with collagen (4 and 20 mg/mL), ADP (5 and 20 mM,) epinephrine (10 mM), arachidonic acid (1 mM), TRAP (25 mM), and ristocetin (1.5 mg/mL) was normal in individuals II-2

Table 1. Clinical and laboratory parameters in GFI1B mutated and non-mutated family members

Subject (Sex) II-2 (F) II-3 (F) III-1 (F) III-2 (F) II-4 (M) II-5 (F) Normal values

GFI1B mutational status

Platelet count (x109/L)

MPV (fL)

MPD (mm)

Hb (g/dL)

MCV (fL)

648+5G>A/+ 648+5G>A/+ 648+5G>A/+ 648+5G>A/+ +/+ +/+

50 149 124 153 94* 215 150-450

10.7 9.2 12.6 11.7 10.8 10 8-13**

3.32 2.95 2.96 3.57 3.09 3.26 2.4-2.7†

13.6 13.4 12 13.9 15.6 14.6 11.7-15.5 (F) 13.2-17.3 (M)

85 90.6 82.3 88.3 98 95.1 82-98

RBC RCDW (%) Leukocyte (x106/mL) count (x109/L) 4.89 14 4.40 13.7 4.10 13.4 4.80 15 4.92 13.9 4.58 14 3.8-5.2 (F) 11.6-16 4.4–5.7 (M)

6.23 4.01 7.62 9.1 5.7 5.42 4-10

WHO bleeding score§ 3 3 0 0 0 0 0

£ F: female; M: male; * in this patient thrombocytopenia was likely due to liver cirrhosis; **MPV range given by the automated cell counter; †MPD obtained in 50 healthy volunteers (95% confidence interval)18; §World Health Organization (WHO) bleeding scale: 0 - no bleeding tendency; 1 - cutaneous bleeding only (including minimal mucosal bleeding); 2 - mild blood loss (any mucosal bleeding not fulfilling the criteria for grade 1 or 3); 3 - gross blood loss, requiring transfusion; 4 - debilitating blood loss (including retinal or cerebral associated with fatality). See patients and methods for details about bleeding symptoms. MPV: mean platelet volume; MDP: myeloproliferative disorders; Hb: hemoglobin; RBC: red blood cell.

262

haematologica | 2022; 107(1)

Novel GFI1B variant dysregulates oncogenic factors

and III-1 carrying the c.648+5G>A variant, and similar to II4 who is wild-type (data not shown).

Expression level of alternative spliced products The patient’s cDNA was used to clone the wild-type (330 amino acids [aa]; p37) and the exon 9 skipped (284 aa; p32) forms of GFI1B tagged to myc into an expression vector. Accordingly to what was observed during the RT-

PCR on patients RNA (Figure 1C), colony screening using primers on exons 8 and 10 (Online Supplementary Table S1) detected the expected products of 246 bp and of 108 bp, as well as two additional bands of 312 bp and 174 bp (Figure 2A). Sanger sequencing showed that the additional products retained the last 66 bp (gggatcccggccgggtccagtcctgagcctgcacctgaccccccggggcctcatttcctccggcag) of intron 9 in both p37 and p32 forms due to the recogni-

Figure 2. GFI1B splicing isoforms. (A) Polymerase chain reaction (PCR) products obtained after amplification of a portion of GFI1B from different plasmids containing subcloned GFI1B cDNA (individual II-2) using primer aligning on exons 8 and 10. Fragments corresponding to the wild-type (246 basepairs [bp]) and skipped exon 9 (108 bp) forms of GFI1B are shown in lanes 1 and 4, and 2 and 5, respectively. Additional fragments of 312 bp (lane 7) and 174 bp (lanes 3 and 6) corresponding to p39 and p34, respectively, were also detected. M: molecular weight. (B) Schematic structure of the genomic GFI1B gene (gDNA) which includes the coding exons (exons 6-11; NM_004188.6). On the bottom, the representation of the full cDNA (p37) and the different isoforms identified. p32 is characterized by skipping of exon 9. p39 and p34 results in p37 and p32, respectively, which also retain the last 66 bp of intron 9 (of which the first and the last are listed in detail) due to recognition of a cryptic acceptor splice site “ag” (in bold). Arrows represent primers used for the amplification. (C) Differential expression level of the four GFI1B isoforms between patients (II-2 and III-1) and healthy controls (HC) perfomed by quantitative PCR using specific primers. Overall the wild-type isoform (p37) is the most represented in controls while p32 is significantly increased in patients (*P<0.05). Error bars represent the standard deviation of three independent experiments. Statistical analysis was performed using t-test. GDNA: genomic DNA; cDNA: coding DNA; aa: amino acids.

haematologica | 2022; 107(1)

263

M. Faleschini et al.

tion of a cryptic acceptor splicing site (ggagtgtcctgttccgcaggggat) with a score of 0.83 predicted by NNSpice tool (Figure 2B). Therefore, the two additional products would correspond to GFI1B molecules of an expected molecular weight of 39 kDa (p39) and 34 kDa (p34), respectively for an in-frame insertion of 22 aa (p.216_217insGIPAGSSPEPAPDPPGPHFLRQ). We investigated the expression level of the four forms of GFI1B (Online Supplementary Table S2). In control samples (n=2), forms p37 and p32 represented the most expressed products, being approximately 83% and 10% of the total GFI1B cDNA, respectively; the other two forms, p34 and p39, were overlooked (Figure 2C). In patients, the expression level of p37 and p32 was significantly different than in controls, corresponding to 54% and 32% of the total cDNA, respectively. Of note, p32 was relatively less expressed than p37, suggesting that c.648+5G>A leads to partial skipping of exon 9 or partial degradation of the mRNA. Finally, no significant difference in the expression of p39 and p34 was observed between patient and control samples.

Pathogenic role of the c.648+5G>A mutation leading to p32 In order to investigate the effect of c.648+5G>A on GFI1B cellular localization, we performed immunofluorescence and western blot analysis in cells transiently transfected with the wild-type or mutant cDNA. Like p37, the p32 isoform enters the HeLa cell nucleus (Figure 3A) and both forms are similarly distributed in the cytoplasmic and nuclear fractions of Hek293 cells (Figure 3B), suggesting that zinc fingers 1 and 2 are dispensable for migration of p32 into the nucleus. Moreover, we determined the effect on the transcriptional activity, using the luciferase gene under the control of three known GFI1B target gene promoters (MEIS, GFI, and GFI1B itself). The luciferase activity was significantly reduced when the Meg-01 cells were co-transfected with myc-tagged p37, confirming the role of GFI1B as a transcription repressor of those three targets.27 On the contrary, the expression of p32 not only abolished the repression but also increased the transcriptional activity of those three promoters, suggesting that the functional defect is likely due to

Figure 3. Pathogenic role of GFI1B p32. (A) Immunofluorescence (IF) and (B) western blot (WB) of nuclear (N) and cytoplasmatic (C) fractionated cellular lysates from Hek293 transiently transfected with wild-type (WT) (p37) and mutated (p32) GFI1B expression vector. Results demonstrate that both isoforms are distributed in the nucleus as well as in the cytoplasm with the same proportions. In IF, propidium iodide (PI) was used to detect nuclei. HSP90 and ORC2 antibodies were used in WB as loading marker for the cytoplasmic and nuclear fraction, respectively. (C) Transcriptional activity of p37 (WT) and the variant p32 GFI1B isoforms measured on GFI, GFI1B and MEIS promoters in Meg01 cells. Results are represented by ratio of the promoter signal (Firefly) and the control promoter (Renilla) and expressed as fold change in luciferase activity. p37 shows the expected repressive activity on targets compared to the empty vector while p32 is an activator (1- to >1.5-fold) in comparison with the p37 for all the three targets (***P<0.001). Error bars represent the standard deviation of three independent experiments. Statistical analysis was performed using t-test.

264

haematologica | 2022; 107(1)

Novel GFI1B variant dysregulates oncogenic factors

a dominant negative effect of the exogenous p32 on the p37 endogenous transcription factor in Meg-01 cells (Figure 3C). Of note, similar data were obtained using non-tagged p37 and p32 forms generated to test the effect of a different mutation (c.648+1_648+8delGTGGGCAC) of GFI1B.10

Transcriptional regulation of the CD34 promoter reflects the aberrant expression of CD34 in patients' platelets Previous investigations reported that GFI1B-related thrombocytopenia is associated with aberrant expression of the stem cell antigen CD34 in platelets and megakaryocytes, which could be caused by the loss of CD34 downregulation during megakaryocytopoiesis due to defective GFI1B function.10 We therefore investigated the transcriptional effect of p32 on the CD34 promoter. Similarly to what observed on MEIS, GFI, and GFI1B promoters, p32 does not repress the CD34 promoter activity (P<0.001) (Figure 4A). Consistent with our in vitro findings, immunofluorescence analysis of peripheral blood smears confirmed the aberrant expression of the CD34 antigen in the proband’s platelets (II-2) (Figure 4B).

p32 dysregulates the transcription of genes involved in oncogenic pathways Since myeloid and lymphoid leukemia cells express higher level of p32 than control cells,26 we hypothesized that p32 could unbalance the transcription of genes involved in malignant transformation. For this reason, we investigated the expression level of other GFI1B targets, including protooncogenes GFI1 and MYC, antiapoptotic factors TGFBR3 and BCL2L1, and hematopoietic master regulators, the GATA family members, whose dysregulation is known to occur in oncogenesis. qPCR performed on patients’ peripheral blood RNA showed that except for GATA2, all the

other genes tested (GFI1, BCL2L1, TGFBR3, MYC, GATA1 and GATA3) were significantly overexpressed in patients II2 and III-1 compared to controls with a fold-change ranging from 1.5 to > 3 (Figure 5), suggesting that patients carrying the c.648+5G>A mutation could have an increased risk for malignancies. In order to confirm the role of p37 and p32 in regulating the oncogenic factors, qPCR was also carried out in HEK cells overexpressing the two GFI1B isoforms. As in patients' peripheral blood cells, the overexpression of p32 is associated with an increased expression level of the oncogenes, indicating that p32 up-regulates their transcription (Online Supplementary Figure S1). The only exception is represented by GATA 3, whose expression was significantly decreased in p32-overexpressing cells instead of being increased as in patients’ samples, suggesting that p32 downregulates this gene at least in the HEK cellular model.

Discussion GFI1B is a transcriptional repressor that plays a fundamental role in megakaryocyte development and erythropoiesis9 and pathogenic variants are responsible for the GFI1B-associated bleeding disorder. The novel heterozygous GFI1B splicing variant (c.648+5G>A) identified in our family causes alternative splicing leading to unbalanced expression level of the GFI1B isoforms. Specifically, we detected four alternative splicing events, leading to the three known isoforms, p37, p32 and p39, as well as the novel p34 isoform. The p37 and p32 isoforms are expressed at different level in the individuals carrying the c.648+5G>A substitution, with the p32/p37 ratio ten times higher in affected than healthy individuals, indicating that this unbalance, due to the presence of the c.648+5G>A mutation,

Figure 4. CD34 expression in c.648+5G>A patients. (A) Transcriptional activity of p37 (wild-type) and the variant p32 GFI1B isoforms measured on the CD34 promoter after transient transfection in Meg01 cells. Results are represented by the ratio of the promoter signal (Firefly) and the control promoter (Renilla) and expressed as fold change in luciferase activity. Significative variations occur on CD34 transcriptional levels as a significant p32 impaired repression activity is detected in comparison to p37 active repression (***P<0.001). Error bars represent the standard deviation of three independent experiments. Statistical analysis was performed using t-test. (B) Peripheral blood slides of the proband III-2 and healthy volunteers were double-labeled for CD41 (red) and CD34 (green). Platelets were identified by morphology and CD41 expression. Platelets of the proband III-2 showed a clear aberrant expression of the CD34 antigen compared to healthy control (HC). Scale bars correspond to 5 mm.

haematologica | 2022; 107(1)

265

M. Faleschini et al.

could be responsible for thrombocytopenia. In fact, considering that p37 has a pivotal role in megakaryopoiesis and platelet production and that p32 is essential for erythroid lineage, distortion of the p32/p37 ratio is likely to dysregulate hemopoiesis. On the contrary, the role of p39 and p34, which represent approximately 3% and 10%, respectively of the total GFI1B mRNA in both specimens of control samples, has not yet been defined.7 We further investigated the role of p37 and p32 in terms of cellular localization and ability to control the transcription of targeted genes. In overexpressed conditions, most of the two isoforms are in the nucleus at comparable level, though their transcriptional effect is opposite. Whereas p37 represses the activity of the luciferase that is under the control of GFI1B target promoters, such as MEIS1, GFI1, and GFI1B itself, p32 increased its activity, reversing the transcriptional repression. The same effect of c.648+5G>A on alternative splicing has previously been reported for c.648+1_648+8del GTGGGCAC, another mutation of GFI1B that was found in members of one family affected with MYH9related disease.10 As in our family, in individuals carrying the GFI1B but not the MYH9 variant, the expression level of p32 was markedly increased, being in at least equal amounts to p37. However, these individuals were not thrombocytopenic (mean value of 228x109/L). Another GFI1B variant, associated with skipping of exon 9 and significant reduction of the p37 expression level, is the synonymous substitution c.576C>T(p.Phe192=) in exon 9, which is reported as rs150813342 with a MAF of 0.0041.28 Indeed, in individuals heterozygous for c.576C>T the platelet count is reduced with an average of 25– 30x109/L in comparison with controls. Finally, unbalance between p32 and p37 was identified in association with another mutation (c.551insG/p.Ser185Leufs*3).29 In patients homozygous for this mutation, while the expression level of p37 was significantly decreased, likely due to degradation by nonsense-mediated decay, the expression of p32 was the

same level as healthy controls. Therefore, in these patients the active form of GFI1B is mainly p32. Of note, whereas individuals homozygous for c.551insG had a severe phenotype, heterozygous individulas were clinically unaffected. Of note, in these patients, as well as in our affected individuals, erythropoiesis was not defective, except for mild anisocytosis, suggesting that red cell production is rather independent of the p32/p37 ratio.29 Taken together all these observations suggested that the expressivity is extremely variable even when the variants exert the same effect. Indeed, bleeding and platelet count vary significantly, from individuals without defects (c.648+1_648+8delGTGGGCAC) or slight reduction of platelet count (c.576C>T) to those with an intermediate phenotype characterized by mild thrombocytopenia with mild bleeding tendency as in our family (c.648+5G>A) or with severe disease when the c.551insG mutation is homozygous.29 Of note, abnormalities in megakaryocyte maturation were observed in isogenic cell clones homozygous for rs150813342 and in hematopoietic stem cell progenitors in which GFI1B p37 was selectively silenced,28 suggesting that both expressivity and penetrance of the platelet phenotype might depend - at least in the cases enlisted above - on a specific threshold of the p32/p37 ratio, a value that cannot be estimated from the literature or our own data. One feature that is likely to be pathognomonic of the GFI1B-associated bleeding disorder is the abnormal expression of CD34 on the platelet surface.10 Indeed, platelets of affected individual III-2, as well as those mentioned above with an altered p32/p37 ratio, express the CD34 antigen. Strong expression has also been reported in association with the Cys168Phe, Arg184Pro, and His181Tyr mutations, all affecting ZNF 1, or H294fsX307, Gln287* and G272fsX2 removing ZNF 5 and 6.3,10–12 Changes in proportion between p37 and p32 significantly increasing p32 levels has been reported in acute and chronic leukemia, which could be interpreted as a triggering event for carcinogenesis.26 This hypothesis is supported by the

Figure 5. Quantitative polymerase chain reaction of GFI1B transcriptional target genes. The GFI, BCL2L1, TGFbR3, MYC, GATA1, and GATA3 mRNA expression levels were significantly increased in patients (II-2 and III-1) compared to controls (two unrelated healthy individuals [HC]) (fold change from 1.5 to >3). No significant difference was observed for GATA2. Error bars represent the standard deviation of three independent experiments. Statistical analysis was performed using t-test.

266

haematologica | 2022; 107(1)

Novel GFI1B variant dysregulates oncogenic factors

significant upregulation of GFI1, BCL2L1, TGFBR3, MYC, GATA1, and GATA3 in cells from the patients' peripheral blood. These genes are also differentially expressed in HEK cells overexpressing p32 or p37, confirming that p32 controls their transcription. Considering that these oncogenes are involved in hematopoietic differentiation and are dysregulated in various types of cancer, including hematological malignancies,27 it would be interesting to establish whether the GFI1B-associated bleeding disorder could be part of the nosological entity called “Inherited thrombocytopenia-associated genes with predisposition to neoplasms”.30 Increasing the detected number of individuals with GFI1B mutations and creating a registry of individuals with GFI1B variants, including a comprehensive medical and family history and regular follow-up, would be of fundamental importance in defining whether there is any increased risk of developing such hematological malignancies. Therefore, an effort should be made to identify patients with this rare form of thrombocytopenia by evaluating the CD34 expression on platelets, a useful assay in

References 1. Noris P, Pecci A. Hereditary thrombocytopenias: a growing list of disorders. Hematology Am Soc Hematol Educ Program. 2017;2017(1):385-399. 2. Stevenson WS, Morel-Kopp MC, Chen Q, et al. GFI1B mutation causes a bleeding disorder with abnormal platelet function. J Thromb Haemost. 2013;11(11):2039-2047. 3. Monteferrario D, Bolar NA, Marneth AE, et al. A dominant-negative GFI1B mutation in the gray platelet syndrome. N Engl J Med. 2014;370(3):245-253. 4. Rabbolini DJ, Morel-Kopp MC, Ward CM, Stevenson WS. GFI1B variants associated with thrombocytopenia. Platelets. 2017;28 (5):525-527. 5. Zweidler-Mckay PA, Grimes HL, Flubacher MM, Tsichlis PN. Gfi-1 encodes a nuclear zinc finger protein that binds DNA and functions as a transcriptional repressor. Mol Cell Biol. 1996;16(8):4024-4034. 6. Vassen L, Fiolka K, Mahlmann S, Möröy T. Direct transcriptional repression of the genes encoding the zinc-finger proteins Gfi1b and Gfi1 by Gfi1b. Nucleic Acids Res. 2005;33(3):987-998. 7. Laurent B, Randrianarison-Huetz V, Frisan E, et al. A short Gfi-1B isoform controls erythroid differentiation by recruiting the LSD1-CoREST complex through the dimethylation of its SNAG domain. J Cell Sci. 2012;125(Pt 4):993-1002. 8. Khandanpour C, Sharif-Askari E, Vassen L, et al. Evidence that growth factor independence 1b regulates dormancy and peripheral blood mobilization of hematopoietic stem cells. Blood. 2010;116(24):5149-5161. 9. Randrianarison-Huetz V, Laurent B, Bardet V, Blobe GC, Huetz F, Duménil D. Gfi-1B controls human erythroid and megakaryocytic differentiation by regulating TGFbeta signaling at the bipotent erythromegakaryocytic progenitor stage. Blood. 2010;115(14):2784-2795. 10. Rabbolini DJ, Morel-Kopp MC, Chen Q, et al. Thrombocytopenia and CD34 expression is decoupled from a-granule deficiency with

haematologica | 2022; 107(1)

defining the pathogenicity of variants that would otherwise be considered of uncertain significance. Disclosures No conflicts of interest to disclose. Contributions MF, NP, MCMK, and RB performed research; CM, TG, and MS performed data analysis; FM, PN and AP collected clinical data; MF, SA, and RB designed the study and wrote the manuscript; VB performed blood cell studies. All contributors have read and approved the manuscript. Acknowledgement The authors would like to thank patients for their participation in this project and Prof. Gughi D.L. for the constructive discussions. Funding This study was supported by IRCCS “Burlo Garofolo” (Ricerca Corrente 01/2018) and AIRC Grant IG-21974.

mutation of the first growth factor-independent 1B zinc finger. J Thromb Haemost. 2017;15(11):2245-2258. 11. van Oorschot R, Marneth AE, Bergevoet SM, et al. Inherited missense variants that affect GFI1B function do not necessarily cause bleeding diatheses. Haematologica. 2019;104(6):e260-e4. 12. Kitamura K, Okuno Y, Yoshida K, et al. Functional characterization of a novel GFI1B mutation causing congenital macrothrombocytopenia. J Thromb Haemost. 2016;14 (7):1462-1469. 13. Ferreira CR, Chen D, Abraham SM, et al. Combined alpha-delta platelet storage pool deficiency is associated with mutations in GFI1B. Mol Genet Metab. 2017;120(3):288294. 14. Uchiyama Y, Ogawa Y, Kunishima S, et al. A novel GFI1B mutation at the first zinc finger domain causes congenital macrothrombocytopenia. Br J Haematol. 2018;181(6):843847. 15. Chen L, Kostadima M, Martens JHA, et al. Transcriptional diversity during lineage commitment of human blood progenitors. Science. 2014;345(6204):1251033. 16. Cheng AN, Bao EL, Gughi DL, Fiorini C, Sankaran VG. Macrothrombocytopenia associated with a rare GFI1B missense variant confounding the presentation of immune thrombocytopenia. Pediatr Blood Cancer. 2019;66(9):e27874. 17. Johnson B, Lowe GC, Futterer J, et al. Whole exome sequencing identifies genetic variants in inherited thrombocytopenia with secondary qualitative function defects. Haematologica. 2016;101(10):1170-1179. 18. Noris P, Biino G, Pecci A, et al. Platelet diameters in inherited thrombocytopenias: analysis of 376 patients with all known disorders. Blood. 2014;124(6):e4-e10. 19. Noris P, Guidetti GF, Conti V, et al. Autosomal dominant thrombocytopenias with reduced expression of glycoprotein Ia. Thromb Haemost. 2006;95(3):483-489. 20. Marconi C, Di Buduo CA, Barozzi S, et al. SLFN14-related thrombocytopenia: identification within a large series of patients with inherited thrombocytopenia. Thromb

Haemost. 2016;115(5):1076-1079. 21. Bottega R, Pecci A, De Candia E, et al. Correlation between platelet phenotype and NBEAL2 genotype in patients with congenital thrombocytopenia and a-granule deficiency. Haematologica. 2013;98(6):868-874. 22. De Rocco D, Melazzini F, Marconi C, et al. Mutations of RUNX1 in families with inherited thrombocytopenia. Am J Hematol. 2017;92(6):E86-E8. 23. Bottega R, Nicchia E, Cappelli E, et al. Hypomorphic FANCA mutations correlate with mild mitochondrial and clinical phenotype in Fanconi anemia. Haematologica. 2018;103(3):417-426. 24. Faleschini M, Melazzini F, Marconi C, et al. ACTN1 mutations lead to a benign form of platelet macrocytosis not always associated with thrombocytopenia. Br J Haematol. 2018;183(2):276-288. 25. Balduini CL, Pecci A, Loffredo G, et al. Effects of the R216Q mutation of GATA-1 on erythropoiesis and megakaryocytopoiesis. Thromb Haemost. 2004;91(1): 129-140. 26. Vassen L, Khandanpour C, Ebeling P, et al. Growth factor independent 1b (Gfi1b) and a new splice variant of Gfi1b are highly expressed in patients with acute and chronic leukemia. Int J Hematol. 2009;89(4):422-430. 27. Anguita E, Candel FJ, Chaparro A, Benet M, Roldán-Etcheverry JJ. Transcription factor GFI1B in health and disease. Front Oncol. 2017;7:54. 28. Polfus LM, Khajuria RK, Schick UM, et al. Whole-exome sequencing identifies loci associated with blood cell traits and reveals a role for alternative GFI1B splice variants in human hematopoiesis. Am J Hum Genet. 2016;99(3):785. 29. Schulze H, Schlagenhauf A, Manukjan G, et al. Recessive grey platelet-like syndrome with unaffected erythropoiesis in the absence of the splice isoform GFI1B-p37. Haematologica. 2017;102(9):e375-e378. 30. Arber DA, Orazi A, Hasserjian R, et al. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood. 2016;127(20):2391-2405.

267

ARTICLE Ferrata Storti Foundation

Haematologica 2022 Volume 107(1):268-283

Red Cell Biology & its Disorders

The epigenetic regulator RINF (CXXC5) maintains SMAD7 expression in human immature erythroid cells and sustains red blood cell expansion Audrey Astori,1,2,3* Gabriel Matherat,1,2,3* Isabelle Munoz,1,2,3 Emilie-Fleur Gautier,1,2,3 Didier Surdez,3,4,5 Yaël Zermati,1,2 Frédérique Verdier,1,2 Sakina Zaidi,3,4,5 Vincent Feuillet,1 Amir Kadi,1 Evelyne Lauret,1,3 Olivier Delattre,3,4,5 Carine Lefèvre,1,2 Michaela Fontenay,1,2,6 Evelyne Ségal-Bendirdjian,7 Isabelle Dusanter-Fourt,1,3 Didier Bouscary,1,3 Olivier Hermine,2,3,8 Patrick Mayeux1,2,3 and Frédéric Pendino1,2,3 Université de Paris, Institut Cochin, INSERM, CNRS; 2Laboratory of Excellence GR-ex; Equipe Labellisée Ligue Nationale Contre le Cancer (LNCC); 4PSL Research University, Institut Curie Research Center, INSERM U830; 5SIREDO: Care, Innovation and Research for Children, Adolescents and Young Adults with Cancer, Institut Curie; 6Service d'Hématologie Biologique, Hôpital Cochin, Assistance Publique-Hôpitaux de Paris, Centre-Université de Paris; 7INSERM UMR-S 1124, Team: Cellular Homeostasis Cancer and Therapies, Université de Paris and 8Université de Paris, Institut Imagine, INSERM, CNRS, Paris, France. 1 3

*AA and GM contributed equally as co-first authors.

ABSTRACT

Correspondence: FRÉDÉRIC PENDINO frederic.pendino@inserm.fr Received: June 18, 2020. Accepted: November 16, 2020. Pre-published: November 26, 2020. https://doi.org/10.3324/haematol.2020.263558

268

he gene CXXC5, encoding a retinoid-inducible nuclear factor (RINF), is located within a region at 5q31.2 commonly deleted in myelodysplastic syndrome and adult acute myeloid leukemia. RINF may act as an epigenetic regulator and has been proposed as a tumor suppressor in hematopoietic malignancies. However, functional studies in normal hematopoiesis are lacking, and its mechanism of action is unknown. Here, we evaluated the consequences of RINF silencing on cytokine-induced erythroid differentiation of human primary CD34+ progenitors. We found that RINF is expressed in immature erythroid cells and that RINF-knockdown accelerated erythropoietin-driven maturation, leading to a significant reduction (~45%) in the number of red blood cells, without affecting cell viability. The phenotype induced by RINF-silencing was dependent on tumor growth factor b (TGFb) and mediated by SMAD7, a TGFb-signaling inhibitor. RINF upregulates SMAD7 expression by direct binding to its promoter and we found a close correlation between RINF and SMAD7 mRNA levels both in CD34+ cells isolated from bone marrow of healthy donors and myelodysplastic syndrome patients with del(5q). Importantly, RINF knockdown attenuated SMAD7 expression in primary cells and ectopic SMAD7 expression was sufficient to prevent the RINF knockdown-dependent erythroid phenotype. Finally, RINF silencing affects 5’-hydroxymethylation of human erythroblasts, in agreement with its recently described role as a TET2-anchoring platform in mouse. Collectively, our data bring insight into how the epigenetic factor RINF, as a transcriptional regulator of SMAD7, may fine-tune cell sensitivity to TGFb superfamily cytokines and thus play an important role in both normal and pathological erythropoiesis.

Introduction In earlier work, we demonstrated that RINF loss-of-function affects granulopoiesis in normal and tumoral human hematopoietic cells.1 In keeping with its gene location at chromosome 5q31.2 (Online Supplementary Figure S1), a commonly deleted region associated with high risk in myeloid neoplasms,2 RINF loss of expression has been

haematologica | 2022; 107(1)

RINF maintenance of SMAD7 sustains human erythropoiesis

proposed to contribute to the development or progression of (pre)leukemia, such as myelodysplastic syndrome (MDS).1,3-5 We and others have demonstrated that RINF mRNA expression is an unfavorable6,7 and independent3 prognostic factor in acute myeloid leukemia (AML) as well as in solid tumors.8,9 However, its role in normal hematopoiesis has hitherto been poorly investigated and its contribution to the erythroid lineage and red blood cell (RBC) expansion is unknown. RINF contains a nuclear localization signal that has been functionally validated10 and, in most studies, its subcellular localization is reported to be mainly or exclusively nuclear and it acts as a transcriptional cofactor.1,3,8,10-17 RINF associates strongly with chromatin1 through its conserved zincfinger domain (CXXC) which plays an essential role in providing the capacity to bind CpG islands.18,19 Interestingly, this domain is almost identical to the one harbored by TET1 and TET3, two epigenetic modulators involved in the erasure of DNA-methylation marks,20 pointing to the possibility that RINF might interfere with TET activities, hydroxymethylation, and gene transcription, as recently demonstrated in mice.12,15 RINF has also been reported to bind ATM, mediate DNA-damage-induced activation of TP533,10 and inhibit the WNT-b-catenin signaling pathway3,21-24 through a cytoplasmic interaction with disheveled proteins DVL and DVL2.21 Transforming growth factor b (TGFb) is a powerful and widespread cell growth inhibitor in numerous mammalian tissues.25,26 In the hematopoietic system, TGFb is known to regulate hematopoietic stem and progenitor cells (HSPC) and is also described as a potent inducer of erythroid differentiation and inhibitor of cell proliferation.27-30 TGFb signals through cell surface serine/threonine kinase receptors, mainly TGFbRI and TGFbRII. Activated TGFbRI phosphorylates SMAD2 and SMAD3 which translocate into the nucleus and form complexes that regulate transcription of target genes. TGFb can also elicit its biological effects by activation of SMAD-independent pathways.31,32 Inhibitory SMAD (SMAD6 and SMAD7) inhibit TGFb signaling. Importantly, a reduced SMAD7 expression sensitizes cells to the antiproliferative effects of TGFb and contributes to anemia in patients suffering from MDS, suggesting, firstly, that identifying transcriptional regulators of SMAD7 could enlighten our understanding of erythropoiesis and, secondly, that inhibiting TGFb signaling could be a therapeutic strategy that would mitigate ineffective hematopoiesis in disease states.33-35 In the present work, we used primary human CD34+ cells to demonstrate that RINF knockdown affects human erythropoiesis and mitigates RBC production through a mechanism that is mediated by SMAD7, the main inhibitor of TGFb signaling.

Methods The methods for flow cytometric cell sorting of megakaryocyte-erythroid progenitor (MEP) cells, immunofluorescence studies, chromatin Immunoprecipitation (ChIP) experiments, and the primer sequences used for quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) and ChIP-qPCR are described in the Online Supplementary Methods.

Primary culture of hematopoietic cells Mononuclear cells were isolated from cord blood (CRB Saint-

haematologica | 2022; 107(1)

Louis Hospital, Paris, France) or adult bone marrow donors with written informed consent for research use, in accordance with the Declaration of Helsinki. The ethics evaluation committee of INSERM, the Institutional Review Board (IRB00003888), approved our research project (n. 16-319). Mononucleated cells were separated by Ficoll-Paque (Life Technologies) and purified with a human CD34 MicroBead Kit (cat. n. 130100453, Miltenyi Biotec). A two-step culture method for cell expansion and erythroid differentiation was used to obtain highly stage-enriched erythroblast populations.36,37 Briefly, CD34+ cells (purity 94-98%) were cultured for 7 days with interleukin (IL)6, IL3 (10 ng/mL), and stem cell factor (SCF, 100 ng/mL) to expand hematopoietic progenitors.37 CD36+ erythroid progenitors were purified using magnetic microbeads (CD36 FA6.152 from Beckman Coulter and anti-mouse IgG1 MicroBeads from Miltenyi Biotec). Then, erythropoietin (EPO) was added for 10-18 days to allow erythroid differentiation. For assays of colony-forming cells (CFC, counted at 11-14 days), CD34+ hematopoietic stem cells were seeded in methylcellulose (H4034, StemCell Technologies). TGFbRI inhibitor SB431542 was from Selleckchem (cat. n. S1067)38 and TGFb was from Peprotech (cat. n. 100-21).

Flow cytometry and differentiation analysis To monitor end-stage erythroid differentiation of primary cells, cells were labeled with anti-human PE/Cy7-conjugated glycophorin A (GPA/CD235a) (cat n. A71564), PE-conjugated CD71 (IM2001U), and APC-conjugated CD49d (cat. n. B01682) from Beckman Coulter, PE-conjugated CD233/Band3 from IBGRL (cat. n. 9439-PE). Fluorescence-activated cell sorting (FACS) analysis was performed on an Accuri™ C6 Flow Cytometer (Becton-Dickinson). For morphological characterization cells were cytospun and stained with May-GrünwaldGiemsa. For the benzidine assay, 2x105 cells were incubated for 5 min with 0.01% benzidine and H O at 0.3% (Sigma-Aldrich). 2

Culture of cell lines and treatments K562 (ATCC#CCL243) and UT7 39 cells were cultured in RPMI 1640 medium and minimum essential medium-a supplemented with 2.5 ng/mL of granulocyte-macrophage colonystimulating factor (GM-CSF; Miltenyi Biotech), respectively. Both media were supplemented with 10% inactivated fetal bovine serum, 2 mM L-glutamine, 50 U/mL penicillin G and 50 mg/mL streptomycin (Life Technologies). To induce hemoglobin production, K562 cells were treated with hemin at 40 mM (Sigma-Aldrich). Erythroid maturation of UT7 was triggered by replacing GM-CSF with EPO 5 U/mL. 5.3

5.3

Lentiviral and retroviral transduction of hematopoietic cells For shRNA-mediated RINF knockdown, we used the pTRIPDU3/GFP lentiviral vector40 which drives the same shRNA sequences downstream of the H1 promoter as our previously described pLKO.1/puroR vector.1 For RINF overexpression, the retroviral vector MigR/IRES-GFP was used with the previously described experimental conditions.1 Green fluorescent protein (GFP) sorting was performed on a FACSAriaIII (BD Biosciences). For SMAD7 overexpression, the retroviral vector pBABE-puro-SMAD7-HA (Addgene plasmid#37044) was used as previously described.1,41 For doxycycline-inducible lentiviral expression, SMAD7-HA cDNA was inserted into the pINDUCER21 vector by gateway technology (Addgene plasmid #46948) and shRNA sequences were inserted in the Tet-pLKOGFP vector (modified from Tet-pLKO-puro, Addgene plasmid#21915, deposited by Dmitri Wiederschain).

269

A. Astori et al.

Quantitative reverse transcriptase polymerase chain reaction Cells were collected and stored directly at -80°C for RNA preparation with the TRIzol (Life Technologies) extraction protocol as previously described.1 First-strand cDNA synthesis (reverse trasncription) was carried out using a Transcriptor First Strand cDNA Synthesis Kit (cat. n. 489703000, Roche). RINF mRNA expression was detected using a Lightcycler® 480 ProbesMaster kit (cat. n. 4707494001, Roche). Relative mRNA expression was normalized to RLPL2, PPIA or ACTB gene expression in a two-color duplex reaction. For SMAD7, PU.1 and c-KIT mRNA detection, qRT-PCR was performed using SYBRGreen on a Light Cycler 480 machine (Roche) and gene expression was calculated by the 2-ΔΔCT method. Primer sequences are available in an Online Supplementary File.

Western blot analysis Total cell extracts were prepared and western blotting was carried out as previously described.1 Blots were incubated with a primary polyclonal antibody and then with an appropriate peroxidase-conjugated secondary antibody (anti-rabbit and antimouse IgG horseradish peroxidase [HRP] antibodies from Cell Signaling, cat. n. 7074S, and 7076S, respectively) and anti-goat IgG HRP antibody from Southern Biotech (cat. n. 6160-05). Proteins were detected using a chemiluminescent system (Clarity Western ECL, cat. n. 170-5060, Biorad). Rabbit polyclonal antibodies were previously described for RINF,1 P85/PI3KR1,42 or commercially purchased for anti-RINF, anti-SMAD7 (cat. n. 16513-1-AP, cat. n. 25840-1-AP, ProteinTech), anti-b-actin (ACTB, cat. n. A1978, Sigma-Aldrich), anti-phospho-SMAD2/3 (cat. n. 8828, Cell Signaling), and anti-HSC70 mouse monoclonal antibody (SC-7298, Santa Cruz Biotechnology).

Statistical analyses All analyses were performed using GraphPad Prism software. When the distribution of data was normal, the two-tailed Student t test was used for group comparisons. Contingency tables were established using the Fisher exact test, and the Pearson correlation coefficient was used to determine the correlation between the normally distributed RINF mRNA and SMAD7 mRNA expression values. Statistics were carried out on a minimum of three independent experiments. The statistical significance of P values is indicated in the Figures: not significant (ns): P>0.05; *P<0.05; **P<0.01; ***P<0.001. Error bars represent confidence intervals at 95% or standard deviations (SD) for the qRT-PCR analyses.

Results RINF is expressed during early human erythropoiesis We employed cytokine-induced erythroid differentiation of CD34+ progenitor cells isolated from human cord blood or adult bone marrow to model human erythropoiesis in vitro. To obtain highly enriched populations of differentiating erythroblasts, cells were grown in a two-phase liquid culture (experimental design in Figure 1A) as previously described.36,37 We found that RINF protein was present at the earliest stages of erythroid differentiation and reached its highest level in progenitors and proerythroblasts (ProE) (Figure 1B). After that, the level of RINF protein decreased in the basophilic erythroblast stage and was barely detectable in late erythroblasts (i.e., polychromatic and orthochromatic erythroblast stages). Similarly, RINF mRNA levels were high in erythroid progenitors but decreased at 270

the basophilic stage and onwards (Figure 1C). This temporal expression pattern mirrors RINF expression data extracted from three independent transcriptome datasets (microarray or RNA-sequencing) reported by Merryweather-Clarke et al.,43 An et al.,44 and Keller et al.45 with adult or cord-blood CD34+ cells (Online Supplementary Figure S2). Taken together, our data show that RINF expression peaks in erythroid progenitors and proerythroblasts during cytokine-induced erythropoiesis.

RINF silencing results in diminished red blood cell production without affecting cell viability We next investigated the consequences of RINF knockdown on erythropoiesis and cell viability (see experimental design in Figure 2A). Knockdown experiments were performed with two previously validated short-hairpin RNA (shRINF#4 and shRINF#3) sequences1 driven by the lentiviral pTRIPDU3/GFP vector (Online Supplementary Figure S3A, B). We found that RINF was extinguished as early as 2 days after transduction (i.e., at day 0 of EPO) with shRINF#4 at both mRNA (Figure 2B) and protein (Figure 2C) levels, without affecting cell viability, assessed by trypan blue cell counting (Figure 2D) or cell growth up to day 10 of EPO (Figure 2E, left panel). However, in the last days of ex vivo culture (day 11-17), we noticed a reduction in total number of RBC produced (average decrease of ~45%) at 17 days with EPO (Figure 2E, right panel), which was particularly marked for four donors out of six studied. To investigate how RINF knockdown affects the clonogenic capacity of HSPC, we next performed comparative colony-forming assays for the development and maturation of erythroid (burst-forming unit erythroid progenitors; BFU-E), myeloid (CFU-G, CFU-M, or CFU-GM) and mixed (CFU-GEMM) colonies. No statistically significant changes were noted in the total number of colonies or the number of CFU-M, CFU-GM, or CFU-GEMM (Online Supplementary Figure S3C). However, the proportions of granulocytic colonies (CFU-G) and BFU-E, were slightly reduced or increased, respectively (P<0.01, Fisher exact test). The number of MEP-sorted cells was also increased under RINF knockdown conditions (Online Supplementary Figure S3E), and this increase corresponded with an increase in small (more mature) BFU-E, findings that suggest an accelerated maturation. A more careful analysis of colony size revealed that the median size of BFU-E derived from CD34+ cells dropped by about 30% (i.e., from 0.101 to 0.071 mm2, unpaired t-test, P=0.007, n=3 donors) after RINF knockdown (Figure 2F). Concomitantly with the size reduction of large BFU-E (the most immature ones), we noted a slight but statistically significant increase in small BFU-E (the most mature ones), in agreement with accelerated maturation (Figure 2G).

Erythroid maturation is affected by RINF We next investigated whether the reduced expansion observed from day 10 of EPO could be the consequence of accelerated maturation. Indeed, we observed that erythroid differentiation was accelerated under conditions of RINF knockdown using quantification by benzidine staining (Figure 3A), morphological analysis of cells stained with May-Grünwald-Giemsa (Figure 3B), and flow cytometry for the erythroid markers GPA and CD71 (Figure 3C, upper panel), and CD49d and Band3 (Figure 3C, lower panel). These differences were statistically significant (P<0.001) for CD34+ cells derived from both cord haematologica | 2022; 107(1)

RINF maintenance of SMAD7 sustains human erythropoiesis

blood and adult bone marrow. The shRNA-RINF-induced acceleration of erythropoiesis was supported by the reduction in cell size (not shown) and the more rapid downregulation of c-KIT and PU.1 mRNA in RINF knockdown cells (Figure 3D). The accelerated maturation was also characterized by a pronounced reduction of ProE cells (∼2.8-fold less) enumerated at day 11 (Figure 3B). Despite a weaker efficiency of shRINF#3 to knockdown RINF expression compared to shRINF#4 (Online Supplementary Figure S3B, Figure 3E, left panel), we also confirmed that a similar phenotype, i.e., accelerated maturation (as shown by increased benzidine staining) (Figure 3E, right panel) and a reduction in the total num-

ber of RBC produced at day 17 (Figure 3F), was obtained with another shRNA sequence targeting RINF.

RINF controls erythroid maturation and red blood cell expansion in a transforming growth factor b-dependent manner We then wondered whether TGFb, a well-known inducer of erythroid maturation and inhibitor of RBC expansion, could be involved in the RINF-dependent phenotype. In agreement with an important role for autocrine TGFb signaling in the early stages of erythropoiesis, both TGFB1 and TGFBRI were highly expressed in erythroid progenitors (Figure 4A). To functionally validate an involvement of the

haematologica | 2022; 107(1)

Figure 1. RINF is expressed in early erythropoiesis and downregulated during maturation of human hematopoietic stem and progenitor cells. (A) Schematic representation of cell culture experiments. CD34+ cells (from cord blood or adult bone marrow) were first amplified for 7 days (D7 to D0, in black) in the presence of stem cell factor (SCF), interleukin-3 (IL3) and interleukin-6 (IL6). Cells were sorted for CD36 expression and erythropoietin (EPO) was added to the medium in order to trigger the late stage of erythropoiesis, a phase lasting 11-17 days (in red). (B) For morphological characterization, cells were cytospun and stained with MayGrünwald-Giemsa (MGG) and the repartition of cells among differentiation stages indicated in percentages (right panel): early progenitor (Prog), proerythroblast (ProE), early and late basophilic erythroblast (Baso), polychromatophilic erythroblast (Poly), orthochromatic erythroblast (Ortho), and reticulocyte (Retic). The scale bar represents 20 mm. The experiment presented here is representative of four distinct experiments which were performed on cells isolated from human cord blood. RINF protein was detected by western-blot analysis (lower panel). HSC70 protein was used as a loading control. For each lane, 40 mg of protein were loaded. Band intensities were quantified with the Multigauge software and the relative expression of RINF protein (normalized to the loading control) is indicated above the image in percentages on day 3. (C) CXXC5 mRNA expression was measured by quantitative reverse transcriptase polymerase chain reaction (qRT-PCR). CXXC5 mRNA expression is downregulated from the basophilic erythroblast stage. Here, CD34+ cells were isolated from adult human bone marrow, but a similar expression profile was observed with cells isolated from cord blood.

271

A. Astori et al.

Figure 2. Legend on following page.

272

haematologica | 2022; 107(1)

RINF maintenance of SMAD7 sustains human erythropoiesis

Figure 2. RINF knockdown leads to a reduction in total number of red blood cells generated from human hematopoietic stem and progenitor cells. (A) Schematic representation of the shRNA-mediated RINF loss of expression experiment. CD34+ cells isolated from cord-blood were transduced at day 5 of the amplification step with the lentiviral vector pTRIPDU3/GFP expressing a shRNA targeting RINF mRNA expression (shRNA/RINF) or a non-target sequence (shRNA/Control). GFP+ cells were sorted by fluorescent activated cell sorting 2 days after transduction. At that time, cell culture kinetics with less than 80% of CD36+ cells or more than 5% of cell death (trypan blue) were not pursued. Erythropoietin (EPO) was added after sorting and maintained for 17-18 days. (B) RINF knockdown efficiency was estimated at the mRNA level by quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) at D0 of EPO (i.e., right after GFP-sorting). Here, the results are shown for one representative experiment out of seven independent experiments performed. A Student t-test was performed. (C) RINF knockdown efficiency was estimated at the protein level (specific band at 33 KDa) by western-blot analysis (see Methods). P85 (PIK3R1) was used as the loading control.42 In order to obtain enough protein material, CD34+ cells were amplified from a pool of five cord-blood units (5 donors). For each lane, protein extracted from 3x105 cells was loaded right after sorting, at D0 of EPO. Band intensities were quantified with Multigauge software and the relative expression of RINF protein (normalized to the loading control) is indicated above the image in percentages of the shControl condition (D) Cell viability was assessed by the trypan blue exclusion method at 2 or 3 days after GFP-sorting. The histogram represents the percentage of trypan blue negative cells. (E) Cell culture growth was monitored in six independent experiments (here, cord-blood donors). The proliferation curve, expressed in cumulative population doublings, is shown for one representative experiment (donor#6, left panel). For most experiments (or “donors”), CD34+ cells from several cord blood units (2-4) were pooled. Fold cell expansion (right panel) was calculated by dividing the absolute output number of expanded red blood cells after 17-18 days of culture with EPO by the respective number on day 0 (of treatment with EPO). An average reduction of ~45% was noted in the RINF knockdown condition (paired t-test, P=0.04, n=6). For two out of the six donors (donors 1 and 2, in gray), the shRNA-RINF#4-induced reduction in the total number of RBC was not obvious in our serum-free conditions (and absence of transforming growth factor b [TGFb]). These two unresponsive donors were also treated with TGFb, and data are presented in Figure 4E, F (F) Photographs of the five larger BFU-E colonies for the two conditions, here for one representative donor at day 11 (left panel). Two days after transduction (day 5 of expansion), GFP-expressing cells were sorted by flow cytometry and 500 cells per well were seeded into methycellulose containing cytokines (IL3, SCF, G-CSF, GM-CSF, and EPO) enabling the formation of erythroid colonies (BFU-E). BFU-E colonies were imaged between 11 and 14 days of culture numerated from three donors (n=103 and n=110, respectively). The size of each BFU-E colony was determined by imaging analysis using ImageJ/Fiji. On average, the colonies were smaller for the shRINF#4-transduced condition (unpaired two-tailed t-test, P<0.01). (G) CD34+ cells were isolated from bone marrow of healthy donors (n=4). Colony-forming cells (CFC) were counted between 11 and 14 days of culture. The histogram illustrates the mean number of small and large BFU-E-derived colonies. The error bars indicate donor-to-donor variability for each vector (± standard error of mean). To assess the effect of shRNAmediated RINF silencing, a paired comparison was performed using the Fisher exact test.

TGFb signaling pathway during RINF knockdown-dependent acceleration of erythropoiesis, we performed liquid cell culture experiments of primary CD34+ cells in the presence of SB431542, a potent and selective inhibitor of TGFbRI38 (Figure 4B). Strikingly, in the presence of SB431542, the maturation of RINF knockdown cells (evidenced by a faster acquisition of erythroid cell surface markers such as CD49d/Band3 at day 10 of EPO) was not accelerated and SMAD2 protein was not phosphorylated/activated after 3 days of EPO (Figure 4C, lanes 5 and 6), supporting the efficiency of the inhibitor. Conversely, in the absence of inhibitor, RINF-silencing gave rise to a faster rate of phosphorylation of SMAD2 protein, here noted at 8 h of EPO (Figure 4C compare lanes 1 and 2, or the kinetics of pSMAD2 waves in both conditions, right panel), indicating a higher sensitivity to autocrine TGFb signaling, which would be mediated through SMAD2, as previously reported in hematopoietic cells.35 A weaker RBC expansion was also noted for RINF knockdown cells, even when TGFb1 (5 ng/mL) was added as late as day 11 of EPO treatment (Figure 4D). Moreover, even for the two donors whose RINF silencing was more subtle (donors 1 and 2, Figure 2E), RINF knockdown cells were more sensitive than control cells to TGFb1 (Figure 4E, F). This was especially pronounced at low doses (0.1 ng/mL), at which we found faster acquisition of GPA (at day 6 of EPO) and reduced RBC production (at day 17 of EPO) (Figure 4F). Taken together, these data suggest that RINF knockdown cells were more sensitive to TGFb and that the level of RINF in erythroid progenitors influences the upcoming RBC production, at least in the presence of TGFb.

Identification of SMAD7 as a RINF target gene candidate To identify molecular mechanisms by which RINF might modulate responsiveness to TGFb, we reanalyzed our previously published gene expression microarray datasets of K562 and UT7 cells transduced with shRINF#4 or shControl.7 A relatively discrete set of 193 gene candidates (Online Supplementary Table S1) were downregulated by RINF knockdown in both cell lines. Strikingly, ingenuity pathway analyses revealed that this gene list was enriched in genes belonging to the TGFb signaling pathway (Online Supplementary Figure S4C). Considering its well-known 5.3

haematologica | 2022; 107(1)

inhibitory function on TGFb-signaling and its low expression in MDS, SMAD7 appeared as a promising candidate for functional investigation.33,34 Encouragingly, SMAD7 was also downregulated by RINF knockdown in a third hematopoietic cell line, MV4-11 (Online Supplementary Figure S4E, right panel) and the downregulation of SMAD7 was confirmed by qRT-PCR analyses in K562 cells, even though the knockdown appeared moderate (~45%, P<0.001) in these independent experiments (Online Supplementary Figure S4D, right panel). We next measured RINF levels in two cell line models of erythroid maturation, the K562 cell line treated with hemin and the UT7 cell line treated with EPO39 after GM-CSF withdrawal (Online Supplementary Figure S4A, B). In agreement with our findings in primary cells, RINF protein expression was downregulated upon treatment-induced erythroid differentiation, as early as 10 h after treatment with hemin for K562 cells, and after 2 days with EPO for UT7 cells (Online Supplementary Figure S4A). Moreover, RINF knockdown accelerated the erythroid maturation program (Online Supplementary Figure S4B) triggered by hemin in K562 cells (noted after 1 day of treatment), or EPO in UT7 cells (at 4 days of treatment). We also investigated whether RINF overexpression could delay hemoglobinization. To this end, we used the retroviral system MigR/IRESGFP vector1 (Online Supplementary Figure S5A, upper panel) to drive ectopic expression from full-length RINF cDNA. We found high constitutive expression of RINF protein for both cell lines (Online Supplementary Figure S5A), without cell toxicity or consequences on cell proliferation (not shown). When we evaluated erythroid differentiation using benzidine staining (Online Supplementary Figure S5B) we found that RINF overexpression reduced hemoglobinization in both cell lines (P<0.001), consistent with our knockdown experiments (Online Supplementary Figure S4A, B). 5.3

5.3

RINF controls SMAD7 transcription directly Our qRT-PCR analyses demonstrated a robust increase (approximately 5-fold) of SMAD7 mRNA levels in cells overexpressing RINF (Figure 5A). To investigate whether RINF-mediated induction of SMAD7 mRNA was direct and occurred at the transcriptional level, we performed ChIP experiments with anti-RINF antibodies. To this end, five sets of primers were designed for regions encompassing the 273

A. Astori et al.

Figure 3. RINF silencing accelerates erythroid maturation of human primary erythroblasts. Erythroid differentiation was monitored in six separate experiments performed with CD34+ cells isolated from either cord blood (n=3) or adult bone marrow (n=3), using different techniques. Despite donor-to-donor variability in the maturation kinetics, we consistently observed accelerated maturation for RINF knockdown conditions. Of note, data presented in Figure 3B, C correspond to two distinct donors. (A) Hemoglobin production of primary cells was estimated by a benzidine assay. CD34+ cells isolated from cord blood were transduced with the lentiviral vector pTRIPDU3/GFP expressing a shRNA targeting RINF mRNA expression (shRNA/RINF) or a non-target sequence (shRNA/Control). GFP+ cells were sorted by fluorescent activated cell sorting (FACS) 2 days after transduction. The black arrow at day 6 indicates the kinetic point selected for one representative picture in the right panel. (B) Morphological analysis was performed after staining with May-Grünwald-Giemsa (MGG). A Fisher exact test was applied to determine whether the observed acceleration of maturation was statistically significant. For this, enumerated erythroid cells (n=454) were segregated into either early (Progenitors, ProE, Baso) or late (Polychro, Ortho, Retic) groups. P<0.0001. (C) Erythroid maturation was also monitored by flow cytometry using anti-CD71 and anti-glycophorin-A (GPA/CD235a) labeled antibodies (upper panel) or using anti-CD49d and anti-Band3 antibodies (lower panel). Scattergrams of two representative experiments are shown (n=6), indicating a more mature population of cells for shRNA-RINF conditions either at day 6 of EPO (GPA+, upper panel), or at day 10 of EPO (Band3+, lower panel). (D) RINF loss of expression in human CD34+ cells is also associated with accelerated downregulation of cKIT and PU.1. A histogram of relative mRNA expression determined by quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) for RINF (left panel), cKIT (middle panel) and PU.1 (right panel). PU.1 and c-KIT mRNA are known to be downregulated upon erythroid maturation and are used here as molecular markers. In RINF knockdown cells (black bars), the downregulation of c-KIT and PU.1 mRNA preceded that observed in control cells by at least 3 days, in agreement with an accelerated maturation. (E, F) In order to validate our data with a second shRNA sequence targeting RINF, CD34+ cells were transduced with the lentiviral vector pLKO-Tet-ON/shRNA vector, allowing doxycycline-inducible expression of shRNA sequences targeting RINF expression (shRINF#3 or shRINF#4). GFP+ cells were sorted by FACS 2 days after transduction and cultured in the presence of doxycycline 0.2 mg/mL. (E) The left histogram represents the levels of RINF mRNA obtained with the two sequences targeting RINF (shRINF#3 or shRINF#4) and detected by qRT-PCR. Values are expressed in percentage of the shControl condition. The right histogram represents the percentages of benzidine positive cells obtained with the three shRNA sequences (shControl, shRINF#3, or shRINF#4). (F) Cell culture growth was monitored in one additional experiment (here, a pool from 2 adult donors) and the cumulative population doublings is indicated (left panel) as well as the fold expansion between D0 and D17 of treatment with erythropoietin (EPO) (right panel).

274

haematologica | 2022; 107(1)

RINF maintenance of SMAD7 sustains human erythropoiesis

SMAD7 promoter (Figure 5B). As shown in Online Supplementary Figure S5C, we found that RINF binds directly to several regions (R1 to R4) of the SMAD7 promoter and that this binding was increased in cells overexpressing RINF. Moreover, H3K4me3 histone marks, quantified by

ChIP-qPCR (Figure 5C), were increased by RINF overexpression in four proximal regions (R2-R4) of the SMAD7 transcription start site, indicating that chromatin was in a more active transcriptional state. To investigate functional relationships between RINF

Figure 4. Legend on following page.

haematologica | 2022; 107(1)

275

A. Astori et al.

Figure 4. The RINF knockdown-induced phenotype is dependent on transforming growth factor b in human primary cells. (A) Transforming growth factor b (TGFβ) ligand and TGFBRI expression profiles were extracted from our proteomic analyses of cord blood-derived CD34+ cells (Gauthier et al.37). TGFB1 and TGFBRI were highly expressed at the Prog1-Prog2 stage (i.e. D1-D3 of erythropoietin [EPO] treatment), suggesting important autocrine signaling in early steps of erythroid maturation in our serum-free culture conditions. (B) Cord blood-derived CD34+ cells were transduced at day 5 of the amplification step, GFP-sorted (2 days after transduction), and EPO was added for 8 to 12 days, in the presence of TGFb inhibitor (SB431542, at 10 mM) or vehicle control (dimethylsulfoxide, DMSO). The histograms (left panel) represents fluorescence activated cell sorting (FACS) analyses of the whole cell suspension for CD49d and Band3 antigens (here at 10 days of EPO). In order to determine whether the observed acceleration of maturation was statistically significant, Student t-tests were performed. Scattergrams (right panel) present one representative experiment out of three separate experiments performed. (C) RINF (left panel) and phosphorylated SMAD2 (right panel) protein levels were estimated at 8 h and 72 h of treatment with EPO by western-blot analysis (see Methods), with the same kinetics as presented in Figure 2C (pool of 5 cord blood units). As expected, the specific band for phospho-SMAD2 (Ser465/467), indicated by an arrow, was not detected in the presence of TGFb inhibitor (SB431542, at 10 mM). P85 (PIK3R1) was used as a loading control (lower panel) from the same blot, and at a shorter exposure time than in Figure 2C.42 For each lane, protein extracted from 3x105 cells was loaded. Band intensities were quantified with Multigauge software and the relative phospho-SMAD2 expression is shown (right panel). (D) TGFb (5 ng/mL) or vehicle control (DMSO) was added at 11 days of EPO and the proliferation monitored during 1 additional week, until day 18 of EPO. For both treated (in red) and untreated (in black) conditions, red blood cell expansion was lower for RINF-silenced cells. The kinetics shown corresponds to Donor#4 of Figure 2E (right panel). Cell culture growth is represented in population doublings (left panel) or cumulative cell number at day 18 of EPO (right panel). (E and F) The two donors (1 and 2) that seemed not sensitive to RINF knockdown in the absence of TGFb (Figure 2E) were treated with variable doses of TGFb1 (ranging from 0.1 to 2.5 ng/mL), and scattergrams (upper panel) represent FACS analyses of the whole cell suspension labeled for both GPA and CD71 antigens. The corresponding dose-effect curves are also shown (lower panel). One representative scattergram at the lowest concentration (0.1 ng/mL) of exogenous TGFb1 is shown. (F) A histogram representing GPA acquisition (upper panel) which is more advanced in RINF knockdown cells (here at day 6 of EPO) in the presence of low doses of TGFb1 (here at 0.1 ng/mL). Cell growth was also monitored, and the total number of red blood cells numerated at day 17 of EPO are presented in the histograms (lower panel).

and SMAD7, we next performed “rescue” experiments in K562 cells. We employed retroviral transduction to express SMAD7 ectopically (Figure 5D). Our data demonstrated that SMAD7 overexpression did not affect endogenous expression of RINF mRNA or protein (Figure 5E, F), but did reduce hemoglobinization (Figure 5G, lane 3), mimicking the effect of ectopic RINF (Online Supplementary Figure S5B). We did not detect any change in cell viability or proliferation in the absence of hemin. Importantly, under conditions in which SMAD7 was ectopically expressed, RINF knockdown did not result in acceleration of hemoglobinization (Figure 5G). Taken together, these data indicate that RINF binds the SMAD7 promoter directly to control its expression, and that SMAD7 is an effector downstream of RINF during erythroid maturation.

RINF and SMAD7 mRNA expression are correlated in bone marrow from adult donors and patients with myelodysplastic syndrome We investigated whether RINF and SMAD7 mRNA expression correlated in primary human CD34+ cells from healthy donors and donors with MDS. To this end, we performed qRT-PCR on CD34+ cells isolated from adult bone marrow donors (n=11) and found a highly significant correlation between RINF mRNA and SMAD7 mRNA (Pearson, rho=0.684, P=0.02) (Figure 6A). We confirmed these data by analyzing an independent microarray dataset from Pellagatti et al.46 As shown in Figure 6B, RINF and SMAD7 mRNA levels correlated strongly in CD34+ cells isolated from healthy donors (Pearson rho=0.784, P<0.001, n=17). Interestingly, this correlation also reached statistical significance in CD34+ cells from MDS -5q patients (Pearson rho=0.603, P<0.001, n=47) but not in other MDS patients, i.e., without del(5q) or 5q-. We then compared the intensity of this correlation with the other 5q genes (n=48 genes) from the commonly deleted regions associated with high risk, and the CXXC5 probe sets were those correlating best with the SMAD7 probe set in both the healthy control group (rho=0.819, and rho=0.784) and the MDS patients’ cohort with del(5q) (Table 1) (Online Supplementary Figure S6A). Reinforcing the previously reported relevance of SMAD7 in MDS pathophysiology,33,34 SMAD7 expression was not only extinguished in CD34+ MDS samples compared to normal samples (in the microarray dataset from Gerstung et al.,47 (Online Supplementary Figure S6C), but patients 276

with the lowest SMAD7 expression had a significantly shorter overall survival (Online Supplementary Figure S6D).

RINF-silencing alters genome-wide hydroxymethylation Since murine RINF has recently been described as a necessary platform for TET2 activity,15,48 we wondered whether RINF-silencing could affect the genome-wide 5hmC of human immature erythroid cells. As shown in Figure 6C, we observed a statistically significant loss of 5hmC detected by immunofluorescence or flow cytometry (Figure 6D) in knockdown primary cells.

A RINF/SMAD7 axis controls erythroid maturation and red blood cell expansion in primary cells We next investigated whether RINF knockdown altered SMAD7 expression in primary erythroid progenitors. As shown in Figure 7A, B, a 60-70% knockdown of RINF led to a statistically significant knockdown of SMAD7 (by 40-45%) in CD34+ cells isolated from cord blood (n=3 donors, paired Student t-test, P<0.001). Accordingly, a 30% knockdown of SMAD7 protein level was detected by immunofluorescence in primary erythroid progenitors (P<0.0001) (Figure 7C). For a direct demonstration of the relevance of this finely tuned regulation of SMAD7 mediated by RINF protein in primary HSPC, we next performed experiments in which we knocked down RINF and induced SMAD7 expression ectopically. To this end, we first transduced CD34+ cells with the doxycyclineinducible vector pInducer21/SMAD7-HA (Figure 7D), and then with the pTRIPUD3 vector that drives shRNA/RINF or shRNA/Control expression. Notably, in the presence of doxycycline, SMAD7 expression (Figure 7E, F) robustly prevented both the RINF knockdown-dependent accelerated maturation (Figure 7G) and reduction of RBC numbers (Figure 7H).

Discussion An increasing body of work suggests that RINF is involved in the maturation, function, or development of various cell types such as neural stem cells,21 endothelial cells,49 myoblasts,14 myofibroblasts,24 osteoblasts,23 as well as in kidney development,22 wound healing,24 and hair regrowth.50 Here, we report that the epigenetic factor RINF is a transcriptional regulator of SMAD7 which fine-tunes TGFb sensitivity of erythroid progenitors, findings that haematologica | 2022; 107(1)

RINF maintenance of SMAD7 sustains human erythropoiesis

bring insight into molecular barriers that may prevent effective erythropoiesis. Our in vitro erythropoiesis experiments demonstrated that RINF is expressed in human erythroid progenitors (i.e., BFU-E and CFU-E) and proerythroblasts (ProE) but not in

the last stages. Loss of RINF expression does not affect cell viability or cell proliferation but accelerates erythroid maturation, and noticeably reduces RBC expansion. The average reduction in RBC number was estimated at ~45% in six experiments (Figure 2F) or one population doubling (i.e.,

Figure 5. SMAD7 is a direct transcriptional target of RINF. (A) The MigR1 retroviral system was used to constitutively express RINF in the K562 cell line. A few days after transduction, cells were sorted for GFP expression and relative expression of RINF and SMAD7 mRNA levels were measured by quantitative reverse transcriptase polymerase chain reaction (qRT-PCR). (B) Schematic representation of SMAD7 gene structure (deduced from the ENCODE database). CpG islands and primer set regions (R0 to R5) used for chromatin immunoprecipitation (ChIP) experiments are also indicated as well as the distance to the transcription start site (TSS) of SMAD7. (C) ChIP-qPCR bars indicate the percentage of SMAD7 promoter co-immunoprecipitated by anti-RINF (left panel) or anti-H3K4me3 (right panel) antibody in the K562 cell line. (D) Schematic representation of the pBABE retroviral vector used for SMAD7 overexpression or empty cassette control. (E) Histograms representing the RINF and SMAD7 mRNA levels detected by qRT-PCR. Values are expressed in percentage of the shControl condition. A Student t-test (unpaired) was performed to assess whether relative expression values were statistically significant from those of the shControl condition. (F) Relative expression of RINF and SMAD7 was measured by western-blot analysis after retroviral transduction of K562 cells with pBABE/SMAD7 or pBABE/Empty and puromycin selection (1 mg/mL) for 5 days. HSC70 was used as a loading control. (G) K562/SMAD7 and K562/Empty cells were then transduced with pTRIPDU3 lentiviral vector. Cells were sorted for GFP expression 2 days after transduction. To analyzed hemoglobin production, the benzidine test was used after 24 h of exposure to hemin. The percentage of benzidine-positive cells was established by counting at least 300 cells per sample. Pictures represent one of three independent experiments. For statistical analysis a Student t-test was performed for qPCR experiments, and hemoglobin production analysis (**P=0.0026; ns: P>0.05).

haematologica | 2022; 107(1)

277

A. Astori et al.

1.03, for the 4 donors #3-6), suggesting that RINF knockdown cells could skip one division (in the presence of TGFb). Our data indicate that RINF-dependent erythroid maturation is also dependent on TGFb signaling at physiological levels. This cytokine, a potent inducer of erythroid maturation and growth inhibition,27,29,51,52 is known to be enriched in the bone marrow microenvironment53 and we surmise that the RINF/SMAD7 regulation axis that we describe here is likely to exert a more pronounced effect in vivo than in our serum-free culture conditions (i.e., without exogenous TGFb). Moreover, since the mechanism of action is SMAD7-dependent, RINF downregulation may also sensitize hematopoietic cells to other cytokines of the TGFb superfamily which may act on late-stage erythropoiesis in vivo (such as activins and GDF11).54-57 In this manner, RINF could be one of the players regulating the dynam-

ic balance between long-term expansion of the erythroid pool versus fast production of RBC for immediate physiological needs. Two recent studies suggest that murine RINF could act as an anchoring platform necessary for TET2 (which lacks a CXXC domain) and its activity at CpG sites (5’hydroxymethylation), in plasmacytoid dendritic cells15 and mouse embryonic stem cells.48 Conversely, an earlier study indicated that RINF inhibits TET2 and hydroxymethylation of genomic DNA in a caspase-dependent manner (in mouse embryonic stem cells), and another one indicated that RINF binds to unmethylated (and not methylated) CpG sites,18 probably reflecting a more complex mechanism of action.12 We here provide the first experimental evidence (to our knowledge) that RINF can regulate genome-wide hydroxymethylation in human

Figure 6. RINF and SMAD7 mRNA closely correlate in primary human CD34+ bone marrow cells isolated from healthy donors or myelodysplastic syndrome patients with del(5q) and RINF-silencing leads to global loss of 5-hydroxymethylation. (A) CD34+ cells were isolated from adult bone marrow (n=11). RINF and SMAD7 mRNA levels were measured by quantitative reverse transcriptase polymerase chain reaction (qRT-PCR). In order to assess whether the expression of these two genes correlated in donors’ samples, a correlation analysis was performed with the Pearson correlation coefficient method (for non-parametric analysis): rho= 0.684), P(twotailed) <0.05 (B) RINF and SMAD7 gene expression data were extracted from a previously described microarray dataset published by Pellagatti et al.46 In this study, CD34+ cells were isolated from adult bone marrow (n=17). For mRNA detection, the 233955_x_at and 204790_at probesets were used for CXXC5 and SMAD7, respectively. A correlation analysis was performed with the Pearson rho (for non-parametric analysis). Strong correlation coefficients were noted for normal healthy controls (rho=0.784) and myelodysplastic syndrome (MDS) patients with del(5q), (rho=0.603), which were both highly significant: P(two-tailed) <0.001. (C) CD34+ cells isolated from cord blood or healthy adult donors were transduced with pTripizi vector expressing a shRNA targeting RINF expression (shRNA/RINF#4) or a “nontarget” shRNA (shRNA/Control). A couple of days later, during the amplification phase (see Figure 1A), cells were GFP-sorted and 5hmC was detected by immunofluorescence (left panel) or flow cytometry (right panel) using an anti-5hmC antibody labeled with a fluorescent dye. The intensity of the labeling was also quantified by cell imaging analysis using Fiji software. Representative images from three independent experiments are shown. Here, the CD34+ cells are from an adult donor at 4 days of expansion (2 days after transduction). Scale bar=50 mm.

278

haematologica | 2022; 107(1)

RINF maintenance of SMAD7 sustains human erythropoiesis

Table 1. List of genes located at 5q and their correlation with SMAD7 expression in primary human CD34+ cells isolated from bone marrow of healthy donors (n=17).

Gene name CXXC5 CXXC5 HSPA9 CXXC5 ETF1 IL17B SPARC ETF1 MATR3 CD74 TGFBI PDGFRB EGR1 SIL1 PSD2 UBE2D2 AFAP1L1 SH3TC2 CSF1R TMEM173 PAIP2 HNRNPA0 CDX1 EGR1 FAM53C AFAP1L1 MZB1 CARMN CDC23 CTNNA1 REEP2 SPATA24 RPS14 LRRTM2

Probeset

Pearson's (correl coeff.)

Gene location at 5q

222996_s_at 233955_x_at 200690_at 224516_s_at 201574_at 220273_at 212667_at 201573_s_at 238993_at 1567627_at 201506_at 202273_at 201694_s_at 218436_at 223536_at 201345_s_at 1555542_at 233561_at 203104_at 224916_at 222983_s_at 229083_at 206430_at 201693_s_at 218023_s_at 226955_at 223565_at 231987_at 223651_x_at 200764_s_at 205331_s_at 238027_at 208645_s_at 206408_at

0.819 0.784 0.729 0.705 0.640 0.554 0.532 0.508 0.452 0.424 0.422 0.391 0.373 0.355 0.352 0.350 0.343 0.337 0.335 0.312 0.289 0.271 0.259 0.214 0.193 0.184 0.180 0.180 0.178 0.176 0.167 0.137 0.134 0.128

High-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR Low-Risk CDR Low-Risk CDR High-Risk CDR High-Risk CDR Low-Risk CDR High-Risk CDR Low-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR Low-Risk CDR Low-Risk CDR Low-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR Low-Risk CDR High-Risk CDR High-Risk CDR Low-Risk CDR High-Risk CDR Low-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR Low-Risk CDR High-Risk CDR

continued from the previous column

NRG2 GFRA3 BRD8 CDC25C SH3TC2 HNRNPA0 KDM3B GRPEL2 TMEM173 EGR1 DNAJC18 FAM13B TRPC7 ABLIM3 NME5 KLHL3 WNT8A SPATA24 LECT2 IL9 ECSCR KLHL3 PKD2L2 ZNF300 SMAD5 TRPC7 SH3TC2 SLC23A1 CDC25C MYOT UBE2D2 PCYOX1L GRPEL2 KIF20A

242303_at 229936_at 242265_at 216914_at 240966_at 201055_s_at 210878_s_at 238427_at 224929_at 227404_s_at 227166_at 218518_at 208589_at 205730_s_at 206197_at 1555110_a_at 224259_at 1558641_at 207409_at 208193_at 227780_s_at 221221_s_at 221118_at 228144_at 225223_at 202363_at 219710_at 223732_at 205167_s_at 219728_at 201344_at 218953_s_at 226881_at 218755_at

0.122 0.114 0.096 0.083 0.052 0.041 0.041 0.040 0.028 0.013 -0.018 -0.025 -0.035 -0.068 -0.081 -0.094 -0.116 -0.137 -0.143 -0.157 -0.169 -0.179 -0.187 -0.191 -0.221 -0.241 -0.315 -0.319 -0.351 -0.386 -0.538 -0.559 -0.614 -0.786

High-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR Low-Risk CDR High-Risk CDR High-Risk CDR Low-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR Low-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR Low-Risk CDR High-Risk CDR High-Risk CDR Low-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR High-Risk CDR Low-Risk CDR Low-Risk CDR High-Risk CDR

All the genes were selected because they are located within one of the two commonly deleted regions (CDR) at chromosome 5q. Their mRNA expression level was extracted from a GDS3795 microarray study performed by Pellagatti et al.46 For each gene, a Pearson correlation coefficient was calculated by comparing its expression to the one of SMAD7, and the list is ranked by descending order.

continued in the next column

cells. The role of RINF that we describe here could pave the way to future studies that will be necessary to better understand the complex molecular epigenetic events that occur during normal and ineffective erythropoiesis12,15,48,58 and that may involve TET enzymes.59 Several correlative studies have implicated RINF as a candidate in the pathogenesis of MDS/AML.1,3,5,60 MDS is a complex and heterogeneous malignant hematologic disorder in which the loss of several genes is suspected to lead to the pathophysiology.4 CXXC5 is located at chromosome 5q31.2, in a commonly deleted region associated with high risk in MDS and AML (Online Supplementary Figure S1).1,4 Even though it is uncertain that RINF loss of expression in itself can cause the development of preleukemia, our functional data in human CD34+ cells indicate that loss of RINF may exacerbate cytopenia of haematologica | 2022; 107(1)

the erythroid lineage. Of importance, the functional contribution of Cxxc5/Rinf loss of function to MDS development was recently described in a mouse model, in which its invalidation (by random insertional mutation) was demonstrated to cooperate with Egr1 haploinsufficiency to promote MDS.4 This study indicated that the Cxxc5 gene could act as a tumor suppressor gene contributing to del(5q) myeloid neoplasms. Although RINF/CXXC5 is not frequently mutated in patients with MDS/AML,60 several recurrent chromosomal anomalies or gene mutations could lead to RINF loss of expression and contribute in the long-term to ineffective erythropoiesis and/or myeloid transformation (legend of Online Supplementary Figure S7). Our findings point to a molecular mechanism in which RINF functions as a negative modulator of TGFb signaling 279

A. Astori et al.

by upregulating and maintaining SMAD7. Consequently, RINF loss of expression will sensitize human erythroid progenitors to autocrine/paracrine TGFb-induced growth inhibition as a mechanism to prevent expansion of the erythroid pool. Our data also suggest that this signaling would be at least partly mediated through SMAD2 activation and phosphorylation (Figure 4C), in agreement with a previous report on human bone marrow.35 The in

vivo relevance of these findings is strengthened by the correlated RINF and SMAD7 mRNA expression in human bone marrow CD34+ cells. This correlation was also observed in MDS patients with del(5q) or 5q- but not in other forms of MDS, in which SMAD7 was silenced in most of the patients’ samples, whatever the RINF expression level. This extinction of SMAD7 in other MDS could be explained by various mechanisms such as miRNA21

Figure 7. Legend on following page.

280

haematologica | 2022; 107(1)

RINF maintenance of SMAD7 sustains human erythropoiesis

Figure 7. The RINF knockdown-induced phenotype is SMAD7-dependent in human primary cells. (A) Histogram representing RINF (left panel) and SMAD7 (right panel) mRNA expression levels in shRNA-transduced CD34+ cells isolated from three cord blood units. In these specific experiments, lentiviral transduction was performed after magnetic sorting of CD34+ cells (i.e., day 0 of the amplification step) and cells were GFP-sorted between 2 and 5 days of the amplification step. Data shown are mean ± standard deviation for each donor (the quantitative reverse transcriptase polymerase chain reaction [qRT-PCR] was performed in triplicate). To assess whether the mRNA level was statistically different between shRNA-RINF and shRNA-Scramble samples, a Student t-test was performed for each donor. Considering the three biological replicates, both RINF and SMAD7 mRNA levels were statistically downregulated by shRNA-RINF (paired Student t-test, P<0.001). (B) This histogram represents the level of SMAD7 mRNA obtained with the two sequences targeting RINF (shRINF#3 and shRINF#4) and detected by qRT-PCR. Values are expressed as percentages of the shControl condition. RINF mRNA levels are shown in Online Supplementary Figure S3B. (C) Since detection of SMAD7 protein level requires a large quantity of cells for western blot analysis and the active endogenous level of this protein can be below the detection threshold, SMAD7 protein levels were estimated by immunofluorescence. Images of stained cells were acquired on a wide-field Nikon Eclipse microscope through a ×20 objective, with an Eclipse TE2000, a cascade CDD camera (Photometrics). Images of SMAD7 had to be acquired with a binning of one, and two images were averaged for each condition. For cell imaging analysis the quantification of the labeling, performed on two images after background subtraction, the mean fluorescent intensity was measured in 100–120 cells for the two conditions, using Fiji software. Representative images from three independent experiments are shown. Scale bar = 50 mm. (D) Schematic representation of the pInducer21/SMAD7 lentiviral vector used for rescue experiments. This vector drives the expression of SMAD7/HA under a tetracycline response element (TRE) promoter that is activated after doxycycline treatment. To perform rescue experiments, cord blood-derived CD34+ cells (n=3 donors) were first transduced with the pInducer21/SMAD7 lentiviral vector, sorted for a low GFP expression, and transduced again with the pTRIPDU3/shRNA vectors targeting RINF or control. After a second step of GFP sorting (this time based on high GFP expression to sort cells transduced with both lentiviral vectors), cells were cultured for 2 more days before adding (or not) doxycycline, which induced SMAD7 expression. (E) Histograms representing RINF and SMAD7 mRNA levels in cells, detected by qRT-PCR. Values are expressed as percentages of the shControl condition. A Student t-test (unpaired) was performed to assess whether the observed relative difference was statistically significant. (F) RINF and SMAD7 were detected by western-blot analysis after doxycycline induction (here after 3 days of treatment with 0.2 µg/mL). ACTB was used as a loading control. (G) Cell maturation was determined by a benzidine assay (upper panel histogram). (H) Cumulative red blood cell number at day 17 of exposure to erythropoietin (EPO) is presented in the histogram in the lower panel.

expression.33 However, since RINF knockdown leads to global loss of 5hmC, it is tempting to speculate that the loss of RINF could lead to hypermethylation of the SMAD7 promoter in HSPC, a process that could have gone unnoticed in previous studies. Interestingly, the SMAD7 promoter is known to be silenced by hypermethylation in non-hematopoietic tissues, and hypomethylating agents such as azacytidine can revert this hypermethylation, and alleviates TGFb-induced diseases in several pathological models such as atherosclerosis,61 renal fibrosis62 and liver fibrosis.63 Thus, further studies should investigate whether the SMAD7 promoter is demethylated after treatment with hypomethylating agents (azacytidine or decitabine) in human HSPC isolated from MDS patients, and to what extend this mechanism could contribute to the efficacy of these epigenetic drugs to sustain erythropoiesis in the long-term. In support of this concept, TET2 mutation is a good indicator of response to azacytidine treatment.64 Previous studies have underscored the importance of SMAD7 regulation in MDS.33,34 Our present data complement these studies by demonstrating, first, a direct transcriptional mechanism by which SMAD7 is regulated in hematopoietic cells and, second, that the residual level of SMAD7 in MDS is associated with the severity of the disease and patients’ survival (Online Supplementary Figure S6D). These data are particularly relevant in the context of understanding how responsiveness to TGFb in the bone marrow microenvironment is fine-tuned,28 and in a context-specific manner.30 Even though this mechanism is unlikely to be solely responsible for TGFb-hypersensitivity, our findings pave the way for novel research avenues in blood disorders characterized by ineffective erythropoiesis such as myelofibrosis,65 Fanconi anemia,66 b-thalassemia,57,67 and MDS,55 and for which novel therapies based on TGFb inhibitors are particularly promising.56,68 Finally, beyond its role during erythropoiesis, the RINF/SMAD7 regulation axis described here could, at least partly, contribute to the pleiotropic effects of RINF described during wound healing,24 fibrosis,61-63,69 immunity,15 and tumor biology,1,3,6-9 and deserves to be investigated in these physiological processes.

haematologica | 2022; 107(1)

Disclosures FP is a holder of patents describing methods for RINF mRNA detection (WO2009/151337 and WO2012/010661).The other authors declare that they have no potential conflicts of interest. Contributions AA and GM performed most of the experiments. DS and SZ performed the ChIP experiments. YZ, VF, IM, and E-FG performed some experiments and discussed the data. AA, GM, FV, EL, MF, AK, OD, and CL discussed the data. ES-B, ID-F, DB, OH, and PM supervised the study. FP conceived the study and supervised the experiments. AA, GM, and FP analyzed the data, created the figures and wrote the manuscript. Acknowledgments The authors thank Pr. Jérôme Larghero and Thomas Domet from Centre de Ressources Biologiques / Banque de Sang de Cordon de l’AP-HP (Saint-Louis Hospital) for providing cord blood units, Anne Dubart-Kupperschmitt for the pTRIPDU3/GFP vector,40 Johan Lillehaug, Øystein Bruserud, Camille Humbert, Pierre de la Grange (Genosplice), Eric Nguyen, Sabrina Bondu, Alexandre Artus, Amandine Houvert, Nadège Bercovici, Emmanuel Donnadieu, Franck Letourneur, Brigitte Izac, Sebastien Jacques, and Alain Trautmann for technical assistance and scientific discussions, as well as Life Science Editors for reading and editing the manuscript. We are grateful to the staff of the IMAG’IC, CYBIO, and GENOM’IC facilities of the Cochin Institute. Funding This work was supported by INSERM, Paris-Descartes University, the Ligue Nationale Contre le Cancer (LNCC), the Cochin Institute, and the Laboratory of Excellence GR-EX. AA was supported by LNCC, Société Française d'Hématologie, Fondation pour la Recherche Médicale, and Boehringer Society (travel grant). GM was supported by a fellowship grant from the Ministère de l’Enseignement Supérieur et de la Recherche and Société Française d’Hématologie. ES-B is supported by the National Center for Scientific Research (CNRS) and her team was supported by LNCC and Fondation de France. We are also grateful for support from the French-Norwegian exchange program (Aurora to FP).

281

A. Astori et al.

References 1. Pendino F, Nguyen E, Jonassen I, et al. Functional involvement of RINF, retinoidinducible nuclear factor (CXXC5), in normal and tumoral human myelopoiesis. Blood. 2009;113(14):3172-3181. 2. Jerez A, Gondek LP, Jankowska AM, et al. Topography, clinical, and genomic correlates of 5q myeloid malignancies revisited. J Clin Oncol. 2012;30(12):1343-1349. 3. Kuhnl A, Valk PJ, Sanders MA, et al. Downregulation of the Wnt inhibitor CXXC5 predicts a better prognosis in acute myeloid leukemia. Blood. 2015;125(19): 2985-2994. 4. Stoddart A, Qian Z, Fernald AA, et al. Retroviral insertional mutagenesis identifies the del(5q) genes, CXXC5, TIFAB and ETF1, as well as the Wnt pathway, as potential targets in del(5q) myeloid neoplasms. Haematologica. 2016;101(6):e232236 5. Treppendahl MB, Mollgard L, HellstromLindberg E, Cloos P, Gronbaek K. Downregulation but lack of promoter hypermethylation or somatic mutations of the potential tumor suppressor CXXC5 in MDS and AML with deletion 5q. Eur J Haematol. 2013;90(3):259-260. 6. Astori A, Fredly H, Aloysius TA, et al. CXXC5 (retinoid-inducible nuclear factor, RINF) is a potential therapeutic target in high-risk human acute myeloid leukemia. Oncotarget. 2013;4(9):1438-1448. 7. Bruserud O, Reikvam H, Fredly H, et al. Expression of the potential therapeutic target CXXC5 in primary acute myeloid leukemia cells - high expression is associated with adverse prognosis as well as altered intracellular signaling and transcriptional regulation. Oncotarget. 2015;6(5): 2794-2811. 8. Benedetti I, De Marzo AM, Geliebter J, Reyes N. CXXC5 expression in prostate cancer: implications for cancer progression. Int J Exp Pathol. 2017;98(4):234-243. 9. Knappskog S, Myklebust LM, Busch C, et al. RINF (CXXC5) is overexpressed in solid tumors and is an unfavorable prognostic factor in breast cancer. Ann Oncol. 2011;22(10):2208-2215. 10. Zhang M, Wang R, Wang Y, et al. The CXXC finger 5 protein is required for DNA damage-induced p53 activation. Sci China C Life Sci. 2009;52(6):528-538. 11. Aras S, Pak O, Sommer N, et al. Oxygendependent expression of cytochrome c oxidase subunit 4-2 gene expression is mediated by transcription factors RBPJ, CXXC5 and CHCHD2. Nucleic Acids Res. 2013;41(4):2255-2266. 12. Ko M, An J, Bandukwala HS, et al. Modulation of TET2 expression and 5methylcytosine oxidation by the CXXC domain protein IDAX. Nature. 2013;497 (7447):122-126. 13. L'Hote D, Georges A, Todeschini AL, et al. Discovery of novel protein partners of the transcription factor FOXL2 provides insights into its physiopathological roles. Hum Mol Genet. 2012;21(14):3264-3274. 14. Li G, Ye X, Peng X, et al. CXXC5 regulates differentiation of C2C12 myoblasts into myocytes. J Muscle Res Cell Motil. 2014;35(5-6):259-265. 15. Ma S, Wan X, Deng Z, et al. Epigenetic regulator CXXC5 recruits DNA demethylase Tet2 to regulate TLR7/9-elicited IFN response in pDCs. J Exp Med. 2017;214(5): 1471-1491.

282

16. Marshall PA, Hernandez Z, Kaneko I, et al. Discovery of novel vitamin D receptor interacting proteins that modulate 1,25dihydroxyvitamin D3 signaling. J Steroid Biochem Mol Biol. 2012;132(1-2):147-159. 17. Yasar P, Ayaz G, Muyan M. Estradiol-estrogen receptor alpha mediates the expression of the CXXC5 gene through the estrogen response element-dependent signaling pathway. Sci Rep. 2016;6:37808. 18. Ayaz G, Razizadeh N, Yasar P, et al. CXXC5 as an unmethylated CpG dinucleotide binding protein contributes to estrogen-mediated cellular proliferation. Sci Rep. 2020;10(1):5971. 19. Long HK, Blackledge NP, Klose RJ. ZFCxxC domain-containing proteins, CpG islands and the chromatin connection. Biochem Soc Trans. 2013;41(3):727-740. 20. Melamed P, Yosefzon Y, David C, Tsukerman A, Pnueli L. Tet enzymes, variants, and differential effects on function. Front Cell Dev Biol. 2018;6:22. 21. Andersson T, Sodersten E, Duckworth JK, et al. CXXC5 is a novel BMP4-regulated modulator of Wnt signaling in neural stem cells. J Biol Chem. 2009;284(6):3672-3681. 22. Kim MS, Yoon SK, Bollig F, et al. A novel Wilms tumor 1 (WT1) target gene negatively regulates the WNT signaling pathway. J Biol Chem. 2010;285(19):14585-14593. 23. Kim HY, Yoon JY, Yun JH, et al. CXXC5 is a negative-feedback regulator of the Wnt/beta-catenin pathway involved in osteoblast differentiation. Cell Death Differ. 2015;22(6):912-920. 24. Lee SH, Kim MY, Kim HY, et al. The Dishevelled-binding protein CXXC5 negatively regulates cutaneous wound healing. J Exp Med. 2015;212(7):1061-1080. 25. Pardali K, Moustakas A. Actions of TGFbeta as tumor suppressor and pro-metastatic factor in human cancer. Biochim Biophys Acta. 2007;1775(1):21-62. 26. Siegel PM, Massague J. Cytostatic and apoptotic actions of TGF-beta in homeostasis and cancer. Nat Rev Cancer. 2003;3(11):807-821. 27. Zermati Y, Varet B, Hermine O. TGF-beta1 drives and accelerates erythroid differentiation in the epo-dependent UT-7 cell line even in the absence of erythropoietin. Exp Hematol. 2000;28(3):256-266. 28. Blank U, Karlsson S. TGF-beta signaling in the control of hematopoietic stem cells. Blood. 2015;125(23):3542-3550. 29. Krystal G, Lam V, Dragowska W, et al. Transforming growth factor beta 1 is an inducer of erythroid differentiation. J Exp Med. 1994;180(3):851-860. 30. Ruscetti FW, Akel S, Bartelmez SH. Autocrine transforming growth factor-beta regulation of hematopoiesis: many outcomes that depend on the context. Oncogene. 2005;24(37):5751-5763. 31. Derynck R, Zhang YE. Smad-dependent and Smad-independent pathways in TGFbeta family signalling. Nature. 2003;425(6958):577-584. 32. Shi Y, Massague J. Mechanisms of TGFbeta signaling from cell membrane to the nucleus. Cell. 2003;113(6):685-700. 33. Bhagat TD, Zhou L, Sokol L, et al. miR-21 mediates hematopoietic suppression in MDS by activating TGF-beta signaling. Blood. 2013;121(15):2875-2881. 34. Zhou L, McMahon C, Bhagat T, et al. Reduced SMAD7 leads to overactivation of TGF-beta signaling in MDS that can be reversed by a specific inhibitor of TGF-beta receptor I kinase. Cancer Res.

2011;71(3):955-963. 35. Zhou L, Nguyen AN, Sohal D, et al. Inhibition of the TGF-beta receptor I kinase promotes hematopoiesis in MDS. Blood. 2008;112(8):3434-3443. 36. Freyssinier JM, Lecoq-Lafon C, Amsellem S, et al. Purification, amplification and characterization of a population of human erythroid progenitors. Br J Haematol. 1999;106 (4):912-922. 37. Gautier EF, Ducamp S, Leduc M, et al. Comprehensive proteomic analysis of human erythropoiesis. Cell Rep. 2016;16 (5):1470-1484. 38. Inman GJ, Nicolas FJ, Callahan JF, et al. SB431542 is a potent and specific inhibitor of transforming growth factor-beta superfamily type I activin receptor-like kinase (ALK) receptors ALK4, ALK5, and ALK7. Mol Pharmacol. 2002;62(1):65-74. 39. Millot GA, Vainchenker W, Dumenil D, Svinarchuk F. Differential signalling of NH2-terminal flag-labelled thrombopoietin receptor activated by TPO or antiFLAG antibodies. Cell Signal. 2004;16(3): 355-363. 40. Amsellem S, Ravet E, Fichelson S, Pflumio F, Dubart-Kupperschmitt A. Maximal lentivirus-mediated gene transfer and sustained transgene expression in human hematopoietic primitive cells and their progeny. Mol Ther. 2002;6(5):673-677. 41. Papageorgis P, Lambert AW, Ozturk S, et al. Smad signaling is required to maintain epigenetic silencing during breast cancer progression. Cancer Res. 2010;70(3):968-978. 42. Lecoq-Lafon C, Verdier F, Fichelson S, et al. Erythropoietin induces the tyrosine phosphorylation of GAB1 and its association with SHC, SHP2, SHIP, and phosphatidylinositol 3-kinase. Blood. 1999;93(8):25782585. 43. Merryweather-Clarke AT, Atzberger A, Soneji S, et al. Global gene expression analysis of human erythroid progenitors. Blood. 2011;117(13):e96-108. 44. An X, Schulz VP, Li J, et al. Global transcriptome analyses of human and murine terminal erythroid differentiation. Blood. 2014;123(22):3466-3477. 45. Keller MA, Addya S, Vadigepalli R, et al. Transcriptional regulatory network analysis of developing human erythroid progenitors reveals patterns of coregulation and potential transcriptional regulators. Physiol Genomics. 2006;28(1):114-128. 46. Pellagatti A, Cazzola M, Giagounidis AA, et al. Gene expression profiles of CD34+ cells in myelodysplastic syndromes: involvement of interferon-stimulated genes and correlation to FAB subtype and karyotype. Blood. 2006;108(1):337-345. 47. Gerstung M, Pellagatti A, Malcovati L, et al. Combining gene mutation with gene expression data improves outcome prediction in myelodysplastic syndromes. Nat Commun. 2015;6:5901. 48. Ravichandran M, Lei R, Tang Q, et al. Rinf regulates pluripotency network genes and Tet enzymes in embryonic stem cells. Cell Rep. 2019;28(8):1993-2003. 49. Kim HY, Yang DH, Shin SW, et al. CXXC5 is a transcriptional activator of Flk-1 and mediates bone morphogenic proteininduced endothelial cell differentiation and vessel formation. FASEB J. 2014;28(2):615626. 50. Lee SH, Seo SH, Lee DH, Pi LQ, Lee WS, Choi KY. Targeting of CXXC5 by a competing peptide stimulates hair regrowth and wound-induced hair neogenesis. J

haematologica | 2022; 107(1)

RINF maintenance of SMAD7 sustains human erythropoiesis

Invest Dermatol. 2017;137(11):2260-2269. 51. Akel S, Petrow-Sadowski C, Laughlin MJ, Ruscetti FW. Neutralization of autocrine transforming growth factor-beta in human cord blood CD34(+)CD38(-)Lin(-) cells promotes stem-cell-factor-mediated erythropoietin-independent early erythroid progenitor development and reduces terminal differentiation. Stem Cells. 2003;21(5):557567. 52. Gao X, Lee HY, da Rocha EL, et al. TGFbeta inhibitors stimulate red blood cell production by enhancing self-renewal of BFUE erythroid progenitors. Blood. 2016;128 (23):2637-2641. 53. Thompson NL, Flanders KC, Smith JM, Ellingsworth LR, Roberts AB, Sporn MB. Expression of transforming growth factorbeta 1 in specific cells and tissues of adult and neonatal mice. J Cell Biol. 1989;108 (2):661-669. 54. Carrancio S, Markovics J, Wong P, et al. An activin receptor IIA ligand trap promotes erythropoiesis resulting in a rapid induction of red blood cells and haemoglobin. Br J Haematol. 2014;165(6):870-882. 55. Bewersdorf JP, Zeidan AM. Transforming growth factor (TGF)-beta pathway as a therapeutic target in lower risk myelodysplastic syndromes. Leukemia. 2019;33(6): 1303-1312. 56. Fenaux P, Kiladjian JJ, Platzbecker U.

haematologica | 2022; 107(1)

Luspatercept for the treatment of anemia in myelodysplastic syndromes and primary myelofibrosis. Blood. 2019;133(8):790-794. 57. Suragani RN, Cawley SM, Li R, et al. Modified activin receptor IIB ligand trap mitigates ineffective erythropoiesis and disease complications in murine beta-thalassemia. Blood. 2014;123(25):3864-3872. 58. Lio CJ, Yuita H, Rao A. Dysregulation of the TET family of epigenetic regulators in hematopoietic malignancies. Blood. 2019;134 (18): 1487-1497. 59. Yan H, Wang Y, Qu X, et al. Distinct roles for TET family proteins in regulating human erythropoiesis. Blood. 2017;129 (14):2002-2012. 60. Gelsi-Boyer V, Trouplin V, Adelaide J, et al. Mutations of polycomb-associated gene ASXL1 in myelodysplastic syndromes and chronic myelomonocytic leukaemia. Br J Haematol. 2009;145(6):788-800. 61. Wei L, Zhao S, Wang G, et al. SMAD7 methylation as a novel marker in atherosclerosis. Biochem Biophys Res Commun. 2018;496(2):700-705. 62. Yang Q, Chen HY, Wang JN, et al. Alcohol promotes renal fibrosis by activating Nox2/4-mediated DNA methylation of Smad7. Clin Sci (Lond). 2020;134(2):103122. 63. Bian EB, Huang C, Wang H, et al. Repression of Smad7 mediated by DNMT1

determines hepatic stellate cell activation and liver fibrosis in rats. Toxicol Lett. 2014;224(2):175-185. 64. Itzykson R, Kosmider O, Cluzeau T, et al. Impact of TET2 mutations on response rate to azacitidine in myelodysplastic syndromes and low blast count acute myeloid leukemias. Leukemia. 2011;25(7):11471152. 65. Yue L, Bartenstein M, Zhao W, et al. Efficacy of ALK5 inhibition in myelofibrosis. JCI insight. 2017;2(7):e90932. 66. Zhang H, Kozono DE, O'Connor KW, et al. TGF-beta inhibition rescues hematopoietic stem cell defects and bone marrow failure in Fanconi anemia. Cell Stem Cell. 2016;18(5):668-681. 67. Piga A, Perrotta S, Gamberini MR, et al. Luspatercept improves hemoglobin levels and blood transfusion requirements in a study of patients with beta-thalassemia. Blood. 2019;133(12):1279-1289. 68. Verma A, Suragani RN, Aluri S, et al. Biological basis for efficacy of activin receptor ligand traps in myelodysplastic syndromes. J Clin Invest. 2020;130(2):582-589. 69. Cheng W, Wang F, Feng A, Li X, Yu W. CXXC5 attenuates pulmonary fibrosis in a bleomycin-induced mouse model and MLFs by suppression of the CD40/CD40L pathway. Biomed Res Int. 2020;2020: 7840652.

283

LETTERS TO THE EDITOR Diabetes mellitus and risk of plasma cell and lymphoproliferative disorders in 94,579 cases and 368,348 matched controls The prevalence of co-occurring diabetes mellitus (DM) and cancer is increasing worldwide. Small studies have shown an association between overall plasma cell and lymphoproliferative disorders (LPD) and DM. We evaluated the association between DM and nine LPD as well as unspecified amyloidosis from a population-based matched case-control study in Sweden, including 94,579 cases and 368,348 controls. We found a significant increase of LPD diagnoses within 6 months of DM diagnosis, but >6 months after DM diagnosis, DM remained associated with increased risk of acute lymphoblastic leukemia (ALL; odds ratio [OR]: 1.47; 95% confidence interval [CI]: 1.04-2.06), chronic lymphocytic leukemia (CLL; OR: 1.18; 95% CI: 1.1-1.25), and amyloidosis (OR: 1.62; 95% CI: 1.44-1.83), as well as decreased risk of Waldenström macroglobulinemia (WM; OR: 0.77; 95% CI: 0.65-0.91). DM was not associated with increased risk of monoclonal gammopathy of undetermined significance (MGUS) after controlling for medical visits (OR: 0.99; 95% CI: 0.92-1.07) or with MGUS progression. Our findings show that DM association is time-dependent and suggest that some previously observed associations may be disease-specific or caused by detection bias. In contrast, our results suggest that there may be a biological mechanism for the association between DM and ALL. Epidemiologic studies suggest patients with DM are at a higher risk for many cancers.1-3 Smaller epidemiologic studies and a meta-analysis have shown an association between DM and LPD, including plasma cell disorders, suggesting increased risk for multiple myeloma (MM), leukemia, and non-Hodgkin lymphoma (NHL).1,3-6 One study noted this risk only in CLL.7 However, an association with DM has not yet been studied in related disorders such as MGUS, amyloidosis (unspecified including AL amyloidosis [AL] and related amyloid disorders), WM, Hodgkin lymphoma (HL), hairy cell leukemia (HCL), ALL, and T-cell malignancies, including T-cell lymphoma and mycosis fungoides (TCL). To our knowledge, this is the largest population-based study to evaluate this risk of specific LPD in relation to prior DM diagnosis. In the main analysis, we conducted a population-based

matched case-control study to evaluate the impact of preceding DM on the development of LPD in adults. We included cases of MM, WM, HL, NHL, TCL, HCL, ALL, and CLL in the Swedish Cancer Registry and cases of amyloidosis (International Classification of Diseases [ICD] codes E85.8, E85.9) and CLL in the Swedish National Patient Registry from 1987 to 2013. Individuals with MGUS were acquired from a network of Swedish hematology and oncology centers and the Swedish National Patient Registry as previsouly described.8 For each case, up to four controls matched by age, sex, and county of residence were included from the general population. Cases with no matching controls were excluded. DM diagnosis was acquired from the Swedish National Patient Registry (ICD E10, E11, E12, E13, E14). We first assessed the risk of LPD at any time after DM diagnosis. Second, because DM may be diagnosed or registered when patients seek care for the symptoms of LPD — which often have considerable diagnostic delay — we assessed the association of LPD diagnosis with DM diagnosis ≤6 months or >6 months before the LPD diagnosis. We also included a sensitivity analysis with a cut-off at 12 months. Conditional logistic regression, conditioned on the matching variables, was performed to estimate odds ratios. A sensitivity analysis was performed for individuals with MGUS. Because MGUS is asymptomatic, it is often diagnosed during medical visits for other disorders, and detection bias of MGUS and DM may result from DM or MGUS follow-up. We restricted the analysis to years with available outpatient visit data (≥2001); excluded participants with no clinical encounters (visit or admission) <1 year before the matched MGUS case diagnosis; and used logistic regression adjusting for number of visits in the year preceding inclusion, in addition to adjusting for sex, age, and year of inclusion. In a secondary analysis, we assessed the risk of progression from MGUS to MM, WM, amyloidosis, or other LPD. Participants were followed from the date of MGUS diagnosis until diagnosis of LPD. In order to avoid immortal time bias we included DM as a time-dependent covariate in a Cox-model adjusting for age, sex, and year of MGUS diagnosis. All analyses were performed in R (v3.6.3; R Core Team, 2020) using the survival and survminer packages. The study was conducted in accordance with the principles of the Declaration of Helsinki.

Table 1. Prevalence of diabetes mellitus preceding the diagnosis of lymphoproliferative disorders compared to matched controls.

Diagnosis

LPD (n) case control

DM diagnosed before LPD DM DM OR 95% CI case control

HCL HL NHL TCL MM ALL WM MGUS# CLL Amyloidosis

890 4,402 32,898 1,784 15,275 1,198 3,380 19,172 18,422 5,373

5.1% 4.2% 7.2% 7% 8.1% 6.5% 7.6% 10% 9.3% 9.6%

3,515 17,410 128,896 7,012 59,695 4,738 13,260 75,032 71,796 21,200

5% 3% 5.9% 5.3% 6.4% 3% 7.4% 6.8% 6.7% 5.4%

1.03 1.46 1.23 1.35 1.30 2.35 1.03 1.55 1.44 1.93

0.73-1.44 1.23-1.74*** 1.17-1.29*** 1.09-1.67** 1.22-1.39*** 1.75-3.15*** 0.89-1.19 1.46-1.64*** 1.36-1.53*** 1.72-2.15***

DM diagnosed ≤6 months before LPD DM DM OR 95%CI case control 1.3% 0.9% 1.4% 1.3% 2.3% 2.5% 2.2% 1.8% 2% 1.8%

0.4% 0.2% 0.4% 0.3% 0.4% 0.1% 0.5% 0.4% 0.4% 0.4%

3.65 5.81 4.08 3.72 5.90 19.19 4.04 4.37 4.82 5.33

1.67-8.00** 3.59-9.4*** 3.58-4.64*** 2.07-6.66*** 4.98-6.98*** 7.97-46.22*** 2.91-5.6*** 3.74-5.11*** 4.14-5.61*** 3.93-7.21***

DM diagnosed >6 months before LPD DM DM OR 95% CI case control 3.7% 3.3% 5.7% 5.7% 5.9% 4% 5.4% 8.2% 7.3% 7.8%

4.6% 2.8% 5.6% 4.9% 6% 2.8% 6.8% 6.3% 6.2% 5%

0.8 1.19 1.03 1.16 0.97 1.47 0.77 1.33 1.18 1.62

0.54-1.18 0.98-1.44 0.97-1.08 0.93-1.47 0.9-1.05 1.04-2.06* 0.65-0.91* 1.25-1.41*** 1.1-1.25*** 1.44-1.83***

Refer to text for MGUS adjustment per sensitivity analysis.*P<0.05; **P<0.01; ***P<0.001. DM: diabetes mellitus; LPD: ymphoproliferative disorders; HCL: hairy cell leukemia; HL: Hodgkin lymphoma; NHL: non-Hodgkin lymphoma; TCL: T-cell lymphoma; MM: multiple myeloma; ALL: acute lymphoblastic leukemia; WM: Waldenström macroglobulinemia; MGUS: monoclonal gammopathy of undetermined significance; CLL: chronic lymphocytic leukemia; OR: odds ratio; CI: confidence interval.

284

haematologica | 2022; 107(1)

Letters to the Editor

Patients with MGUS, MM, amyloidosis, HL, NHL, TCL, ALL, and CLL were more likely to have a preceding DM diagnosis compared to matched controls, whereas patients with WM and HCL were not. These results remained significant for MGUS, amyloidosis, ALL, and CLL when DM was diagnosed >6 months prior to the LPD (Table 1). Large Israeli and Canadian studies showed an attenuated long-term risk for MM, NHL, and leukemia >1 year after DM diagnosis, although the association remained significant.3,4 Our study had substantially more cases. Other studies that have not evaluated time-dependent risk showed an overall association of DM with MM, NHL, and CLL.5-7 Prior data suggesting that maternal DM increases childhood ALL risk in offspring indicate biologic mechanisms behind this long-term increased risk of ALL associated with DM, thus warranting further study in adults.9 Diagnosis of DM >6 months prior was associated with decreased risk of development of WM (OR: 0.77; 95% CI: 0.65-0.91; P<0.05), which is unique compared to all other LPD and has not been reported before. This may be related to the protective effect of anti-diabetic drugs10 although no data is available on their effect in patients with WM. However, further research is needed to expand on this finding of the study. Risk of MGUS and CLL remained increased after excluding DM diagnoses <6 months prior. These disorders are often asymptomatic, particularly MGUS, and their diagnosis might be associated with medical followup. In a sensitivity analysis controlling for the number of clinic visits, we found that MGUS was not increased in DM (OR: 0.99; 95% CI: 0.92-1.07; P=0.89), indicating that detection bias during medical follow-up likely explains the association we had initially observed in MGUS. CLL can present both as an asymptomatic or symptomatic disease, meaning that a similar analysis for CLL would be difficult to interpret. However, a large proportion of CLL patients are asymptomatic, so a similar detection bias in part may be reasonably conjectured in CLL. Long-term risk of amyloidosis was also increased in DM (OR: 1.62; 95% CI: 1.44-1.83; P<0.001). DM is a chronic inflammatory disorder associated with amyloid deposition in pancreatic islets11 and kidneys.12 However, these sites are not routinely biopsied, so these findings are likely to be incidental and underrepresented by international classification of diseases (ICD) codes. This suggests that the increased risk of AL may be related to DM. Light chain glycosylation is associated more frequently with AL than with other LPD13 and DM is associated with higher incidence of protein glycosylation,14 suggesting a possible mechanism for the role of DM in the development of AL. Short-term risk of cancer diagnosis after DM diagnosis may result from detection bias, particularly because risk factors (aging, obesity, physical activity, diet, alcohol, and smoking) are common to both DM and cancer. In symptomatic conditions, patients may have DM diagnosed <6 months prior through clinical encounters for similar symptoms of fatigue and weight loss. In asymptomatic disorders such as MGUS and CLL, clinical encounters for DM may lead to increased rates of LPD diagnosis. Alternatively, a direct biological link via hyperinsulinemia (insulin like growth factor-1), chronic inflammation (cytokines), or hyperglycemia (epigenetic changes or protein glycation/glycosylation) may lead to development or acceleration of early-stage cancer, particularly as patients may have prediabetes or undiagnosed DM for many years.15 Patients with DM for >6 months are likely to be haematologica | 2022; 107(1)

Table 2. Monoclonal gammopathy of undetermined significance progression to more advanced disease by diabetes mellitus state.

n Median age (years) Age range (years) male (%) Diagnosed with DM after MGUS Progressed n (%) MM WM Amyloidosis unspecified Other LPD Cox models (no DM as reference) Progressed (overall) MM WM Amyloidosis unspecified Other LPD

Diabetes mellitus

No diabetes mellitus

2,224 73 26-95 60 739 (33%) 220 (9.9%) 137 (6.2%)# 35 (1.6%) 14 (0.6%)# 35 (1.6%)

12,563 72 18-99 52 2,031 (16.2%) 1,115 (8.9%)# 380 (3.0%) 153 (1.2%)# 384 (3.1%)

0.89 (95% CI: 0.78-1.03); P=0.11 1.06 (95% CI: 0.88-1.26); P=0.54 0.72 (95% CI: 0.51-1.02); P=0.06 0.74 (95% CI: 0.43-1.29); P=0.29 0.70 (95% CI: 0.50-1.00); P=0.047

These numbers include two individuals - one with diabetes mellitus (DM) with concomitant multiple myeloma (MM) and amyloidosis and one without DM with concomitant Waldenström macroglobulinemia (WM) and amyloidosis diagnosed on the same day. They have been counted once for each diagnosis. MGUS: monoclonal gammopathy of undetermined significance; LPD: lymphoproliferative disorder.

on anti-diabetic medications including metformin, which protects against MGUS progression,10 and insulin, which does not increase the risk for NHL and MM compared to untreated diabetics.6 Sensitivity analysis using a 12 month cut-off for time between DM and LPD diagnosis showed essentially the same results (data not shown). Patients with DM are not more likely to progress from MGUS to MM, WM, amyloidosis, or other LPD (Table 2). However, we could not control for the use of anti-diabetic drugs that may lower rates of MGUS progression.10 Interestingly, we found decreased risk of progression from MGUS to WM of borderline significance (P=0.06) which is consistent with the findings of the main analysis showing an association of DM and decreased risk of WM. The large sample size to detect differences in rare diseases is a major strength of this study. Limitations include lack of granular information related to DM (including DM subtypes), body mass index, LPD, and race/ethnicity, as well as a likely lack of racial/ethnic diversity in this population. Also, the diagnoses are based on ICD codes, relying on the correct registration by physicians. This is the largest population-based study of preceding DM diagnosis on the time-dependent risk of individual LPD, showing there is an attenuated long-term risk with ALL, CLL, and amyloidosis. Furthermore, DM was not associated with MGUS progression. Urvi A. Shah,1,* Sæmundur Rögnvaldsson,2,* Andriy Derkach,3 Magnus Björkholm,4 Ingemar Turesson,5 Yael David,6 Malin Hultcrantz,1 Carlyn Tan,1 Hani Hassoun,1 Neha Korde,1 Alexander Lesokhin,1 Sham Mailankody,1 Sigurður Yngvi Kristinsson2,4,# and C. Ola Landgren7,# 1 Department of Medicine, Myeloma Service, Memorial Sloan Kettering Cancer Center, New York, NY, USA; 2Department of Medicine, University of Iceland, Reykjavík, Iceland; 3Department of Epidemiology and Biostatistics, Memorial Sloan Kettering Cancer

285

Letters to the Editor

Center, New York, NY, USA; 4Department of Medicine, Karolinska University Hospital and Karolinska Institutet, Stockholm, Sweden; 5 Department of Hematology, Skane University Hospital Malmö/Lund, Malmö/Lund, Sweden; 6Department of Chemical Biology, Sloan Kettering Institute, New York, NY, USA and 7Department of Medicine, Sylvester Comprehensive Cancer Center, University of Miami, Miami, FL, USA *UAS and SR contributed equally as co-first authors #

SYK and OL contributed equally as co-senior authors

Correspondence: URVI A. SHAH - shahu@mskcc.org doi:10.3324/haematol.2021.278772 Received: March 16, 2021. Accepted: August 26, 2021. Pre-published: September 2, 2021. Disclosures: UAS has received research funding from Celgene/Bristol Myers Squibb and Janssen, to her institution and honorariums for continuing medical education activity from the Physicians Education Resource, all outside of the submitted work. M.B. reports the following, all outside of the submitted work: research funding from Takeda; member of independent data monitoring committee for Mundipharma; lecture honoraria from Roche, Pfizer and BMS, Grant committee work for Incyte. COL has received grant support from: LLS, Rising Tide Foundation, Memorial Sloan Kettering Cancer Center, NIH, FDA, MMRF, IMF, Perelman Family Foundation, Amgen, Celgene, Janssen, Takeda, Glenmark, Seattle Genetics, Karyopharm; has received honoraria for scientific talks/participated in advisory boards for: Adaptive, Amgen, Binding Site, BMS, Celgene, Cellectis, Glenmark, Janssen, Juno, Pfizer; and served on Independent Data Monitoring Committees (IDMC) for international randomized trials by: Takeda, Merck, Janssen, Theradex. Presented in abstract form at the 62nd annual meeting of the American Society of Hematology, 6 December 2020. Contributions: all authors participated in the writing and review of the manuscript. Additionally UAS and SR designed the study and performed data analysis; AD performed data analysis and designed statistical methods; MB and IT curated the data and were involved in study design conception; YD, SP, MH, CT, HH, NK, AL, and SM were involved in study design conception; SYK and COL curated the data, designed the study, and supervised the research project. Acknowledgments: the authors would like to thank Hannah Rice, BA, ELS for editorial assistance. Funding: this research was funded in part through the NIH/NCI Cancer Center Support Grant P30 CA008748 (UAS) and supported by grants from Swedish Cancer Society (MB), Parker Institute of Cancer Immunotherapy Career Development Award (YD, UAS), International Myeloma Society Career Development Award, Paula

286

and Rodger Riney Foundation, American Society of Hematology Clinical Research Training Institute Award and TREC Training Workshop R25CA203650 (PI: Melinda Irwin) (UAS). Data sharing statement: The data included in study is built on multiple population-based registries including the medical records of Swedish citizens. Although the data has been made unidentifiable before analysis, ethical board approval does not permit data sharing.

References 1. Mitri J, Castillo J, Pittas AG. Diabetes and risk of non-Hodgkin's lym-

phoma: a meta-analysis of observational studies. Diabetes Care. 2008;31(12):2391-2397. 2. Pearson-Stuttard J, Bennett J, Cheng YJ, et al. Trends in predominant causes of death in individuals with and without diabetes in England from 2001 to 2018: an epidemiological analysis of linked primary care records. Lancet Diabetes Endocrinol. 2021;9(3):165-173. 3. Gong IY, Cheung MC, Read S, Na Y, Lega IC, Lipscombe LL. Association between diabetes and haematological malignancies: a population-based study. Diabetologia. 2021;64(3):540-551. 4. Dankner R, Boffetta P, Balicer RD, et al. Time-dependent risk of cancer after a diabetes diagnosis in a cohort of 2.3 million adults. Am J Epidemiol. 2016;183(12):1098-1106. 5. Castillo JJ, Mull N, Reagan JL, Nemr S, Mitri J. Increased incidence of non-Hodgkin lymphoma, leukemia, and myeloma in patients with diabetes mellitus type 2: a meta-analysis of observational studies. Blood. 2012;119(21):4845-4850. 6. Fortuny J, Benavente Y, Bosch R, Garcia-Villanueva M, de Sevilla AF, de Sanjose S. Type 2 diabetes mellitus, its treatment and risk for lymphoma. Eur J Cancer. 2005;41(12):1782-1787. 7. Khan AE, Gallo V, Linseisen J, et al. Diabetes and the risk of nonHodgkin's lymphoma and multiple myeloma in the European Prospective Investigation into Cancer and Nutrition. Haematologica. 2008;93(6):842-850. 8. Kristinsson SY, Bjorkholm M, Goldin LR, McMaster ML, Turesson I, Landgren O. Risk of lymphoproliferative disorders among firstdegree relatives of lymphoplasmacytic lymphoma/Waldenstrom macroglobulinemia patients: a population-based study in Sweden. Blood 112(8):3052-6. 9. Yan P, Wang Y, Yu X, Liu Y, Zhang ZJ. Maternal diabetes and risk of childhood malignancies in the offspring: a systematic review and meta-analysis of observational studies. Acta Diabetol. 2021; 58(2):153-168. 10. Chang SH, Luo S, O'Brian KK, et al. Association between metformin use and progression of monoclonal gammopathy of undetermined significance to multiple myeloma in US veterans with diabetes mellitus: a population-based retrospective cohort study. Lancet Haematol. 2015;2(1):e30-36. 11. Ueberberg S, Nauck MA, Uhl W, et al. Islet amyloid in patients with diabetes due to exocrine pancreatic disorders, type 2 diabetes, and nondiabetic patients. J Clin Endocrinol Metab. 2020;105(8):dgaa176. 12. Diez R, Madero M, Gamba G, Soriano J, Soto V. Renal AA amyloidosis in patients with type 2 diabetes mellitus. Nephron Extra. 2014; 4(2):119-126. 13. Omtvedt LA, Bailey D, Renouf DV, et al. Glycosylation of immunoglobulin light chains associated with amyloidosis. Amyloid. 2000;7(4):227-244. 14. Reily C, Stewart TJ, Renfrow MB, Novak J. Glycosylation in health and disease. Nat Rev Nephrol. 2019;15(6):346-366. 15. Huang Y, Cai X, Qiu M, et al. Prediabetes and the risk of cancer: a meta-analysis. Diabetologia. 2014;57(11):2261-2269.

haematologica | 2022; 107(1)

Letters to the Editor

In classical Hodgkin lymphoma the combination of the CCR5 antagonist maraviroc with trabectedin synergizes, enhances DNA damage and decreases three-dimensional tumor-stroma heterospheroid viability Several preclinical and clinical studies have demonstrated a key role of the CCL5/CCR5 axis in cancer, providing the rationale for clinical trials with the CCR5 antagonist maraviroc.1 Indeed, maraviroc, approved by the Food and Drug Administration as an antiretroviral for the treatment of human immunodeficiency virus infections, has been repurposed as a potential therapeutic option for cancer treatment since it was demonstrated to inhibit tumor growth, metastasis formation, and the creation of a tumor-promoting microenvironment (TME).2,3

In breast cancer, CCR5+ tumor cells are characterized by increased levels of DNA repair genes and maraviroc enhanced both the cell killing and the DNA damage mediated by the DNA-damaging agent doxorubicin.4 Tumor cells from classical Hodgkin lymphoma (cHL) (the Hodgkin and Reed-Sternberg cells, HRS) secrete CCL5,5 and express a functional CCR5 receptor,6 and patients with high CCL5 levels have a poor prognosis.5 CCR5 is activated in both autocrine6 and paracrine5 manners by CCR5-ligands secreted by HRS cells, monocytes and cHL-mesenchymal stromal cells (MSC). Maraviroc decreases HRS cell growth in vitro and in tumor xenografts, synergizes with doxorubicin and brentuximab vedotin and decreases the self aggregation of HRS cells, monocytes and MSC in three-dimensional (3D) heterospheroids, used as an in vitro model to study the TME.5 Trabectedin covalently binds DNA, blocking transcrip-

Figure 1. Maraviroc synergized with trabectedin in Hodgkin and Reed-Sternberg cells. L-1236, L-428, KM-H2 and HDLM-2 cells (2.0x105 cells/mL), were exposed to increasing concentrations of maraviroc, trabectedin and their combination. (A) After 72 h cell viability was evaluated by trypan blue dye exclusion. Results (percentage of control) are mean and standard deviation of three independent experiments each run in triplicate. (B) Synergy was determined using Calcusyn software. Values of bar charts are mean combination index values and standard deviation of three experiments each run in triplicate. MVC: maraviroc; TB: trabectedin.

haematologica | 2022; 107(1)

287

Letters to the Editor

Figure 2. Legend on following page.

288

haematologica | 2022; 107(1)

Letters to the Editor

Figure 2. Maraviroc enhanced DNA damage induced by trabectedin. Hodgkin and Reed-Sternberg (HRS) cells were treated with maraviroc (100 mM), trabectedin (500 pM) and their combination for 24 h. (A) Western blot for g-H2AX and a-tubulin protein expression in cHL cell lines. Membrane strips were incubated with mouse anti-phospho-histone H2A.X (Ser139) (clone JBW301) (Millipore) and mouse anti-a-tubulin antibody clone B-5-1-2 (Sigma Aldrich) and revealed with donkey anti-mouse IgG (H+L chain) A90-137P (Bethyl Laboratories). Images were acquired using a ChemiDoc XRS system (Bio-Rad). Data are representative of four experiments. (B) Bar charts showing densitometric analysis of g-H2AX expression normalized to a-tubulin as a loading control. Protein quantification was performed using ImageJ software. Values are means and standard deviation of four experiments. *P<0.05, treatments vs. control, One-way analysis of variance followed by the Dunnett test. °P<0.05, trabectedin vs. trabectedin and maraviroc in combination, Student t-test. (C) Immunofluorescence images (confocal microscopy) of g-H2AX foci after drug treatment. HRS cells adherent to coverslips, were fixed, permeabilized and incubated with anti-phospho-histone H2A.X (Ser139) (clone JBW301) (Millipore), followed by Alexa Fluor-488 anti-mouse secondary antibody (Thermo Scientific). Images were acquired with a Leica TCS SP8 Confocal system (Leica Microsystems Heidelberg, Mannheim, Germany), using Leica Confocal Software (LCS). MVC: maraviroc; TB: trabectedin.

tion and interfering with DNA repair, and modifies the TME by selectively reducing tumor-associated macrophages.7 Its anticancer activities have been demonstrated in solid tumors and in preclinical models of hematologic malignancies.8 In cHL, trabectedin induces a potent in vitro antitumor activity, decreases cytokine levels and viability of heterospheroids formed by HRS cells and MSC, and inhibits tumor xenograft growth, monocyte infiltration and angiogenesis.9 Considering that HRS cells express a functional CCR5 receptor and maraviroc synergizes with doxorubicin and affects TME interactions,5 our hypothesis is that this CCR5 antagonist, by promoting DNA damage and disrupting the cross-talk between tumor cells and MSC in 3D heterospheroids could potentiate the cytotoxic effects of trabectedin. Here we found that maraviroc in combination with trabectedin exerted synergistic effects, enhanced DNA double-strand breaks, and cooperated to decrease the viability of 3D tumor-stroma-heterospheroids. In this study we used a panel of four authenticated cHL-derived cell lines L-1236, L-428, KM-H2, and HDLM-2 (obtained from the DSMZ, Germany), and cHLMSC obtained from frozen lymph nodes and generated as previously described.5 The effects of drug combinations were evaluated using the Chou-Talalay method, calculating the combination index (CI) with CalcuSyn software (Biosoft, Ferguson, MO, USA).5 CI values <0.9 indicate synergy, the lower the value the stronger the synergism.5,9 Phosphorylation of the H2AX histone (g-H2AX) and g-H2AX-foci formation were used to study the lethal double-strand DNA lesions.10 The combination of maraviroc (Sigma-Aldrich) with trabectedin (PharmaMar) was also tested in 3D heterospheroids. Heterospheroids were generated by co-culturing HRS cells and cHL-MSC (1.0 x 104/mL of each cell type) in RPMI-1640 medium containing 2% fetal calf serum, using plates coated with 20 mg/mL poly-HEMA (Sigma) to prevent cell adhesion.5 Statistical analysis was carried out using GraphPad Prism version 6.0 software (GraphPad, La Jolla, CA, USA). A Student t-test was used to compare two groups and one-way analysis of variance, followed by the Dunnett test, to compare each of a number of treatments with a single control. A P-value <0.05 was considered statistically significant. First, we performed drug-combination studies with maraviroc and trabectedin using CCR5+ cHL-derived cell lines (L-1236, L-428, KM-H2, HDLM-2). Maraviroc reduced the CCR5-CCL5 mediated autocrine growth6 of HRS cells in a dose-dependent manner (Figure 1A), without an apparent correlation with CCR5 expression (Online Supplementary Figure S1A, B) or levels of secreted CCL5 (Online Supplementary Figure S1C). The combination of maraviroc with trabectedin resulted in a strong synergism (i.e., an interaction between two or more drugs that determines a greater effect than the sum of the individual effects of each drug) in KM-H2 (CI <0.3) and haematologica | 2022; 107(1)

in HDLM-2 cells (CI <0.3), and a moderate synergism in L-1236 cells (CI ranging from 0.47 to 0.63) (Figure 1B). In L-428 cells, we observed synergy only at low drug concentrations (CI ranging from 0.57 to 0.98) (Figure 1B). Genetic lesions in members of the NF-kB and JAK/STAT pathways or TP53 alterations could be a possible explanation for the lower synergistic effects in L-428 cells.11 We then evaluated DNA fragmentation in HRS cells treated with maraviroc alone, trabectedin alone, or their combination. Phosphorylation of the histone H2AX (g-H2AX) is one of the early events associated with the detection and processing of DNA double-strand breaks10 and the formation of g-H2AX-foci, arising when the cell identifies DNA lesions. For this purpose, we analyzed g-H2AX induction and g-H2AX-foci formation after a brief period of incubating cells with the various drugs. Western blot assay showed that treatment of HRS cells with maraviroc alone did not induce or only slightly induced DNA fragmentation. On the other hand, trabectedin induced g-H2AX, an effect that was greatly enhanced by the combination with maraviroc (Figure 2A, B). To further validate that maraviroc in combination with trabectedin enhanced DNA damage, we evaluated the formation of nuclear g-H2AX foci by confocal microscopy.10 Consistently with the findings of the western blot assay (Figure 2A, B), after treatment with maraviroc only rare g-H2AX foci were detected, whereas the combination with maraviroc enhanced the formation of g-H2AX-foci by trabectedin (Figure 2C), confirming that the CCR5-antagonist in combination with trabectedin further promoted DNA fragmentation (Figure 2). The strong cytotoxic effects of the drug combination with respect to the limited activity of single treatments were confirmed by phase contrast photographs and by annexin-V/7-aminoactinomycin D staining (Online Supplementary Figure S2). Although maraviroc alone induced a very modest or no DNA fragmentation in HRS cells, unlike in breast cancer cells,4 we cannot exclude a possible role of CCR5 in enhancing DNA repair also in cHL. Alternatively, maraviroc could increase the concentration of trabectedin in HRS cells by competing for the drug transporter P-glycoprotein, a substrate for both drugs.12 In cHL, a few HRS cells are surrounded by a protective and immunosuppressive TME, capable of decreasing drug activity and counteracting the immune control of tumor growth.13 Heterospheroids represent a 3D model in which different cell types are cultured under nonadherent conditions. This in vitro model, developed to mimic the cross-talk of HRS cells with the TME, was recently used to study the antitumor activity of trabectedin alone,9 and of maraviroc in combination with doxorubicin.5 cHL-MSC can exert protective effects against anticancer drugs by their direct contact with tumor cells13,14 or by secreting tumor-promoting molecules, including CCL5.5 Since maraviroc reduces the self-assembly of HRS 289

Letters to the Editor

cells with cHL-MSC,5 it could cooperate with trabectedin not only by enhancing DNA damage in CCR5+ cells but also by counteracting the protective effects of the crosstalk between HRS and cHL-MSC. We evaluated the efficacy of the maraviroc-trabectedin combination in formed

heterospheroids (direct contact). We cultured HRS cells (L-1236 or HDLM-2 cells) with cHL-MSC under nonadherent conditions. After 24 h (the time necessary to obtain cell aggregation as heterospheroids), we added maraviroc and different concentrations of trabectedin9

Figure 3. Maraviroc cooperated with trabectedin to decrease cell viability in three-dimensional heterospheroids formed by Hodgkin and Reed-Sternberg cells and classical Hodgkin lymphoma mesenchymal stromal cells. (A) Schedule of treatment of Hodgkin and Reed-Sternberg (HRS) cells and classical Hodgkin lymphoma mesenchymal stromal cells (cHL-MSC). (B) L-1236, or (C) HDLM-2 cells were cultured in non-adherent conditions with cHL-MSC (1.0 x 104/mL of each cell type) in 24-well plates and treated with trabectedin (360 pM) alone, maraviroc (100 mM) alone or in combination. Drugs were added immediately (add. T= 0) and after heterospheroid formation (add. T=24h). After 6 days, cell viability was evaluated using the PrestoBlue Cell Viability Reagent (Invitrogen). Values are mean and standard deviation of three experiments. °P<0.05 for T=0 vs. T=24h, Student t-test. *P<0.05 treatments vs. medium, One-way analysis of variance followed by the Dunnett test. (D) Representative phase contrast micrographs of heterospheroids cultured with trabectedin, maraviroc and their combination (T=0). MVC, maraviroc; TB, trabectedin. add., added.

290

haematologica | 2022; 107(1)

Letters to the Editor

(Online Supplementary Figure S3). After 3 days (Online Supplementary Figure S3A, B) and 6 days (Online Supplementary Figure S3C, D) of treatment, we evaluated drug effects. Both maraviroc and trabectedin decreased heterospheroid cell viability in a dose-dependent manner and their combination further increased cytotoxicity. These effects were more evident after 6 days of treatment in HDLM-2 and in particular in L-1236 cells (Online Supplementary Figure S3). To demonstrate that the direct contact with cHL-MSC can protect against the effects exerted by maraviroc, trabectedin and their combination, HRS cells and cHL-MSC were cultured under non-adherent conditions and drugs were added before (at T=0) and after (at T=24h) their spontaneous aggregation in heterospheroids (see Figure 3A). After 6 days cell viability of heterospheroids formed by L-1236/cHL-MSC (Figure 3B) and by HDLM-2/cHLMSC (Figure 3C) was evaluated. Both maraviroc and trabectedin alone reduced cell viability of heterospheroids (Figure 3B, C) and their combination was significantly more efficacious than single treatments. Cytotoxicity was more evident when drugs were added before cell aggregation (heterospheroid formation, T=0) (black histograms), suggesting a protective role of HL-MSC and cooperation between the two cell types, including the increased secretion and expression of pro-survival factors.13 Representative phase contrast photographs demonstrating the increased cytotoxic activity of the combination trabectedin-maraviroc in heterospheroids are shown in Figure 3D. Taken together our results suggest that maraviroc can enhance trabectedin activity by reducing the tumor-promoting effects of the direct contact of HRS cells with cHL-MSC13 and by inhibiting autocrine and paracrine effects induced by CCL5 secreted by HRS cells6 and tumor-educated cHL-MSC.5 In conclusion, maraviroc synergized with trabectedin, enhanced trabectedin-induced DNA double-strand breaks and decreased the protective effects of cHL-MSC in heterospheroids. Therefore, this study offers an additional preclinical rationale for the use of maraviroc as a new therapeutic option to affect tumor microenvironmental interactions, to enhance the cytotoxic activity of DNA-damaging agents and, by decreasing their doses, to reduce adverse side effects.5 These findings provide insights for future research to investigate the possibility of potentiating the antitumoral activity of DNA-damaging agents, including gamma radiation, currently used for the treatment of cHL.15 Naike Casagrande, Cinzia Borghese and Donatella Aldinucci Division of Molecular Oncology, Centro di Riferimento Oncologico di Aviano (CRO), IRCCS, Aviano, Italy Correspondence: DONATELLA ALDINUCCI- daldinucci@cro.it

haematologica | 2022; 107(1)

doi:10.3324/haematol.2021.279389 Received: June 29, 2021. Accepted: August 27, 2021. Pre-published: September 9, 2021. Disclosures: no conflicts of interest to disclose. Contributions: NC and CB generated and interpreted the data; DA and NC drafted the manuscript; DA supervised the study. All authors reviewed, revised, and approved the final manuscript. Acknowledgments: the authors thank Dr. Alfonso Colombatti for his help with reviewing the manuscript. Funding: this work was supported in part by grant IG 15844 from the Italian Association for Cancer Research (to DA) and by the Italian Ministry of Health (Ricerca Corrente).

References 1. Aldinucci D, Borghese C, Casagrande N. The CCL5/CCR5 axis in

cancer progression. Cancers (Basel). 2020;12(7):1765. 2. Miao M, De Clercq E, Li G. Clinical significance of chemokine recep-

tor antagonists. Expert Opin Drug Metab Toxicol. 2020;16(1):11-30. 3. Jiao X, Nawab O, Patel T, et al. Recent advances targeting CCR5 for

cancer and its role in immuno-oncology. Cancer Res. 2019; 79(19): 4801-4807. 4. Jiao X, Velasco-Velázquez MA, Wang M, et al. CCR5 governs DNA damage repair and breast cancer stem cell expansion. Cancer Res. 2018;78(7):1657-1671. 5. Casagrande N, Borghese C, Visser L, Mongiat M, Colombatti A, Aldinucci D. CCR5 antagonism by maraviroc inhibits Hodgkin lymphoma microenvironment interactions and xenograft growth. Haematologica. 2019;104(3):564-575. 6. Aldinucci D, Lorenzon D, Cattaruzza L, et al. Expression of CCR5 receptors on Reed-Sternberg cells and Hodgkin lymphoma cell lines: involvement of CCL5/Rantes in tumor cell growth and microenvironmental interactions. Int J Cancer. 2008;122(4):769-776. 7. D’Incalci M, Badri N, Galmarini CM, Allavena P. Trabectedin, a drug acting on both cancer cells and the tumour microenvironment. Br J Cancer. 2014;111(4):646-650. 8. Jimenez PC, Wilke DV, Branco PC, et al. Enriching cancer pharmacology with drugs of marine origin. Br J Pharmacol. 2020;177(1):3-27. 9. Casagrande N, Borghese C, Favero A, Vicenzetto C, Aldinucci D. Trabectedin overcomes doxorubicin-resistance, counteracts tumorimmunosuppressive reprogramming of monocytes and decreases xenograft growth in Hodgkin lymphoma. Cancer Lett. 2021;500:182193. 10. Ruprecht N, Hungerbühler MN, Böhm IB, Heverhagen JT. Improved identification of DNA double strand breaks: γ-H2AX-epitope visualization by confocal microscopy and 3D reconstructed images. Radiat Environ Biophys. 2019;58(2):295-302. 11. Cuceu C, Hempel WM, Sabatier L, Bosq J, Carde P, M’kacher R. Chromosomal Instability in Hodgkin lymphoma: an in-depth review and perspectives. Cancers (Basel). 2018;10(4):91. 12. Tupova L, Ceckova M, Ambrus C, et al. Interactions between maraviroc and the ABCB1, ABCG2, and ABCC2 transporters: an important role in transplacental pharmacokinetics. Drug Metab Dispos. 2019;47(9):954-960. 13. Aldinucci D, Celegato M, Casagrande N. Microenvironmental interactions in classical Hodgkin lymphoma and their role in promoting tumor growth, immune escape and drug resistance. Cancer Lett. 2016;380(1):243-252. 14. Bankov K, Döring C, Ustaszewski A, et al. Fibroblasts in nodular sclerosing classical Hodgkin lymphoma are defined by a specific phenotype and protect tumor cells from brentuximab-vedotin induced injury. Cancers (Basel). 2019;11(11):1687. 15. Connors JM, Cozen W, Steidl C, et al. Hodgkin lymphoma. Nat Rev Dis Primers. 2020;6(1):61.

291

Letters to the Editor

Vecabrutinib inhibits B-cell receptor signal transduction in chronic lymphocytic leukemia cell types with wild-type or mutant Bruton tyrosine kinase Ibrutinib as monotherapy or in combination has transformed the treatment landscape of chronic lymphocytic leukemia (CLL).1,2 This drug covalently tethers to the cysteine-481 residue in Bruton tyrosine kinase (BTK), which is a pivotal enzyme in the B-cell receptor (BCR) pathway.3 Ibrutinib treatment results in long-term overall survival in patients with CLL; however, disease can relapse, particularly in previously treated patients. At a median of 3.4 years of follow-up, the cumulative incidence of progression was 19%, and 85% of these patients had acquired mutations of BTK or PLCG2.4,5 The predominant BTK mutation is cysteine to serine (BTKC481S), with the second most frequent alteration being cysteine to arginine (BTKC481R), both of which preclude covalent bond formation by ibrutinib, resulting in resistance to this drug.4 Reversible BTK inhibitors, such as vecabrutinib, have been developed that bind to BTK and maintain inhibitory activity against WT and mutant BTK. In this study, we characterized the activity of vecabrutinib on the BCR pathway using a CLL cell line model system engineered to overexpress BTK C481 WT or two mutant variants, BTKC481S and BTKC481R. 6 We also used primary CLL cells resistant to ibrutinib to further extend our investigations of vecabrutinib. A phase 1b clinical trial was recently completed in B-cell malignancies in which vecabrutinib was well-tolerated with some evidence of activity including in CLL patients with the C481S mutation.7 Vecabrutinib is a highly selective reversible BTK inhibitor (half maximal inhibitory concentration [IC50] = 3 nM). In a panel of 234 kinases and kinase variants, vecabrutinib demonstrated a biochemical IC50 of <100 nM for seven kinases (Figure 1A; Online Supplementary Figure S1A, B). The IC50 of vecabrutinib against WT BTK was similar to that of ibrutinib3 while it was more potent than ibrutinib on ITK and TEC kinases. In a direct kinase assay, vecabrutinib inhibited WT and mutant C481S variants with similar potency. Data from healthy donors’ whole blood (n=145) further established the potency of vecabrutinib in inhibiting BTK, although there was a high degree of variability (mean ± standard deviation values, 50 nM ± 39 nM, range 2.8 nM – 216 nM) (Figure 1B). Vecabrutinib inhibited phosphorylation of the BTK downstream target PLCg2 in Ramos Burkitt lymphoma cells with IC50 values of 13 ± 6 nM (Figure 1C). Collectively, these data suggest that vecabrutinib inhibits phosphorylation of BTK and of PLCg2 at nanomolar concentrations. In MEC-1, a CLL cell line, treatment with both vecabrutinib and ibrutinib did not alter cell cycle profiles and resulted in 15-20% cell death at 1 mM (Online Supplementary Figure S1C-E). Vecabrutinib decreased BTK phosphorylation at a dose of 0.1 mM. Consistent with the observed decline in BTK phosphorylation, decreases in phosphorylation of PLCg2 and ERK were also observed. Phospho-S6 levels did not change in the MEC-1 cell line after treatment with vecabrutinib (Online Supplementary Figure S1F). To mimic ibrutinib resistance, we transduced MEC-1 to generate cell lines that stably expressed green fluorescent protein and overexpressed either WT BTK (BTKWT) or the mutated variant C481S (BTKC481S), or C481R (BTKC481R) BTK.6 Cell death induced by ibrutinib or acalabrutinib has been limited when the drugs have been tested in vitro 292

in B-cell lines or primary CLL cells,8,9 and similarly, vecabrutinib did not affect cell viability or cell cycle profile (Online Supplementary Figure S2A-D). Several known proteins within the BCR signal transduction pathway, including BTK, can be used as biomarkers to monitor ibrutinib response or biological activity, including ERK and S6.8,10 We evaluated these proteins by immunoblot for phosphorylated proteins to show response to vecabrutinib and ibrutinib in cell lines that harbor WT overexpression or mutant BTK overexpression. Vecabrutinib at a dose of 1 mM decreased phospho-ERK more effectively than ibrutinib did in ibrutinib-resistant MEC-1 cells that overexpress mutant BTK (Figure 1D-F). This was observed in both BTKC481S and BTKC481R variants. Changes in phospho-ERK were consistent with those in our previous study, in which we showed that phospho-ERK is a superior biomarker to determine ibrutinib response upon overexpression of mutant BTK in MEC-16 (Figure 1E, F). It is important to note that ibrutinib also decreased phospho-proteins in cells with mutated BTK. There are two explanations for this finding. First, the cell lines express endogenous WT BTK and second, ibrutinib can bind reversibly to BTK, albeit with reduced potency. In the clinic, ibrutinib’s poor pharmacokinetic properties (peak plasma level along with initial and terminal elimination half-lives) preclude activity as a noncovalent inhibitor. Vecabrutinib treatment at a dose of 1 mM decreased Bcl-2 levels in cells harboring BTKC481S and Mcl-1 in cells overexpressing BTKC481R (Online Supplementary Figure S2E-G). To evaluate protein changes more extensively as well as to compare the effects of vecabrutinib with those of ibrutinib, we compared protein profiles using the reversephase protein array (RPPA). Since we observed the largest effect of vecabrutinib after treatment at 1 mM, in all cell lines we compared this concentration with samples treated with dimethylsulfoxide (DMSO) vehicle. The top ten canonical pathways were identified by Ingenuity Pathway Analysis (Online Supplementary Figure S3A-C). Several pathways were commonly affected in all three cell types. The extent of change and significance were different (Figure 2A). The top three canonical pathways with maximal change after vecabrutinib in cells with BTKWT overexpression were FLT3 signaling in hematopoietic progenitor cells, EGF and HGF signaling (Online Supplementary Figure S3A), whereas in cells with BTKC481S overexpression they were HGF signaling, regulation of epithelial-mesenchymal transition by growth factors pathway and B-cell receptor signaling (Online Supplementary Figure S3B) and in cells with BTKC481R overexpression they were epithelial-mesenchymal transition, ERK/MAPK signaling, and a senescence pathway (Online Supplementary Figure S3C). We also classified the types of target proteins using pie charts (Online Supplementary Figure S3D-F). Kinase and transcription regulator protein groups constituted >60% of proteins affected by vecabrutinib in all BTK subtypes. BCR pathway inhibition generally affects signal transduction, measured as phospho-proteins, proteins involved in transcription factors, cell proliferation, B-cell proteins, and apoptosis. Among the 258 proteins evaluated by RPPA, eight phospho-proteins and five other proteins were affected by vecabrutinib and ibrutinib (Figure 2B). SHP-2 (PTPN11), a phosphatase that plays a critical role at several junctures in the BCR pathway, has been shown to interact with many proteins.11 The tyrosyl phosphorylation of this protein has been shown to be haematologica | 2022; 107(1)

Letters to the Editor

Figure 1. Vecabrutinib kinome profile and effect on B-cell receptor pathway protein phosphorylation. (A) Vecabrutinib and ibrutinib kinase inhibition profiles. Vecabrutinib kinase selectivity and kinase inhibitory activities were evaluated in the KinaseProfiler™ panel of 234 individual kinases (Eurofins Pharma Discovery Services, Dundee, UK). For TEC and LCK, vecabrutinib activity was assessed against the activated kinase. (B) Phospho-BTK inhibition in human whole blood ex vivo. Peripheral blood mononuclear cells were isolated from peripheral blood of healthy donors and were incubated with 0.3 to 10,000 nM vecabrutinib for 0.5 h. Experiments were conducted on samples from 145 individuals to calculate IC50 values. Mean and standard deviation (SD) values are shown with horizontal lines and error bars. (C) Phospho-PLCg2 inhibition in the Ramos cell line. Cells were incubated with 0.3 to 10,000 nM vecabrutinib. Experiments were conducted on four biological replicates in the Ramos cell line to calculate IC50 values. Mean and SD values are shown with horizontal lines and error bars. (D-F). Effects of vecabrutinib on BCR pathway proteins in MEC-1 cells that overexpress wild-type or mutant BTK. Green fluorescence protein (GFP)-labeled MEC-1 cell lines that stably overexpress wild-type BTK (BTKWT) and mutant BTK (BTKC481S or BTKC481R) were generated by using standard lentiviral transfection methods. Cells were sorted by a BD FACSAria (BD Biosciences) for the enrichment of transduced GFP+ cell populations in each cell line. For all experiments, >75% GFPpositive cell populations were used. Cells were treated with indicated concentrations of vecabrutinib or ibrutinib and incubated for 24 h. Protein extracts were subjected to immunoblot assays to determine levels of phospho-BTK (Y223), BTK, phospho-ERK (T202/Y204), ERK in MEC-1 cells overexpressing (D) BTKWT, (E) BTKC481S, and (F) BTKC481R cells. Vinculin was used as the loading control.

haematologica | 2022; 107(1)

293

Letters to the Editor

Figure 2. Effect of vecabrutinib treatment on functional protein profiles of wild-type and mutant BTK-overexpressing MEC-1 cells. (A) Effect of vecabrutinib on common pathways in all three cell lines. Exponentially growing MEC-1 cells with either wild-type or mutant BTK were treated for 24 h with vecabrutinib at 1 mM. Experiments were performed in biological triplicates (n=3/cell line). At the end point, cells were collected, and protein was extracted and was subjected to the reverse-phase protein array (RPPA) that included 258 antibodies. Changes in expression (RPPAtreatment[1 mM]-RPPADMSO) values were used to determine the top canonical pathways identified with Ingenuity Pathway Analysis and associated with BTKWT (red bars), BTKC481S (light blue bars), and BTKC481R (navy blue bars) cells. (B) Effect of vecabrutinib treatment on proteins in wild-type and mutant BTK-overexpressing MEC-1 cells. Exponentially growing MEC-1 cells with either wild-type or mutant BTK were treated with three concentrations of vecabrutinib or ibrutinib and then their proteins were extracted and subjected to RPPA assays as described in the Methods section. Graphs were generated using the log2 fold change in expression (RPPAtreatment-RPPADMSO) values obtained by analysis of the RPPA data. Legends used in all graphs in Figure 2B are included in the upper left corner. Gene names and phosphorylation sites are indicated on top of each figure. Statistical comparisons were made between dimethylsulfoxide (DMSO)- versus ibrutinib- or vecabrutinib-treated cells, and asterisks depict P values <0.05.

294

haematologica | 2022; 107(1)

Letters to the Editor

Figure 3. Inhibition of the B-cell receptor pathway in chronic lymphocytic leukemia cells from patients with wild-type or mutant BTK. For blood sample collections, patients (n=5) provided written informed consent for the protocol, which was approved by the Institutional Review Board of The University of Texas MD Anderson Cancer Center, in accordance with the Declaration of Helsinki. Blood samples were collected into Vacutainer glass green-top blood collection tubes and cells were isolated by Ficoll-Hypaque density centrifugation and incubated with vecabrutinib at two or three concentrations (0.01, 0.1, and 1 mM) for 24 h. (A) Table presenting the patients’ numbers and BTK mutation status for five patients with chronic lymphocytic leukemia (CLL). BTK mutations were identified in patients’ samples using End CLL or End Lymphoma panels that included C481F, C481R, C481S, L528W, T474I, and T474F mutations. The same patients’ numbers are used in Figure 3B-E. (B) Apoptotic cell death in primary CLL lymphocytes of five patients. Freshly isolated cells (patients 1-3) or cryopreserved cells (patients 4 and 5) were incubated for 24 h with indicated concentrations of vecabrutinib. Cells were stained with annexin V-FITC and propidium iodide (PI) and apoptotic cells were determined with flow cytometry. Cell death in dimethylsulfoxide (DMSO)-treated samples was subtracted from that of treated samples. Percent apoptosis in DMSO was 14.6, 0.3, 3.3, 73, and 41.9 in patients 1 to 5, respectively. (C-E) Effect of vecabrutinib on B-cell receptor pathway proteins. Protein extracts were subjected to immunoblot assays to determine levels of phospho-BTK (Y223), BTK, phospho-ERK (T202/Y204), ERK, phospho-S6 (Ser235/236), and S6. (C, D) Patients with BTKWT or BTKT474F mutant CLL cells, (E) Patients with BTKC481S and BTKC481R variants.

haematologica | 2022; 107(1)

295

Letters to the Editor

essential for the activation of ERK, a downstream molecule in BCR signaling.12 Compared to DMSO-treated cells, BTK inhibition substantially decreased phosphoSHP2(Y542) in WT cells, whereas only vecabrutinib specifically inhibited SHP-2 phosphorylation in both mutant cell lines (Figure 2B). Other proteins that showed a change after drug treatment were ribosomal protein pS6Ka1(T573); FAK protein pPTK2(Y397); pHSB1(S82); transcription factor pERK1/2(T202/Y204); transcription factors pSTAT3(S727); pSTAT3(Y705); transcription factors ETS1 and GATA1; cell cycle proteins PLK1, and phosphop27(T198); B-cell protein PD1; and cleaved caspase 7 (Figure 2B). In general, among WT and mutant variants, changes in these proteins were more pronounced in WT cells than in mutants. Of the two drugs, vecabrutinib produced deeper or similar responses compared to those to ibrutinib in WT cells for all proteins (Figure 2B). Consistent with a prior report,12 treatment with ibrutinib and vecabrutinib caused a 2-log decrease in phospho-ERK in BTKWT cells after treatment. Furthermore, vecabrutinib treatment resulted in a 1-log decline in cells harboring mutant BTK, whereas ibrutinib had no effect (Figure 2B). A hallmark protein that is affected by ibrutinib is p70S6K or S6 kinase. The target protein of this kinase is S6 ribosomal protein, which initiates protein synthesis upon phosphorylation. Impressively, compared to ibrutinib, vecabrutinib profoundly decreased phospho-S6K in all three cell types. Among all proteins, phosphorylated ERK and S6K were consistently and substantially decreased in all three cell types. These two proteins may serve as biomarkers for the effect of vecabrutinib (Figure 2B). Ibrutinib produced a larger decrease in PD1 levels than vecabrutinib did. Parallel to PD1 protein levels, serine(727) and tyrosine(705) phosphorylation of STAT3 has been shown to be reduced in CLL cells after ibrutinib treatment.13 STAT3 and STAT5 activation through phosphorylation has been associated with inflammation and carcinogenesis.14 STAT3 is constitutively active in CLL cells, but this activation is mitigated by ibrutinib treatment.13,15 Vecabrutinib treatment decreased Y705 and S727 phosphorylation on STAT3 in WT and mutant cell lines. Interestingly, in mutant cells, a decline in S727 phosphorylation occurred only with vecabrutinib (Figure 2B). Although extensive cell death was not seen with either of these drugs, RPPA data revealed that vecabrutinib treatment increased cleaved caspase 7 in all transduced cell lines. Finally, we tested vecabrutinib in primary CLL cells from five patients with either WT or mutant BTK (Figure 3A). Cell death ranged from 0 up to 21% after 24 hours of vecabrutinib, being higher in BTKWT than in BTKmutant samples (Figure 3B). BTK with C481S and C481R alterations had the lowest apoptosis. In concert, immunoblot results showed that vecabrutinib inhibited BCR pathway (phosphorylation of BTK, ERK, and S6) signaling in BTKWT and in BTKT474F (gate-keeper mutation) (Figure 3C, D). Consistent with the RPPA data, cells expressing the BTKC481S and BTKC481R variants had minor changes (patient 3). In patients 4 and 5, phospho-ERK either remained the same or increased; these two patients had C481S (catalytic domain) and T474I (gate-keeper) double mutations (Figure 3E). Super-resistance to irreversible BTK inhibitors or variable sensitivity to reversible noncovalent BTK inhibitors has been reported for cells harboring T474 variants along with the cysteine 481 substitution.16 In summary, our study provides a nonclinical character296

ization of vecabrutinib. This reversible BTK inhibitor is selective in kinome profiling and inhibits phosphorylation of BTK and PLCg2 in preclinical whole-cell assays. Ibrutinib-sensitive and -resistant model systems demonstrated inhibition of the BCR signal transduction cascade in both BTKWT and BTKmutant B-cell lines.6 Data were consistent in CLL cells, but the impact on BTKC481S and BTKC481R was modest in primary cells. The results of a clinical trial of vecabrutinib in B-cell malignancies will be reported soon.7 Burcu Aslan,1 Stefan Edward Hubner,2 Judith A. Fox,3 Pietro Taverna,3 William G. Wierda,2 Steven M. Kornblau2 and Varsha Gandhi1,2 1 Department of Experimental Therapeutics, The University of Texas MD Anderson Cancer Center, Houston, TX; 2Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX and 3Sunesis Pharmaceuticals, Inc., South San Francisco, CA, USA Correspondence: VARSHA GANDHI - vgandhi@mdanderson.org. doi:10.3324/haematol.2021.279158 Received: May 6, 2021. Accepted: September 1, 2021. Pre-published: September 9, 2021. Disclosures: related to this work, VG has sponsored research agreements with Sunesis Pharmaceuticals and Loxo Oncology. Outside of this work, VG has received research support from AbbVie, Acerta, AstraZeneca, ClearCreek Bio, Dava Oncology, Gilead, and Pharmacyclics. JAF and PT were employees of Sunesis Pharmaceuticals. The other authors have no conflicts of interest. Contributions: BA designed and performed the experiments, analyzed the results, and wrote the first draft of the manuscript; SEH performed the bioinformatics analyses of the RPPA data; JF and PT contributed other nonclinical and clinical investigations of vecabrutinib, provided data that are included in Figure 1, and reviewed the manuscript;WGW identified ibrutinib-resistant patients and provided samples; SMK supervised SEH and analyzed RPPA data; VG conceptualized and supervised the research, obtained funding, analyzed the data, and wrote and revised the manuscript. Acknowledgments: the authors thank Drs. Joseph R. Marszalek, Christopher P. Vellano, Michael Peoples, and Mikhila Mahendra for the creation of MEC-1 cell lines that overexpress WT and mutant BTK. The authors thank LaKesla Iles for technical assistance and Tamara Locke, Scientific Editor, Research Medical Library, for editing this article. Funding: this work was supported in part by The University of Texas MD Anderson Cancer Center Moon Shot Program, a Sponsored Research Agreement from Sunesis Pharmaceuticals, and by the NIH/NCI through award number P30CA016672.

References 1. Burger JA. Treatment of chronic lymphocytic leukemia. N Engl J Med. 2020;383(5):460-473. 2. Timofeeva N, Gandhi V. Ibrutinib combinations in CLL therapy: scientific rationale and clinical results. Blood Cancer J. 2021;11(4):79. 3. Honigberg LA, Smith AM, Sirisawad M, et al. The Bruton tyrosine kinase inhibitor PCI-32765 blocks B-cell activation and is efficacious in models of autoimmune disease and B-cell malignancy. Proc Natl Acad Sci U S A. 2010;107(29):13075-13080. 4. Woyach JA, Ruppert AS, Guinn D, et al. BTKC481S-mediated resistance to ibrutinib in chronic lymphocytic leukemia. J Clin Oncol. 2017; 35(13):1437. 5. Burger JA, Landau DA, Taylor-Weiner A, et al. Clonal evolution in patients with chronic lymphocytic leukaemia developing resistance to BTK inhibition. Nat Commun. 2016;7(1):1-13. 6. Aslan B, Kismali G, Chen LS, et al. Development and characterization of prototypes for in vitro and in vivo mouse models of ibrutinib-resistant

haematologica | 2022; 107(1)

Letters to the Editor

CLL. Blood Adv. 2021;5(16):3134-3146. 7. Allan JN, Patel K, Mato AR, et al. Ongoing results of a phase 1B/2 doseescalation and cohort-expansion study of the selective, noncovalent, reversible bruton's tyrosine kinase inhibitor, vecabrutinib, in B-cell malignancies. Blood. 2019;134(Supplement_1):3041. 8. Herman SE, Mustafa RZ, Gyamfi JA, et al. Ibrutinib inhibits BCR and NF-kB signaling and reduces tumor proliferation in tissue-resident cells of patients with CLL. Blood. 2014;123(21):3286-3295. 9. Patel V, Balakrishnan K, Bibikova E, et al. Comparison of acalabrutinib, a selective Bruton tyrosine kinase inhibitor, with ibrutinib in chronic lymphocytic leukemia cells. Clinl Cancer Res. 2017;23(14):3734-3743. 10. Ponader S, Burger JA. Bruton's tyrosine kinase: from X-linked agammaglobulinemia toward targeted therapy for B-cell malignancies. J Clin Oncol. 2014;32(17):1830. 11. Neel BG, Gu H, Pao L. The ‘Shp'ing news: SH2 domain-containing tyrosine phosphatases in cell signaling. Trends Biochem Sci. 2003;28(6):284293.

haematologica | 2022; 107(1)

12. Araki T, Nawa H, Neel BG. Tyrosyl phosphorylation of Shp2 is required for normal ERK activation in response to some, but not all, growth factors. J Biol Chem. 2003;278(43):41677-41684. 13. Kondo K, Shaim H, Thompson PA, et al. Ibrutinib modulates the immunosuppressive CLL microenvironment through STAT3-mediated suppression of regulatory B-cell function and inhibition of the PD-1/PDL1 pathway. Leukemia. 2018;32(4):960-970. 14. Loh C-Y, Arya A, Naema AF, et al. Signal transducer and activator of transcription (STATs) proteins in cancer and inflammation: functions and therapeutic implication. Front Oncol. 2019;9:48. 15. Rui L, Drennan AC, Ceribelli M, et al. Epigenetic gene regulation by Janus kinase 1 in diffuse large B-cell lymphoma. Proc Natl Acad Sci U S A. 2016;113(46):E7260-E7267. 16. Estupiñán HY, Wang Q, Berglöf A, et al. BTK gatekeeper residue variation combined with cysteine 481 substitution causes super-resistance to irreversible inhibitors acalabrutinib, ibrutinib and zanubrutinib. Leukemia. 2021;35(5):1317-1329.

297

Letters to the Editor

Stored blood has compromised oxygen unloading kinetics that can be normalized with rejuvenation and predicted from corpuscular side-scatter During storage, red blood cells (RBC) undergo major biochemical and morphological changes.1,2 Progression of this so-called storage lesion varies between donor units,3,4 but key manifestations are a decrease in [2,3-diphosphoglycerate] ([2,3-DPG]), which becomes depleted by 3 weeks,5 and spherical remodeling.6 These changes are predicted to affect gas handling by reducing the amount of O2 off-loaded from RBC and expanding the diffusion path-length across the cytoplasm,7 respectively. Since RBC have only a few seconds to exchange gases in capillaries, impaired O2 handling could compromise the efficacy of transfusions, particularly when an immediate improvement to tissue oxygenation is required, such as with major hemorrhage. Although multiple randomized controlled trials have failed to identify clinical consequences related to the storage lesion,8 several limitations of these trials have been pointed out,9 including the concern that the intervention being tested is storage duration, rather than a functional appraisal of O2 handling by RBC. Consequently, studies that randomize according to storage duration may lack the power to resolve the real effects of storage lesion. It is therefore important to determine how storage affects the gas-handling properties of RBC. The effect of storage on the kinetics of O2 (un)loading is poorly understood because appropriate measurements are not routinely performed. For example, estimates of O2-binding affinity cannot predict ‘on’ and ‘off’ rates because they relate to the steady-state. A recent study10 found that storage paradoxically hastens gas exchange, despite also increasing O2-affinity, and that normalizing [2,3-DPG] has no effect on O2 release rate. However, these inferences were based on measuring the effect of RBC on extracellular O2 dynamics and are, therefore, indirect. To assess gas exchange more directly, we used single-cell HbO2 saturation imaging,7 which measures the time-constant (t) of O2-unloading and the O2-carrying capacity (k) evoked by anoxia. This technique combines ratiometric fluorescence imaging of HbO2 saturation using CellTracker Green and Deep-Red dyes, with rapid switching between oxygenated and anoxic microstreams delivered to RBC. Whole-blood units from consenting donors were collected in citrate phosphate dextrose, depleted of leukocytes, and manufactured as red cell concentrates (RCC) in saline adenine glucose-mannitol (SAG-M) within 27 h of venipuncture according to National Health Service Blood and Transplant procedures. Reference measurements were made on freshly collected venous blood from six volunteers (CUREC R46540/RE001). Data were fully anonymized. The first study investigated storage-related changes in RBC O2 handling, and how these vary between different units. RCC from six donors (A-F) were sampled at various points during storage (Figure 1A). Dye-loaded RBC were plated in a superfusion chamber mounted on a Leica LCS confocal system (excitation at 488/633 nm, emission at 500-550/650-700 nm) and O2 unloading was triggered by rapidly changing the microstream bathing cells from an oxygenated to an anoxic solution (N2-bubbled, containing 1 mM of O2 scavenger dithionite). Example time courses of the HbO2 response are shown in Figure 1B. With longer storage, cells from the RCC of donor C released O2 more slowly and in smaller quantities. This trend continued beyond week 3, at which point 298

2,3-DPG is depleted. In cells from the RCC of donor F, O2-handling reached its nadir already at day 2. Frequency distributions for the six RCC are presented in Figure 1C, ranked by progression of the kinetic decline. After 1 week of storage, some RCC (C, D) maintained fast O2-unloading kinetics, whereas others reached their nadir (F). With storage, the width of t distributions expanded, indicating increasing heterogeneity, but there was no evidence of bimodality. These distributions were transformed mathematically into cumulative frequency diagrams that inform about the fraction of RBC capable of unloading 95% of O2 in a given time-frame, T95 (Online Supplementary Figure S1A). The majority of freshly-collected RBC released their O2 cargo within 1 s, but with longer storage, RBC needed more time to complete gas exchange, reaching levels (T95 >2 s) that can result in diffusion-limited gas exchange at tissues. O2-carrying capacity (k) was also compromised by storage (Figure 1D). The second study tested whether defective O2 handling is reversible with biochemical rejuvenation, a treatment previously shown to normalize [2,3-DPG].11 To determine the effect of such treatment, ABO-matched RCC units were pooled in order to reduce donor-dependent variation in measured responses. Two RCC pools were rejuvenated at specified time-points under storage, and time-matched controls received no treatment (Figure 1E). Single-cell HbO2 saturation imaging showed that biochemical rejuvenation hastened O2-handling kinetics irrespectively of storage duration (Figure 1F, G). The normalizing effects of rejuvenation on t, k and T95 are summarized in Figure 1G and H and Online Supplementary Figure S1B. Proxies for O2-handling kinetics were sought by comparing the variables t and k against standard biochemical ([2,3-DPG], [ATP]) and flow-cytometric parameters (Sysmex XN1000 hematology analyzer), including RETRBC-Y and RET-RBC-Z channels that correspond to corpuscular forward-scatter and side-scatter (SSC) measured after treating RBC with Cellpack-DFL. [2,3-DPG] became depleted by day 21, ATP, mean corpuscular hemoglobin concentration and SSC decayed more slowly, while mean corpuscular volume increased (Figure 2A, Online Supplementary Figure S1C). The efficacy of rejuvenation was confirmed by the increases in [2,3-DPG], [ATP] and SSC (Figure 2A, Online Supplementary Figure S1D). The most significant correlates to t and k (Spearman test) were storage duration, [2,3DPG], [ATP] and SSC (Figure 2B). The inverse relationship between t and k (Online Supplementary Figure S2A) suggests a common underlying cause of remodeling. A process anticipated to reduce k and raise t is [2,3-DPG] degradation; indeed, an inverse correlation was apparent between [2,3-DPG] and t (Online Supplementary Figure S2B). However, the power of [2,3-DPG] to predict t became compromised beyond day 21, after which t continued to increase, despite [2,3-DPG] depletion. An additional factor, such as the expanding path-length in spherically-remodeled RBC, must be contributing to dysfunctional O2 handling. Intriguingly, SSC was found to have good predictive power for t (Online Supplementary Figure S2C), and the power improved further when sample-level variation was factored into the regression (Figure 2C, Online Supplementary Figure S2D). Crucially, SSC was a considerably better predictor of changes in O2 handling than was storage duration. In contrast to [2,3DPG], SSC continued to fall beyond day-21, is more easily measured, and provides data at single-cell resolution. SSC measured conventionally is influenced by RBC shape and composition, but upon haematologica | 2022; 107(1)

Letters to the Editor

Cellpack-DFL treatment, it also becomes sensitive to membrane permeability, and hence metabolic status. The association between SSC and 2,3-DPG (metabolism) and forward-scatter (volume) was confirmed by a metaanalysis of multiple RCC11 (Online Supplementary Figure S2E, F). Metabolomic analyses were performed to relate specif-

ic metabolites to O2-handling dysfunction4,12 (Figure 3A, Online Supplementary Figure S3A). Established markers of the storage lesion13 were validated, including lactate, hypoxanthine, glucose and reduced glutathione. Donor unit-dependent variation was manifested in the responses of sphingosine 1-phosphate, methionine, S-adenosylmethionine, spermidine, and 20:3 and 20:5 fatty acids

Figure 1. The kinetics of O2 handling in red cell concentrates are dependent on storage duration and donor, and can be normalized with rejuvenation treatment. (A) The protocol of study #1 showing color-coding for donors and indicating the days at which red cell concentrates (RCC) were sampled for analyses. (B) Example time course of HbO2 saturation in response to a rapid solution maneuver that reduces extracellular O2 tension to anoxic levels. Measurements on RCC from donor C (left) or F (right) at days 2, 7, 35, 42 and 49 of storage. Each line represents the mean of ten red blood cells (RBC); error bars are the standard deviation (SD). (C) Histograms of O2-unloading time constant t grouped by donor, and plotted for storage days 2, 7, 35, 42 and 49 (red transitioning to blue as storage duration increases). Each histogram was constructed from at least 300 RBC. Reference data from freshly collected venous blood are shown in yellow. (D) Frequency histograms of O2-carrying capacity k. (E) The protocol of study #2 showing color-coding for RCC pools #1 and #2 and days at which biochemical rejuvenation and analyses were performed. (F) Example time course of HbO2 saturation response to anoxia (mean±SD of ten cells from RCC pool #2). (G) Histograms of t, color-coded by storage duration (red to blue). Dashed lines denote measurements from rejuvenated units. Each histogram was constructed from a total of at least 300 RBC. Reference data from freshly collected venous blood are shown in yellow. (H) Frequency histograms of k.

haematologica | 2022; 107(1)

299

Letters to the Editor

(Online Supplementary Figure S3). The metabolic consequences of rejuvenation (Figure 3B, Online Supplementary Figure S3B) provided additional data-points for establishing correlations against t and k (Figure 3C, D). The highest-ranking metabolic correlates to t, equal to or better than [2,3-DPG], are listed in Figure 3E.

Our findings add O2-unloading kinetics to the otherwise static definition of storage lesion. We provide metabolite correlates to O2 handling beyond 2,3-DPG and ATP for future investigations, and show that RBC SSC, a variable influenced by metabolism and corpuscular morphology, is a useful proxy for O2-exchange kinet-

Figure 2. Storage-related dysfunction of O2 handling relates to the level of 2,3- diphosphoglycerate but is best predicted by side-scatter measured on the RETRBC-Z channel of hematology analyzers. (A) Biochemical, blood count, and flow-cytometric parameters measured in red cell concentrates (RCC). Color-coding refers to the donor (study #1) and RCC pool (study #2). Parameters include concentration of diphosphoglycerate ([2,3-DPG]), [ATP], blood [Hb], mean corpuscular hemoglobin concentration (MCHC), mean corpuscular hemoglobin (MCH), mean corpuscular volume (MCV), corpuscular forward- and side-scatter (FSC, SSC) after treatment with Cellpack DFL. (B) Summary of parameters (scaled to a range 0-1) determined by single-cell HbO2 saturation imaging (t, k), biochemistry (2,3-DPG, ATP), RET-channel flow cytometry (FSC, SSC), and standard hematologic indices ([Hb], MCHC, MCH, and MCV). Correlations with t or k determined by Spearman test for datasets in study #1 alone, or the combined datasets from both studies. Filled bars indicate significant correlation (P<0.05). (C) Regression analyses determining the ability of storage duration, [2,3-DPG] normalized to [Hb], [ATP] normalized to [Hb], or SSC to predict actual t. Sample-level effects were modeled as a random intercept. Circles and stars indicate data from study #1 and #2, respectively. Goodness-of-fit was quantified by the χ2 test.

300

haematologica | 2022; 107(1)

Letters to the Editor

Figure 3. Metabolomic analyses of red cell concentrates under storage and after rejuvenation identify correlates to O2-handling kinetics. (A) Mass spectrometry analyses (MetaboAnalyst) of red cell concentrates (RCC) from study #1. Partial least square-discriminant analysis (PLS-DA) indicates a significant impact of storage duration and inter-donor variability on metabolic phenotypes. Metabolites with the highest loading weights in principal components (PC) 1 and 2 were affected by storage duration across all samples (~35% of the total variance) and biological variability between donors (~11%). (B) Metabolomic analyses of RCC before and after rejuvenation. Rejuvenation had a significant impact on metabolites with the highest loading weights in PC1 (43%). (C,D) Metabolic correlates (Spearman test) to t or k for units under study #1 (C) and study #2 (D). Metabolites-of-interest are highlighted. These include 5-oxoproline and lactate as correlating positively with t; C18/20 free fatty acids (e.g. linoleic) as correlating positively with t and negatively with k; fructose 1,6-bisphosphatase (FBP) and inosine triphosphate (ITP; in rejuvenated pools, possibly a result of the inosine boost), the pentose phosphate metabolite erythrose 4-phosphate (EryP) and sphingosine 1-phosphate (S1P), dicarboxylates (2-oxoglutarate, fumarate, succinate, malate), ascorbate/dehydroascorbate, and the amino acids taurine and phenylalanine (Phe) as correlating negatively with t. DPG: 2,3-diphosphoglycerate; FA: fatty acid; Gly: glycine; Lac: lactate; SAM, S-adenosyl methionine. (E) Top five metabolites correlating positively (top, red) and negatively (bottom, blue) with t, using data from study #1 (circles) and study #2 (stars). These correlations were equal to or better than for [2,3-DPG].

haematologica | 2022; 107(1)

301

Letters to the Editor

ics. Our assay identified major differences between donor units in the progression of O2-handling dysfunction under storage. The extent to which a gene-function correlation underpins this variation is testable with our method. The normalization of biochemical and flowcytometric parameters with rejuvenation explains how O2 handling is also restored to physiological levels. These observations are significant because of renewed interest in optimizing blood storage regimes for patients receiving large transfusions (in whom cumulative effects of storage lesion can be profound), or patients undergoing procedures in which sub-optimal tissue oxygenation may result in organ injury. Several high-quality randomized controlled trials14 have shown that transfusion of blood after prolonged storage is not associated with higher rates of adverse clinical outcomes than those after transfusion of blood stored for shorter periods. However, these trials mainly recruited participants with stable anemia and without an imminent threat to organ perfusion, and did not include patients requiring massive transfusion or those with major trauma,14 i.e., cohorts for whom the consequences of impaired gas exchange kinetics are most relevant. In the case of transfusion for major hemorrhage, the clinical imperative is to restore tissue oxygenation immediately. To achieve this, donated RBC must be capable of exchanging gases efficiently at the point of transfusion. This imperative is less relevant in patients with stable anemia because there is evidence that [2,3-DPG] recovers upon re-exposure to a physiological milieu, albeit slowly over up to 3 days.15 A functional assessment of the recovery of O2-handling kinetics in vivo is warranted to determine how the efficacy of transfusion depends on the clinical context of the recipient. In the aforementioned trials, randomizations were based on storage duration rather than high-quality measurements of RBC physiology. We find that time in storage does not adequately describe the deterioration in O2-handling kinetics, and therefore the use of storage duration for treatment allocation may have compromised the statistical power to detect adverse clinical outcomes related to storage lesion. These considerations may explain the lack of a statistically significant effect of different storage regimes in participants with stable anemia. Given that O2-handling kinetics are mechanistically related to the physiological quality of transfused units, we propose that future investigations should consider RBC function, as determined from O2-unloading kinetics or suitable proxies. Killian Donovan,1 Athinoula Meli,2 Francesca Cendali,3 Kyung Chan Park1 Rebecca Cardigan,2,4 Simon Stanworth,5,6,7 Stuart McKechnie,8 Angelo D’Alessandro,3 Peter A. Smethurst2 and Pawel Swietach1 1 Department of Physiology, Anatomy & Genetics, University of Oxford, Oxford, UK; 2Component Development Laboratory, NHS Blood and Transplant, Cambridge, UK; 3Department of Biochemistry and Molecular Genetics, University of Colorado Anschutz Medical Campus, Aurora, CO, USA; 4Department of Haematology, University of Cambridge, Cambridge, UK; 5Transfusion Medicine, NHS Blood and Transplant, Oxford, UK; 6Department of Haematology, Oxford University Hospitals NHS Foundation Trust,

302

Oxford, UK; 7Radcliffe Department of Medicine, University of Oxford, and Oxford BRC Haematology Theme, Oxford, UK and 8 Adult Intensive Care Unit, John Radcliffe Hospital, Oxford University Hospitals NHS Foundation Trust, Oxford, UK Correspondence: PAWEL SWIETACH - pawel.swietach@dpag.ox.ac.uk doi:10.3324/haematol.2021.279296 Received: May 26, 2021. Accepted: September 1, 2021. Pre-published: September 9, 2021. Disclosures: no conflicts of interest to disclose. Contributions: KD, AM, FC, KCP, AD'A, PAS and PS performed research; PS, PAS, AD'A, AM, KD and RC designed and supervised the study; KD, PS, FC and AD'A analyzed the data; PS, KD, AM, SS, SM, AD'A and PAS wrote the manuscript. Acknowledgments: we thank Drs Jarob Saker and Jean-Pierre Perol (Sysmex, Europe) for discussions on interpreting SSC measurements.

References 1. Bennett-Guerrero E, Veldman TH, Doctor A, et al. Evolution of

adverse changes in stored RBCs. Proc Natl Acad Sci U S A. 2007; 104(43):17063-17068. 2. Yoshida T, Prudent M, D'Alessandro A. Red blood cell storage lesion: causes and potential clinical consequences. Blood Transfus. 2019; 17(1):27-52. 3. Tzounakas VL, Georgatzakou HT, Kriebardis AG, et al. Donor variation effect on red blood cell storage lesion: a multivariable, yet consistent, story. Transfusion. 2016;56(6):1274-1286. 4. D'Alessandro A, Fu X, Kanias T, et al. Donor sex, age and ethnicity impact stored red blood cell antioxidant metabolism through mechanisms in part explained by glucose 6-phosphate dehydrogenase levels and activity. Haematologica. 2021;106(5):1290-1302. 5. Hess JR. Measures of stored red blood cell quality. Vox Sang. 2014; 107(1):1-9. 6. Roussel C, Dussiot M, Marin M, et al. Spherocytic shift of red blood cells during storage provides a quantitative whole cell-based marker of the storage lesion. Transfusion. 2017;57(4):1007-1018. 7. Richardson SL, Hulikova A, Proven M, et al. Single-cell O2 exchange imaging shows that cytoplasmic diffusion is a dominant barrier to efficient gas transport in red blood cells. Proc Natl Acad Sci U S A. 2020;117(18):10067-10078. 8. Carson JL, Stanworth SJ, Alexander JH, et al. Clinical trials evaluating red blood cell transfusion thresholds: an updated systematic review and with additional focus on patients with cardiovascular disease. Am Heart J. 2018;200:96-101. 9. Trivella M, Stanworth SJ, Brunskill S, Dutton P, Altman DG. Can we be certain that storage duration of transfused red blood cells does not affect patient outcomes? BMJ. 2019;365:l2320. 10. Chng KZ, Ng YC, Namgung B, et al. Assessment of transient changes in oxygen diffusion of single red blood cells using a microfluidic analytical platform. Commun Biol. 2021;4(1):271. 11. Smethurst PA, Jolley J, Braund R, et al. Rejuvenation of RBCs: validation of a manufacturing method suitable for clinical use. Transfusion. 2019;59(9):2952-2963. 12. Stefanoni D, Shin HKH, Baek JH, et al. Red blood cell metabolism in Rhesus macaques and humans: comparative biology of blood storage. Haematologica. 2020;105(8):2174-2186. 13. Paglia G, D'Alessandro A, Rolfsson O, et al. Biomarkers defining the metabolic age of red blood cells during cold storage. Blood. 2016; 128(13):e43-50. 14. McQuilten ZK, French CJ, Nichol A, Higgins A, Cooper DJ. Effect of age of red cells for transfusion on patient outcomes: a systematic review and meta-analysis. Transfus Med Rev. 2018;32(2):77-88. 15. Scott AV, Nagababu E, Johnson DJ, et al. 2,3-Diphosphoglycerate concentrations in autologous salvaged versus stored red blood cells and in srgical patients after transfusion. Anesth Analg. 2016;122(3): 616-623.

haematologica | 2022; 107(1)

Letters to the Editor

Elastic compression stockings for prevention of the post-thrombotic syndrome in patients with and without residual vein thrombosis and/or popliteal valve reflux The efficacy of elastic compression stockings (ECS) for the prevention of post-thrombotic syndrome (PTS) arising after a proximal deep-vein thrombosis (DVT) is controversial.1,2 Although most randomized studies showed a substantial reduction of PTS with the use of ECS,3-5 the recent, large SOX clinical trial that used sham stockings as a comparator failed to confirm these findings.6 Accordingly, most international guidelines no longer recommend routine use of ECS for the prevention of PTS.1,7 Nevertheless, ECS are still commonly prescribed in clinical practice.8 A recent systematic review showed that patients with DVT who at 6 weeks or later had either residual vein thrombosis (RVT) or popliteal valve reflux (PVR) had a higher risk of subsequent PTS than those without these findings.9 As these ultrasound features are associated with longstanding venous hypertension that could be modified by compression therapy,10 the early identification of RVT and PVR has the potential to identify a subgroup of patients who may still benefit from the use of ECS. In a prospective cohort study of 869 patients with a proximal DVT that was either unprovoked or associated with transient risk factors, we observed an increased risk of PTS in those with RVT.11 Briefly, all patients received initial treatment with unfractionated or low-molecularweight heparin followed by vitamin K antagonists according to international guidelines, with individual treatment duration (ranging between 3 and 24 months)

based on each patient’s preferences and risk profile. Patients with recent (<2 years) ipsilateral DVT and those requiring indefinite anticoagulation were not eligible.11 Patients were advised to wear ECS (30-40 mmHg at the ankle) for at least 2 years, and were followed-up for 3 years. They were instructed to report in a booklet how long they wore the stockings, the use of not permitted stockings, and the occurrence of any adverse effects impairing their use. At 3 months, an ultrasound assessment was done to document the presence of RVT (vein diameter under maximum compressibility >4 mm)11 and PVR (retrograde flow through the popliteal valve after manual compression of the mid-thigh >0.5 seconds, which persisted after repeating the maneuver with a tourniquet).12 The Villalta scale was used to assess the development of PTS every 6 months. A score of 5 to 14 in two, even non-consecutive, assessments indicated non-severe PTS, while a score ≥15 or the presence of a skin ulcer in a single assessment indicated severe PTS.1,11,13 Here we report the risk of PTS in relation to therapeutic adherence to ECS and the presence of RVT, PVR or both in the 861 patients who survived at least 6 months. Two trained physicians who were unaware of the patients’ other details or study outcomes assessed the adherence to ECS. Patients who used the ECS for ≥70% of daytime for the first year were considered adherent (‘stockings’ group). Patients who did not accept the advised ECS, discontinued ECS use during the first year, or used the ECS <70% of daytime were considered nonadherent (‘non-stockings’ group). The main demographic and clinical characteristics of the two groups were compared using standard methods. The hazard ratio (HR) for the effect of ECS on PTS development in the whole population, as well as in patients

Figure 1. Cumulative incidence of patients free of post-thrombotic syndrome in each of the four subgroups according to the presence of residual vein thrombosis and/or popliteal valve reflux and the use of elastic compression stockings. PTS: post-thrombotic syndrome; ECS: elastic compression stockings; RVT: residual vein thrombosis; PVR: popliteal valve reflux.

haematologica | 2022; 107(1)

303

Letters to the Editor

Table 1. Demographic and clinical characteristics of the study patients, separately in the ‘stockings’ and ‘non-stockings’ groups.

Characteristics Age, years, mean ± SD median (range) Males, n(%) Obesity, n(%) Unprovoked DVT, n(%) Previous VTE, n(%) Symptoms of PE, n(%) DVT location, n(%) common femoral vein ± popliteal vein popliteal vein only Vein abnormalities, n(%) RVT (alone or combined with PVR) PVR (alone or combined with RVT) Recurrent DVT and/or PE, n(%) overall ipsilateral DVT Deaths, n(%) Duration of treatment, months mean ± SD median (range) Duration of follow-up, months mean ± SD median (range)

Non-ECS group (N=350)

ECS group (N=511)

62.4 ± 16.5 62 (15-91) 163 (46.6) 48 (13.7) 180 (51.4) 52 (14.9) 59 (16.9)

58.1 ± 18.3 65.5 (18-89) 252 (49.3) 56 (11.0) 258 (50.5) 51 (10.0) 67 (13.1)

0.001 0.429 0.223 0.787 0.030 0.127

166 (47.4) 184 (52.6)

261 (51.1) 250 (48.9)

0.293 -

176 (50.3) 108 (30.9)

248 (48.5) 132 (25.8)

0.613 0.106

53 (15.1) 14 (4.0) 21 (6.0)

74 (14.5) 35 (6.8) 36 (7.0)

0.788 0.098 0.545

5.1 ± 4.0 3 (1-24)

5.4 ± 4.1 3 (3-24)

0.207 -

34.0 ± 6.3 36 (6-36)

34.2 ± 5.9 36 (6-36)

0.670 -

ECS: elastic compression stockings; SD: standard deviation; DVT: deep vein thrombosis; VTE: venous thromboembolism; PE: pulmonary embolism; RVT: residual vein thrombosis; PVR: popliteal valve reflux; Numbers in parentheses indicate percentages unless otherwise specified.

with and without RVT and/or PVR was estimated using the proportional hazard Cox regression model. Interaction terms were defined between RVT and ECS, and between PVR and ECS. A minimal significant model was achieved by a likelihood ratio-guided forward stepwise variable selection method. In each of the four subgroups of patients with or without RVT and/or PVR in the ‘stockings’ or ‘non stockings’ group, the cumulative incidence of PTS was estimated by the Kaplan-Meier method, tested by the log-rank test, and graphically represented by product-limit survival estimates by the final minimal significant model. Of the 861 patients, 511 (59.3%) belonged to the ‘stockings’ group, and the remaining 350 (40.7%) to the ‘non-stockings’ group. The two groups had substantially comparable demographic and clinical characteristics (Table 1). RVT and/or PVR was detected in 539 patients (62.6%). Of these, RVT alone was found in 299 (55.5%), PVR alone in 115 (21.3%) patients, and the combination of RVT with PVR in 125 (23.2%). PTS developed in 249 of the 539 patients (46.2%; severe in 35, 6.5%) with RVT and/or PVR, and in 90 of the 322 (28.0%; severe in 11, 3.4%) without RVT and/or PVR (HR=2.18; 95% confidence interval [95% CI], 1.732.74). PTS developed in 162 of the 511 patients (31.7%; severe in 19, 3.7%) in the ‘stockings’ group, and in 177 of the 350 (50.6%; severe in 25, 7.1%) in the ‘non-stockings’ group (HR=0.64; 95% CI: 0.51-0.79; P<0.001). In patients with RVT and/or PVR, PTS developed in 114 of the 328 (34.8%) in the ‘stockings’ group (severe in 304

14, 4.3%), and in 135 of the 211 (64.0%) in the ‘nonstockings’ group (severe in 19, 9.0%), for hazard ratios of all and severe PTS of 0.52 (95% CI: 0.41-0.66; P<0.001) and 0.41 (95% CI: 0.21-0.83; P=0.013), respectively. In patients without RVT and/or PVR, PTS developed in 48 of the 183 (26.2%) patients in the ‘stockings’ group (severe in 5, 2.7%), and 42 of the 139 (30.2%) patients in the ‘non-stockings’ group (severe in 6, 4.3%), for hazard ratios of all and severe PTS of 0.95 (95% CI: 0.62-1.44; P=0.80) and 0.59 (95% CI: 0.18-1.95; P=0.59), respectively (Table 2). In patients with RVT and/or PVR, the 36month PTS-free survival figures were 35.2% (95% CI: 28.7-41.7) in the ‘non-stockings’ group, and 64.0% (95% CI: 58.7-69.3) in the ‘stockings’ group (P<0.001). In patients without RVT and/or PVR, the respective figures were 69.3% (95% CI: 61.5-77.1) and 73.5% (95% CI: 67.0-78.0), respectively (P=0.43) (Figure 1). Of utmost importance, while the interaction term for the use of ECS and presence of RVT was highly significant (P=0.037), the term for the use of ECS and presence of PVR was not (P=0.46). Our study strongly suggests that in patients with proximal DVT, adequate use of ECS provides a clinically important reduction in any and severe PTS in patients with ultrasound evidence of RVT (with or without PVR) at 3 months, whereas in patients without RVT such an effect is absent. In a clinical context dominated by persistent uncertainty on the necessity for compression therapy to prevent PTS, our study has the potential to revive a stalled discussion. haematologica | 2022; 107(1)

Letters to the Editor

Table 2. Incidence of post-thrombotic syndrome in patients according to the use of compression elastic stockings and the presence of residual vein thrombosis and poplitieal valve reflux.

PTS

Mild and severe P Severe alone P

No RVT and/or PVR No ECS ECS (N=139) (N=183) 42 (30.2)

48 (26.2)

RVT and/or PVR No ECS ECS (N=211) (N=328) 135 (64.0)

0.430 6 (4.3)

114 (34.8)

PVR alone No ECS ECS (N=35) (N=80) 10 (28.6)

0.001 5 (2.7)

0.625

19 (9.0)

18 (22.5) 0.488

14 (4.3) 0.001

2 (5.7)

3 (3.8) 0.763

RVT alone No ECS ECS (N=103) (N=196) 70 (68.0)

68 (34.7) 0.001 9 9 (8.7) (4.6) 0.001

RVT and PVR No ECS ECS (N=73) (N=52) 55 (75.3)

28 (53.8) 0.001 8 2 (11.0) (3.8) 0.028

PTS: post-thrombotic syndrome; RVT: residual vein thrombosis; PVR: popliteal valve reflux; ECS: elastic compression stockings. Numbers in brackets are percentages.

Recently, a hypothesis-generating meta-analysis showed a more than two-fold higher incidence of PTS in patients with ultrasound evidence of RVT at least 6 weeks after the index DVT.9 In patients with PVR, the incidence of PTS was also increased but only by onethird. In hindsight, this small increase could be easily explained by the confounding effect of RVT, which can also occur in combination with PVR. Indeed, in our study 48% of patients with PVR also had RVT, but in the multivariate minimal-significant Cox model, there was no independent effect of PVR on the incidence of PTS, or on the relative efficacy of adequate use of ECS. Our observation is pathophysiologically plausible. Indeed, in the absence of longstanding vascular damage, venous hypertension and subsequently PTS are unlikely to occur.9,14 This is also consistent with the demonstration that PTS is unlikely to develop in individuals with a limited thrombotic burden and in those with isolated calf DVT.1 Our results are robust, as they are based on a prospective observation of patients with proximal DVT who were followed up for as long as 3 years.11 Furthermore, the assessment of the adequacy of use of ECS was done by physicians who were unaware of clinical outcomes or potential confounders. To minimize the effect of potential confounders, patients with recent ipsilateral DVT were excluded, as were patients with a short life expectancy and those requiring indefinite anticoagulation. The main limitation of our study was the lack of random allocation to ECS or no ECS. However, the two study groups were virtually comparable for demographic and clinical characteristics. The clinical implication of our observations in combination with results of contemporary studies is that in patients without RVT at 3 months, ECS can be safely withheld as long as leg complaints have disappeared, ideally after completing the first 6 months following the thrombotic episode.15 However, there is still uncertainty regarding patients with substantial damage to their venous system, as demonstrated by the presence of RVT at 3 months. Hence, we believe that trials should be initiated to assess the effect of ECS in patients with proximal DVT and RVT. As our patients were instructed to use ECS from the beginning, and the likelihood of RVT at 3 months was substantial (>50%), we think that both in clinical practice and in further confirmatory studies ECS should be given as soon as possible after the acute DVT to all patients while awaiting the ultrasound test, which has the potential to help decisions on the subsequent approach. In conclusion, our results show that the ultrasound assessment of RVT in patients with proximal DVT has haematologica | 2022; 107(1)

the potential to identify those who are likely to benefit from the long-term use of ECS. While awaiting confirmation from properly randomized clinical trials, they are, in our opinion, plausible enough to inform the long-term management of patients with proximal DVT. Paolo Prandoni,1 Anthonie W.A. Lensing,1 Martin H. Prins,2 Sabina Villalta,3 Raffaele Pesavento,4 Daniela Tormene,5 Franco Noventa6 and Gualtiero Palareti1 1 Arianna Foundation on Anticoagulation, Bologna, Italy; 2 Department of Clinical Epidemiology and Technology Assessment, University of Maastricht, Maastricht, the Netherlands; 3Division of Internal Medicine, Civic Hospital of Castelfranco Veneto, Italy; 4 Division of Internal Medicine, Civic Hospital of Montebelluna, Italy; 5 Department of Medicine, University Hospital of Padua, Italy and 6 Department of Molecular Medicine, University of Padua, Italy Correspondence: PAOLO PRANDONI - prandonip@gmail.com doi:10.3324/haematol.2021.279680 Received: July 21, 2021. Accepted: September 1, 2021. Pre-published: September 9, 2021. Disclosures: no conflicts of interest to disclose. Contributions: PP, AWAL and MHP designed the study; SV, RP and DT contributed patients, and analyzed and interpreted data; FN and MHP performed the statistical analysis; GP provided the administrative support; PP, AWAL and MHP drafted the manuscript; all authors critically wrote or revised the intellectual content of the manuscript, reviewed and/or commented on each draft, and approved the final version for submission.

References 1. Kahn SR, Comerota AJ, Cushman M, et al. The postthrombotic syn-

drome: evidence-based prevention, diagnosis, and treatment strategies: a scientific statement from the American Heart Association. Circulation. 2014;130(18):1636-1661. 2. Subbiah R, Aggarwal V, Zhao H, Kolluri R, Chatterjee S, Bashir R. Effect of compression stockings on post thrombotic syndrome in patients with deep vein thrombosis: a meta-analysis of randomised controlled trials. Lancet Haematol. 2016;3(6):e293-300. 3. Brandjes DP, Büller HR, Heijboer H, et al. Randomised trial of effect of compression stockings in patients with symptomatic proximalvein thrombosis. Lancet. 1997;349(9054):759-762. 4. Prandoni P, Lensing AW, Prins MH, et al. Below-knee elastic compression stockings to prevent the post-thrombotic syndrome: a randomized, controlled trial. Ann Intern Med. 2004;141(4):249-256. 5. Mol GC, van de Ree MA, Klok FA, et al. One versus two years of elastic compression stockings for prevention of post-thrombotic syndrome (OCTAVIA study): randomised controlled trial. BMJ. 2016;353:i2691. 6. Kahn SR, Shapiro S, Wells PS, et al. Compression stockings to prevent post-thrombotic syndrome: a randomised placebo-controlled trial. Lancet. 2014;383(9920):880-888.

305

Letters to the Editor

306

7. Stevens SM, Woller SC, Baumann Kreuziger L, et al. Antithrombotic

12. Vedovetto V, Dalla Valle F, Milan M, Pesavento R, Prandoni P.

therapy for VTE disease: second update of the CHEST guideline and expert panel report - executive summary. Chest. 2021;S00123692(21)01507-5. 8. Rabe D, Partsch H, Heidl G, et al. Compression treatment in acute symptomatic proximal deep venous thrombosis - results of a worldwide survey. Phlebology. 2021;36(7):526-534. 9. Dronkers CEA, Mol GC, Maraziti G, et al. Predicting post-thrombotic syndrome with ultrasonographic follow-up after deep vein thrombosis: a systematic review and meta-analysis. Thromb Haemost. 2018;118(8):1428-1438. 10. Ten Cate-Hoek AJ. Stockings to prevent post-thrombotic syndrome. BMJ. 2016;353:i2897. 11. Prandoni P, Lensing AW, Prins MH, et al. The impact of residual thrombosis on the long-term outcome of patients with deep venous thrombosis treated with conventional anticoagulation. Semin Thromb Hemost. 2015;41(2):133-140.

Residual vein thrombosis and trans-popliteal reflux in patients with and without the post-thrombotic syndrome. Thromb Haemost. 2013;110(4):854-855. 13. Kahn SR, Partsch H, Vedantham S, Prandoni P, Kearon C. Definition of post-thrombotic syndrome of the leg for use in clinical investigations: a recommendation for standardization. J Thromb Haemost. 2009;7(5):879-883. 14. Amin EE, Bistervels IM, Meijer K, et al. Reduced incidence of vein occlusion and postthrombotic syndrome after immediate compression for deep vein thrombosis. Blood. 2018;132(21):2298-2304. 15. Ten Cate-Hoek AJ, Amin EE, Bouman AC, et al. Individualised versus standard duration of elastic compression therapy for prevention of post-thrombotic syndrome (IDEAL DVT): a multicentre, randomised, single-blind, allocation-concealed, non-inferiority trial. Lancet Haematol. 2018;5(11):e25-33.

haematologica | 2022; 107(1)

Letters to the Editor

Targeting Wnt signaling in acute myeloid leukemia stem cells In this work we tested whether targeting Wingless and Int-1 (Wnt) signaling in acute myeloid leukemia (AML) leukemic stem cells (LSC) by using a porcupine inhibitor (WNT974) is a good strategy to eradicate LSC. Wnt/b-catenin is an evolutionarily conserved pathway that is involved in embryonic development and stem cells by regulating cell fate and differentiation decisions.1 In normal conditions, the Wnt/b-catenin pathway mediates normal hematopoietic stem cell self-renewal, proliferation and differentiation and is tightly controlled.2,3 While b-catenin is highly expressed in normal hematopoietic stem cells, it is downregulated during myeloid differentiation.2,4 Aberrant activation of this pathway gives rise to the accumulation of b-catenin in the nucleus, and promotes the transcription of many oncogenes such as c-Myc (MYC). b-catenin is highly expressed in AML patients and overexpression of b-catenin in normal CD34+ hematopoietic precursors in vitro leads to a myeloproliferative disorder of immature cells.5,6 Further work identified a crucial role of the Wnt/b-catenin pathway in the development of LSC,7 and different authors proposed targeting the Wnt pathway in AML blasts.8-11 Thus, we

hypothesize that targeting Wnt in LSC, using a novel Wnt inhibitor called WNT974, could be an efficient strategy to eradicate LSC, prevent relapse and improve overall outcomes in AML. WNT974 is an inhibitor of the enzyme porcupine, which is a membrane-bound O-acyltransferase located in the endoplasmic reticulum, and it is required for the palmitoylation of Wnt ligands. Inhibition of porcupine prevents the secretion and activity of Wnt ligands outside the cell, leading to a decrease in Wnt ligand cell surface receptor phosphorylation and a reduction in the expression of Wnt target genes.12,13 Several studies documented that WNT974 is able to target the Wnt pathway in solid tumors and the LSC in chronic myeloid leukemia.14 Based on this information, we hypothesized that Wnt inhibition by using WNT974 may block the aberrant Wnt activation in AML LSC and prevent leukemogenesis. To verify this hypothesis, we first evaluated whether WNT974 treatment can inhibit the Wnt signaling pathway using the K562 cell line, which is the cell line used initially to report the activity of this drug in myeloid leukemias. Treatment of K562 cells with dimethylsulfoxide (DMSO), as a control, or WNT974 at different concentrations (0.5, 2.5 and 10 mM) for 24 and 48 h showed a reduction of Wnt pathway targets such as ROR2, LRP6 and GSK3B protein (Figure 1A), and AXIN2 mRNA

Figure 1. Validation of WNT974 activity in cell lines. (A) Western blot assay of ROR2, LRP6 and GSK3B after WNT974 treatment or control (dimethylsulfoxide) for 24 and 48 h. Densitometry analysis was conducted using ImageJ software. Band intensity is reported relative to b-actin (ACTB). (B) AXIN2 mRNA fold change expression of K562 cells treated with WNT974 at different concentrations, or control (DMSO), for 24 h. *P<0.05, **P<0.01 (t-test). (C) Luciferase Renilla assay of HCT116 cells transfected with a wild-type or mutant b-catenin activity reporter vector. Thirty hours after transfection, the cells were treated with Wnt3a ligand (200 ng/mL), WNT974 5 µM, or a combination for 24 h. Data are normalized to the control wild-type. *P<0.05, **P<0.01, ***P<0.001 (t-test). (D) MYC mRNA relative expression of bone marrow CD34+ selected primary acute myeloid leukemia cells treated with WNT974 1 mM for 24 h (*P<0.05, **P<0.01 by the ttest). GAPDH was used as a normalizer.

haematologica | 2022; 107(1)

307

Letters to the Editor

Figure 2. WNT974 treatment of primary CD34+ acute myeloid leukemia patients’ samples decreased self-renewal but not apoptosis or quiescence. (A) Colonyforming unit (CFU) assay of CD34+ primary acute myeloid leukemia (AML) cells treated with WNT974 drug at different concentrations or vehicle (dimethylsulfoxide, DMSO 0.1% volume). After 2 weeks, colonies were scored and replated. A second scoring was done 2 weeks after replating. The graph shows mean and standard deviation values from a technical triplicate. *P<0.05, **P<0.01, ***P<0.001 (t-test). (B) CFU assay of CD34+ cells from bone marrow of adult healthy donors treated with WNT974 drug at different concentrations or vehicle (DMSO 0.1% volume). After 2 weeks, colonies were scored and replated. A second scoring was done 2 weeks after replating. All tests (t-tests) were non-significant. (C) CellTrace Violet assay of three CD34+ selected primary AML cells treated with WNT974 at various concentrations, or vehicle (DMSO 0.1%). 7-AAD staining was conducted to select only the living cells. The graphs show the different counts at day 0 (dark blue, only control), and at day 6 (3 days of incubation, plus 3 days of treatment or vehicle). No values are statistically significant except for that for patient 2 at the higher dose of WNT974 (P=0.0077, t-test). (D) Annexin V-FITC and propidium iodide (PI) analysis by flow cytometry of CD34+ selected primary AML cells treated with WNT974 or control at different concentrations for 72 h. Living (Annexin–, PI–), Early apoptosis (Annexin+, PI–), Late apoptosis (Annexin+, PI+) and Necrosis (Annexin–, PI+). Each analysis was conducted in technical triplicates. *P<0.05, **P<0.01 (t-test).

308

haematologica | 2022; 107(1)

Letters to the Editor

expression (Figure 1B). We also confirmed that the drug efficiently targets the Wnt pathway by using a luciferase reporter vector in HCT116, a colorectal cancer cell line that shows constitutive activation of the Wnt pathway (Figure 1C). We also confirmed Wnt targeting by WNT974 in primary AML samples (Online Supplemental Table S1 for patients’ details). As shown in Figure 1D, WNT974 decreased mRNA expression of the MYC gene, which is considered one of the main oncogenes driven by

abnormal Wnt/b-catenin signaling. We also observed a similar reduction in mRNA expression of other Wnt target genes such as CTNNB1 and GSK3B after WNT974 treatment of primary AML samples (Online Supplementary Figure S1A). Next, we wanted to investigate whether the treatment was able to affect LSC functions in primary AML patients’ samples. To assess whether targeting the Wnt pathway affects leukemia cell self-renewal, bone marrow

Figure 3. WNT974 activity in a primary murine model of acute myeloid leukemia. (A) Colony-forming unit assay of CD117+ m MllPTD/WT/Flt3ITD/WT bone marrow cells treated with WNT974 at 0.5 and 1 mM or vehicle (dimethylsulfoxide, DMSO 0.1%) for 2 weeks. Results from the first scoring are shown in the left panel. After scoring, the cells were replated in fresh Methocult media and scored after another 2 weeks (right panel). Student t-test analysis: *P<0.05, **P<0.01 and ***P<0.001. (B) Flow chart of the in vivo re-transplantations experiment. (C) Overall survival of secondary transplanted mice (n=5 mice per group). (D) Evaluation of engraftment at 1 month. *P<0.05 (t-test).

haematologica | 2022; 107(1)

309

Letters to the Editor

CD34+-selected cells from five AML patients were treated with WNT974 or the control (DMSO) and used in colony-forming unit (CFU) assays. We found no significant differences in the number of colonies in primary CFU after 2 weeks of culture (Figure 2A, Online Supplementary Figure S1B, left panels). Primary colonies were harvested and then re-plated in methylcellulose for an additional 14 days. We found significant decreases in the number of secondary colonies in WNT974-treated cells compared with cells treated with the control, suggesting that WNT974 targeting has an impact on LSC self-renewal (Figure 2A, Online Supplementary Figure S1A, right panels). We performed the same experiment using bone marrow from three adult healthy donors and CD34+ cells from three cord blood specimens. As shown in Figure 2B and Online Supplementary Figure S1C, the treatment did not affect normal cells. Next, we analyzed whether WNT974 has a role in regulating LSC quiescence by using cell membrane labeling retention assays. For these experiments, CD34+-selected samples from three AML patients were stained with cell trace violet (CTV) dye, which incorporates into the cell membrane and can be detected by flow cytometry. CD34+-selected AML patients’ cells were isolated, labeled with CTV, and treated for 3 days with WNT974 and control. The number of viable (7-AADneg), quiescent (CTVhigh/CD34+) cells was then determined by flow cytometry. CTVhigh/CD34+ cells have been shown previously to increase leukemia-initiating ability, and enhanced engraftment in NSG mice. However, in all analyzed patients’ samples, we did not find a significant decrease in the number of CTVhigh/CD34+ cells after treatment with WNT974, compared with controls, except for patient 2 at the highest WNT974 dose (t-test; P=0.0077) (Figure 2C). In order to assess whether the treatment can affect the bulk cell population, we performed an annexin-V/propidium iodide assay on the same CD34+ AML patients’ samples treated with WNT974 after 72 h, and found only a slightly increase of apoptosis at the higher concentration (Figure 2D). Finally, we investigated whether WNT974 could also effectively target LSC in vivo. To test the effects of WNT974 in vivo, we used our well-characterized MllPTD/WT/Flt3ITD/WT double knock-in CN-AML mouse model.15 This model develops an aggressive AML with 100% penetrance and leads to death within 5 to 8 weeks in secondary bone marrow transplantation.15 First, we confirmed that WNT974 treatment has a similar effect on the primary murine AML cells in this model by conducting a CFU in vitro assay on CD117+ murine cells from bone marrow of AML mice. We observed that the treatment was able to reduce proliferation after the re-plating (Figure 3A) similar to the treatment in human AML cells. Next, we transplanted MllPTD/WT/Flt3ITD/WT leukemic cells (CD45.2+) into lethally irradiated wild-type (WT)-BoyJ (CD45.1+) mice together with whole bone marrow cells from WT-BoyJ donors (CD45.1+). Two weeks after engraftment, mice were treated with WNT974 or vehicle (control) at a dose of 5 mg/kg twice a day by oral gavage for 1 week. The design of the experiment is shown in Figure 3B. After the last dose, mice were sacrificed and bone marrow and spleen cells were obtained. We measured the Wnt target gene MYC as a surrogate for Wnt pathway targeting in leukemic cells in vivo. We found significant downregulation of murine Myc in spleen cells that were used during the primary transplants (Online Supplementary Figure S2A). Next, we determined whether Wnt pathway knock-down by WNT974 affected leukemia engraftment in secondary bone marrow trans310

plantation, a key feature of stem cells. Leukemic donor cells obtained from mice spleen treated with WNT974 or vehicle were transplanted into lethally irradiated BoyJ recipients. Despite observing a reduction of Wnt pathway target gene expression in the spleens of secondary transplanted mice (Online Supplementary Figures S2B-E), we did not observe any differences between overall survival (Figure 3C), or engraftment at 1 month (Figure 3D), after re-transplantation. In conclusion, the data show that WNT974 treatment was able to reduce Wnt pathway activity in leukemic cells and decreased self-renewal of primary AML LSC in vitro. However, WNT974 treatment alone was unable to impact LSC functions in vivo. For these experiments, we used high doses of WNT974, higher than the equivalent doses used in human clinical trials with this drug. Thus, it is unlikely that the lack of in vivo effect is due to poor pharmacokinetics or pharmacodynamics. This is supported by the fact that we observed Wnt target downregulation in leukemic cells in the treated mice. Another possibility is that other pathways or even Wnt reactivation could compensate for the initial Wnt downregulation in LSC, and this could be driven by the microenvironment. It is likely that a combination of WNT974 with other active agents in AML could lead to sustained targeting and eradication of LSC in AML as was shown in chronic myeloid leukemia.14 Felice Pepe,1 Marius Bill,1 Dimitrios Papaioannou,1 Malith Karunasiri,1 Allison Walker,1 Eric Naumann,1 Katiri Snyder,1 Parvathi Ranganathan,1,2 Adrienne Dorrance1,2 and Ramiro Garzon1,2 1 The Ohio State University Comprehensive Cancer Center and 2 Division of Hematology, Department of Internal Medicine, The Ohio State University, Columbus, OH, USA Correspondence: RAMIRO GARZON - ramiro.garzon@osumc.edu. doi:10.3324/haematol.2020.266155 Received: July 7, 2020. Accepted: September 8, 2021. Pre-published: September 16, 2021. Disclosures: no conflicts of interest to disclose. Contributions: RG conceived the idea; RG and FP designed the study; FP, MK and EN performed the in vitro experiments; FP, MB and KS performed the in vivo experiments; FP, DP, PR, AD and RG analyzed and interpreted the data; RG supervised the study; FP, AD and RG drafted the manuscript.

References 1. Steinhart Z, Angers S. Wnt signaling in development and tissue

homeostasis. Development. 2018;145(11):dev146589. 2. Luis TC, Ichii M, Brugman MH, Kincade P, Staal FJT. Wnt signaling strength regulates normal hematopoiesis and its deregulation is involved in leukemia development. Leukemia. 2012;26(3):414-421. 3. Staal FJ, Sen JM. The canonical Wnt signaling pathway plays an important role in lymphopoiesis and hematopoiesis. Eur J Immunol. 2008; 38(7):1788-1794. 4. Luis TC, Naber BAE, Roozen PPC, et al. Canonical wnt signaling regulates hematopoiesis in a dosage-dependent fashion. Cell Stem Cell. 2011;9(4):345-356. 5. Staal FJ, Famili F, Garcia Perez L, Pike-Overzet K. Aberrant Wnt Signaling in Leukemia. Cancers (Basel). 2016;8(9):78. 6. Ashihara E, Takada T, Maekawa T. Targeting the canonical Wnt/beta-catenin pathway in hematological malignancies. Cancer Sci. 2015;106(6):665-671. 7. Wang Y, Krivtsov AV, Sinha AU, et al. The Wnt/beta-catenin pathway is required for the development of leukemia stem cells in AML. Science. 2010;327(5973):1650-1653. 8. Minke KS, Staib P, Puetter A, et al. Small molecule inhibitors of WNT signaling effectively induce apoptosis in acute myeloid leukemia cells. Eur J Haematol. 2009;82(3):165-175. 9. Fiskus W, Sharma S, Saha S, et al. Pre-clinical efficacy of combined

haematologica | 2022; 107(1)

Letters to the Editor

therapy with novel b-catenin antagonist BC2059 and histone deacetylase inhibitor against AML cells. Leukemia. 2015;29(6):12671278. 10. Griffiths EA, Golding MC, Srivastava P, et al. Pharmacological targeting of b-catenin in normal karyotype acute myeloid leukemia blasts. Haematologica. 2015;100(2):e49-e52. 11. Jiang X, Mak PY, Mu H, et al. Disruption of Wnt/b-catenin exerts antileukemia activity and synergizes with FLT3 inhibition in FLT3mutant acute myeloid leukemia. Clin Cancer Res. 2018;24(10):24172429. 12. Proffitt KD, Madan B, Ke Z, et al. Pharmacological inhibition of the Wnt acyltransferase PORCN prevents growth of WNT-driven mam-

haematologica | 2022; 107(1)

mary cancer. Cancer Res. 2013;73(2):502-507. 13. Liu J, Pan S, Hsieh MH, et al. Targeting Wnt-driven cancer through the inhibition of Porcupine by LGK974. Proc Natl Acad Sci U S A. 2013;110(50):20224-20229. 14. Agarwal P, , Zhang B, Ho Y, et al. Enhanced targeting of CML stem and progenitor cells by inhibition of porcupine acyltransferase in combination with TKI. Blood. 2017;129(8):1008-1020. 15. Zorko NA, Bernot KM, Whitman SP, et al. Mll partial tandem duplication and Flt3 internal tandem duplication in a double knock-in mouse recapitulates features of counterpart human acute myeloid leukemias. Blood. 2012; 120(5):1130-1136.

311

Letters to the Editor

utero THS exposure on hemostasis and thrombosis. Thus, the tail bleeding time assay revealed that the THSexposed males and females exhibit substantially shortened bleeding time compared to clean air exposed controls (Figure 1A). In fact, the average bleeding time in males was 395 ± 65.14 seconds (sec) in clean air group versus 68.80 ± 14.87 sec in the THS group; whereas in females it was 449.40 ± 45.07 seconds and 44.50 ± 12.56 sec for clean air and THS groups, respectively. As the time needed for cessation of bleeding was significantly reduced in the in utero THS–exposed mice, we therefore hypothesized that these mice are more vulnerable to thrombosis. This was tested by employing the FeCl3 carotid artery injury–induced thrombosis model. As depicted in Figure 1B, in utero THS mice of both sexes displayed a significant reduction in the occlusion time, with the average in males being 1,080 ± 60.00 sec in clean air group versus 210.80 ± 79.37 sec in THS group; whereas the females recorded 1,150 ± 114.30 sec and 281 ± 116.50 sec for the control and experimental groups, respectively. Taken together, these results show that in utero THS exposure enhances hemostasis and renders mice at higher risk of developing thrombosis. However, when males are compared with females, the results did not show any sex-based differences for either hemostasis or thrombus formation. Notably, the platelet and other blood cells count was measured in both in utero THS- and clean air-exposed mice, as changes in platelet number may contribute to the hemostasis and thrombosis phenotype observed. The in utero THS exposure did not affect the platelet count or other hematological parameters (Table 1). In light of the bleeding time and thrombosis data, another set of experiments evaluated the manifestation of the potential prothrombotic phonotype at the level of platelet physiology by studying platelet functional parameters in vitro. Hence, we first determined the effect of in utero THS exposure on agonist-induced platelet aggregation, which was found (Figure 2A) to be substantially increased, in response to either thrombin (0.1 U/mL) or ADP (1 µmol/L) in male and female mice exposed to THS in utero. However, when comparing platelet aggregation of both sexes, our analysis did not

In utero thirdhand smoke exposure modulates platelet function in a sex-dependent manner Thirdhand smoke (THS), the persistent residue of tobacco smoke that remains after a cigarette is extinguished, materialized as a threat for human health over the last decade. These toxic residues end up depositing on surfaces and objects where tobacco has been used (e.g., homes) and persist for weeks/months after the last smoking.1 THS toxicants undergo chemical reactions and changes over time potentially making them more toxic.2 Given that the routes of exposure to THS involve skin absorption, inhalation and ingestion,3 it is thought to be more toxic by producing more toxicants in the blood of the exposed person.4 Indeed, there is a growing body of evidence documenting THS-induced health risks,5 including cardiovascular disease (CVD). For example, we previously showed that THS exposure modulates platelet function and enhances thrombogenesis in adult exposed mice.6 However, it has not yet been established whether prenatal/in utero THS exposure impacts platelet function and related disorders, which is paramount since the developing embryo is especially sensitive to environmental toxicants, including cigarette smoke.7 Therefore, this study was designed to address this issue, utilizing the offspring of exposed females. In addition, we also examined whether sex differences exist in THS-induced effects. We employed our innovative THS exposure approach which has been peer-reviewed4,6 and accepted as one that provides exposure conditions that mimic those in multiple real-life human situations.4 According to our experimental design, the female breeders were exposed to THS smoke or clean air starting 1 week before mating and throughout the whole pregnancy period by placing them in cages that are furnished with either THS or clean-air exposed materials. After delivery (post-natal day 4), the offspring were moved to clean air cages and housed until 8-10 weeks of age, before experimentation. All animal experimental protocols were approved by the Institutional Animal Care and Use Committee. We first sought to investigate the in vivo effect of in

Figure 1. In utero thirdhand smoke exposure shortens the bleeding time in the tail bleeding time assay, and the time to occlusion in the ferric chloride in vivo thrombosis model both in males and females. (A) Tail bleeding time assay in the in utero thirdhand smoke (THS)- and clean air (CA)–exposed mice compared in males and females. Each point represents the tail bleeding time of a single animal. (B) Ferric chloride–induced thrombosis model (time to occlusion) in the in utero THS- and clean air–exposed compared in males and female mice. Each point represents the occlusion time of a single animal. *Male and female data compared using two-way ANOVA while THS vs. CA comparison within the same sex was done using Student’s t-test.

312

haematologica | 2022; 107(1)

Letters to the Editor

Figure 2. Effect of in utero thirdhand smoke exposure on platelet physiology in the exposed male and female mice compared to clean air-exposed controls. (A) Thirdhand smoke (THS) in utero exposure enhanced platelet aggregation in the exposed male and female mice compared to clean air (CA)-exposed controls. Platelets from THS- in utero exposed and CA–exposed mice were stimulated with 1 μmol/L ADP or 0.1 U/mL thrombin before their aggregation response was measured. The experiment was repeated 3 times, with blood pooled from at least 6-8 mice each time. (B) In utero THS exposure enhanced ATP secretion in the in utero exposed male and female mice, compared to CA-exposed controls. Platelets from THS in utero exposed and CA–exposed mice were stimulated with 1 μmol/L ADP or 0.1 U/mL thrombin before their dense granule secretion (ATP secretion) was determined. Dense granules secretion responses were measured in a lumi-aggregometer. Platelets were incubated with luciferase/luciferin (12.5 μL) for the dense granule measurements. The experiment was repeated 3 times, with blood pooled from at least 6-8 mice each time. *Female and male data compared and P value calculated by two-way ANOVA while comparison within the same sex was done using Student’s t-test. Error bars represent standard deviation.

haematologica | 2022; 107(1)

313

Letters to the Editor

Table 1. Peripheral blood cell counts in in utero thirdhand smoke- and clean air–exposed male and female mice.

Cell type Platelets MPV Red blood cells Lymphocytes Monocytes Granulocytes HCT

Clean air

627.40 ± 88.57 4.84 ± 0.27 6.92 ± 0.50 1.40 ± 0.40 0.13 ± 0.06 1.86 ± 0.52 32.00 ± 2.25

Male in utero THS

649.20 ± 42.13 4.82 ± 0.08 7.45 ± 0.27 1.93 ± 0.70 0.09 ± 0.02 2.39 ± 0.80 34.13 ± 1.41

P value 0.63 0.89 0.11 0.18 0.12 0.24 0.16

Clean air

504.80 ± 46.98 4.82 ± 0.13 6.71 ± 0.83 1.60 ± 0.49 0.13 ± 0.06 1.86 ± 0.50 31.98 ± 4.05

Female in utero THS

517.30 ± 7 7.80 4.86 ± 0.11 6.69 ± 0.57 2.17 ± 0.59 0.09 ± 0.02 2.51 ± 0.67 31.06 ± 3.17

P value 0.77 0.62 0.96 0.13 0.12 0.14 0.69

All counts are expressed as thousands per microliter, except for red blood cells, which are expressed as millions per microliter. Data are presented as mean ± standard deviation. HCT: hematocrit; MPV: mean platelet volume; THS: thirdhand smoke.

reveal a statistical significance, with either of the agonists. Given that platelet granule secretion is known to contribute significantly to platelet activity,8 we investigated agonist-induced ATP release and P-selectin surface expression as markers for dense and a-granules release, respectively. Dense granules as well as a-granules secretion were increased in platelets obtained from in utero THS exposed mice, in response to either ADP or thrombin (Figure 2B; Online Supplementary Figure S1i and ii). These data revealed that platelet secretion contributes to the THS prothrombotic phenotype. In terms of sexdependent differences, the in utero THS-exposed males showed much higher ADP-induced dense granule secretion compared to females, but no statistical difference was observed in the clean air-exposed mice (Figure 2B). Moreover, no differences between the two sexes were observed with thrombin regardless of exposure type (Figure 2B). In contrast, a-granules secretion was significantly elevated in THS exposed females compared to males following stimulation by thrombin; but this was not the case with ADP (Online Supplementary Figure S1i and ii). Next, we investigated the impact of in utero THS on aIIbb3 activation; which was more pronounced in the THS exposed mice, in response to 5 µmol/L ADP and 0.1 U/mL thrombin (Online Supplementary Figure S1iii and iv); which is in accordance with the enhanced aggregation response and was demonstrated in both sexes. Interestingly, our analysis revealed a significant sexbased difference (higher in males) with both agonists. As platelets are activated, phosphatidylserine (PS) becomes exposed at their outer surface, for the assembly of coagulation factor complexes.9 Subsequently, we determined the impact of in utero THS exposure on PS expression. We found PS expression to be markedly enhanced upon stimulation with thrombin or ADP following THS in utero exposure (Online Supplementary Figure S1v and vi), which was documented in males and females. However, when both sexes were compared, it was found that the ADP effects were more pronounced in THS-exposed females compared to males. In contrast, when thrombin was analyzed, the effects in males were found to be higher than females. This discrepancy between the release of dense granules versus a-granules in males and females might be attributed to the fact that the former was performed using platelet-rich plasma whereas the latter using washed platelets; given that the presence of other plasma factors makes it difficult to assess whether the sex difference is inherent to platelets or related to plasma.10 As for PS exposure, female platelets showed more significant elevation in compari314

son to those from males with ADP. However, this trend is completely reversed when thrombin-stimulated platelets were utilized with a significant elevation in PS in males compared to females. This could be explained by the variation in dose-response between males and female platelets.11 It also should be noted that these are different platelet functional responses. In addition, analogy could be inferred from the race disparity of thrombin PAR4 receptor that triggers enhanced platelet aggregation as well as calcium mobilization when activated in African lineage compared to Caucasian.12 Similarly, a sex disparity in the receptors of different agonists or their downstream signaling pathways could explain the aforementioned discrepancies. Collectively, our functional assays provide evidence that in utero exposure to THS triggers a state of platelet hyperactivity and contributes to the prothrombotic phenotype in the offspring mice. In summary, these data provide evidence that the negative health effects of maternal THS exposure extend to the “non-exposed” offspring. Thus, our findings document for the first time that in utero THS exposure drives platelets into a state of hyperactivity, that manifests in a host of enhanced functional responses (e.g., aggregation). Together, these effects ultimately lead to a prothrombotic phenotype. It is noteworthy that this danger, according to our current and published data, does not only affect “directly” THS-exposed mice as we have shown before6 but expands to the offspring of the exposed pregnant mice as well. Interestingly and importantly, this prothrombotic phenotype endured despite the fact that offspring mice were not exposed to THS as they were moved to clean-air exposed cages until they reached 8-10 weeks of age. These data also highlight the underestimated risk of exposure to THS toxicants that persist up to several months after the last smoking has taken place.1,13 It is also important to note that this phenotype is consistent with the state of hyperactive platelets we reported previously as a result of exposure to other forms of tobacco that are perceived as safe, namely e-cigarettes14 and hookah/waterpipe.15 As for the comparisons between males and females, although no sex differences could be demonstrated in bleeding time, thrombosis or platelet aggregation, we did observe significant differences in dense and a-granule secretion, aIIbb3 activation as well as PS exposure when compared sex-wise. In conclusion, our data clearly demonstrates for the first time that in utero THS exposure modulates the platelet biology in the non-exposed offspring, making them more susceptible to cardiovascular diseases.

haematologica | 2022; 107(1)

Letters to the Editor

Hamdy E. A. Ali,1 Ahmed B. Alarabi,2 Zubair A. Karim,3 Victor Rodriguez,3 Keziah R. Hernandez,3 Patricia A. Lozano,2 Medhat S. El-Halawany,1 Fatima Z. Alshbool2 and Fadi T. Khasawneh1. 1 Department of Pharmaceutical Sciences, Irma Lerma Rangel College of Pharmacy, Texas A&M University, Kingsville; 2 Department of Pharmacy Practice, Irma Lerma Rangel College of Pharmacy, Texas A&M University, Kingsville and 3Department of Pharmaceutical Sciences, School of Pharmacy, The University of Texas at El- Paso, El Paso, TX, USA. Correspondence: FADI T. KHASAWNEH - fkhasawneh@tamu.edu. doi:10.3324/haematol.2021.279388 Received: June 7, 2021 Accepted: September 8, 2021 Pre-published: September 16, 2021 Disclosure: no conflicts of interest to disclose. Contributions: FK conceived studies, revised the manuscript, analyzed data; FA conceived studies, revised the manuscript, analyzed data; HA wrote the manuscript, performed research and data analysis; AA, ZK, PL, KH, VR and ME performed research and data analysis. Funding: research reported in this publication was supported by the National Institute of Environmental Health Sciences and the National Heart, Lung, And Blood Institute of the National Institutes of Health under Awards Number R21ES029345 and R01HL145053. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

References 1. Becquemin MH, Bertholon JF, Bentayeb M, et al. Third-hand smoking: indoor measurements of concentration and sizes of cigarette smoke particles after resuspension. Tob Control. 2010;19(4):347-348. 2. Sleiman M, Gundel LA, Pankow JF, Jacob P 3rd, Singer BC, Destaillats H. Formation of carcinogens indoors by surface-

haematologica | 2022; 107(1)

mediated reactions of nicotine with nitrous acid, leading to potential thirdhand smoke hazards. Proc Natl Acad Sci U S A. 2010;107(15):6576-6581. 3. Adhami N, Starck SR, Flores C, Martins Green M. A health threat to bystanders living in the homes of smokers: how smoke toxins deposited on surfaces can cause insulin resistance. PLoS One. 2016;11(3):e0149510. 4. Martins-Green M, Adhami N, Frankos M, et al. Cigarette smoke toxins deposited on surfaces: implications for human health. PLoS One. 2014;9(1):e86391. 5. Jacob P, 3rd, Benowitz NL, Destaillats H, et al. Thirdhand smoke: new evidence, challenges, and future directions. Chem Res Toxicol. 2017;30(1):270-294. 6. Karim ZA, Alshbool FZ, Vemana HP, et al. Third hand smoke: impact on hemostasis and thrombogenesis. J Cardiovasc Pharmacol. 2015;66(2):177-182. 7. Talbot P, Lin S. The effect of cigarette smoke on fertilization and pre-implantation development: assessment using animal models, clinical data, and stem cells. Biol Res. 2011;44(2):189-194. 8. Dawood BB, Wilde J, Watson SP. Reference curves for aggregation and ATP secretion to aid diagnose of platelet-based bleeding disorders: effect of inhibition of ADP and thromboxane A(2) pathways. Platelets. 2007;18(5):329-345. 9. Gao C, Xie R, Yu C, et al. Procoagulant activity of erythrocytes and platelets through phosphatidylserine exposure and microparticles release in patients with nephrotic syndrome. Thromb Haemost. 2012;107(4):681-689. 10. Leng XH, Hong SY, Larrucea S, et al. Platelets of female mice are intrinsically more sensitive to agonists than are platelets of males. Arterioscler Thromb Vasc Biol. 2004;24(2):376-381. 11. Johnson M, Ramey E, Ramwell PW. Sex and age differences in human platelet aggregation. Nature. 1975;253(5490):355-357. 12. Edelstein LC, Simon LM, Montoya RT, et al. Racial differences in human platelet PAR4 reactivity reflect expression of PCTP and miR-376c. Nat Med. 2013;19(12):1609-1616. 13. Matt GE, Quintana PJ, Zakarian JM, et al. When smokers move out and non-smokers move in: residential thirdhand smoke pollution and exposure. Tob Control. 2011;20(1):e1. 14. Qasim H, Karim ZA, Silva-Espinoza JC, et al. Short-term e-cigarette exposure increases the risk of thrombogenesis and enhances platelet function in mice. J Am Heart Assoc. 2018;7(15):e00264. 15. Alarabi AB, Karim ZA, Ramirez JEM, et al. Short-term exposure to waterpipe/hookah smoke triggers a hyperactive platelet activation state and increases the risk of thrombogenesis. Arterioscler Thromb Vasc Biol. 2020;40(2):335-349.

315

Letters to the Editor

Ikaros deficiency is associated with aggressive BCR-ABL1 B-cell precursor acute lymphoblastic leukemia independent of the lineage and developmental origin The B-cell lineage is established from waves of differentiation that begin in the fetal liver (FL) and continue in the postnatal bone marrow (BM). In mice, FL lymphopoiesis gives rise to B1 cells, which produce natural polyreactive immunoglobulin M (IgM) antibodies and are implicated in innate immunity. Conversely, BM lymphopoiesis favors conventional B2-cell development over B1, and is essential for adaptive immunity.1,2 The features that distinguish B1- and B2-cell differentiation remain partially understood. Ikaros, encoded by IKZF1, is a key transcriptional regulator of B lymphopoiesis in mice and humans.3 It promotes B2-cell development at multiple stages, including large pre-B-cell (Hardy fraction C') differentiation where it antagonizes IL-7/STAT5 signaling.4 Ikaros is also a tumor suppressor in B2 progenitors, and deleterious IKZF1 alterations are prominent in BCR-ABL1+ B-cell precursor acute lymphoblastic leukemias (BCP-ALL) with constitutively active STAT5.5 While Ikaros is likewise important for fetal and adult B1cell development, the exact stage where it is required has remained unclear, due to a dearth of in-depth knowledge about B1-cell differentiation. Whether Ikaros acts as a tumor suppressor in B1 cells, or in fetal B cells in general, is currently unknown. Here we show that B1-cell differentiation parallels that of B2 cells in terms of phenotype and cell cycle status. We show that Ikaros promotes B1cell differentiation at a stage equivalent to fraction C' of B2 cells, by modulating the expression of genes involved in cell proliferation and migration, pre-B-cell receptor

(BCR) signaling, Ig-HC rearrangement and IL-7/STAT5 signaling. We also show that Ikaros inhibits the expansion of murine BCR-ABL1+ B1 progenitors from the FL and BM, and suggest that this function might be conserved in human FL-derived B cells. In order to determine the role of Ikaros in adult B1-cell differentiation, we analyzed the BM of Ikf/fMb-1-Cre+ conditional knockout (cKO) and Ikf/fMb-1-Cre- wild-type (WT) mice (Figure 1A), where the Cre recombinase is controlled by the endogenous Cd79a promoter.4 While the percentage of WT B1 progenitors (Lin-IgMCD93+CD19+B220-/lo) remained low with age, the same population in the cKO mice was ~20x higher until 4 weeks of age, before decreasing to WT levels by week 16 (Figure 1B). This increase was not due to a compensatory decrease in B2 progenitors, as the percentage of BM B2 progenitors (B220+CD19+CD43+) was stable over time (Online Supplementary Figure S1A). Ikaros loss in B1 progenitors was confirmed by intracellular staining (Online Supplementary Figure S1B). These results showed that Ikaros is required for differentiation in B1 progenitors from an early pro-B1 cell stage. In order to pinpoint the stage where Ikaros promotes B1-cell differentiation, we dissected the above B1 progenitor cell population, by analyzing Ig-HC rearrangement, and evaluating CD24 and BP-1 expression, typically used to define B2-cell development.6 WT or cKO B1 progenitors were heterogenous with unrearranged, germline HC genes, and rearranged DJ as well as proximal and distal VDJ genes (Figure 2A). WT B1 progenitors were mostly CD24-BP-1- and CD24+BP-1-, resembling Hardy fractions A and B among B2 progenitors (Figure 2B). Interestingly, cKO cells were mainly CD24+BP-1+ and CD24++BP-1+, and resembled B2 fractions C and C'. Hence, B1 progenitors are more heterogenous than pre-

Figure 1. B1 progenitors accumulate in Ikaros conditional knockout bone marrow. (A) Representative flow cytometry analysis of bone marrow (BM) B1 progenitors (Lin-IgM-CD93+CD19+B220-/lo) from 4-week-old mice. Numbers correspond to % of parental population. (B) % BM B1 progenitors in wild-type (WT) or conditional knockout (cKO) mice with age (D: days; W: weeks after birth). Significance was calculated using unpaired Student's t-test. ns: not significant; *P<0.05; ***P<0.001; ****P<0.0001. Ikf/fMb-1-Cre mice4 were housed in specific pathogen-free (SPF) conditions or kept in single-ventilated cages. IgM: immunoglobulin M.

316

haematologica | 2022; 107(1)

Letters to the Editor

viously described, and correspond to phenotypic stages like those of B2 fractions.7 These results suggested that Ikaros deficiency blocks B1-cell differentiation at the fraction C' stage, similar to its effect on B2-cell development, using the same Cre transgene.4 In order to evaluate the molecular pathways regulated by Ikaros, we analyzed the gene expression changes

between WT and cKO BM B1 progenitors by RNA sequencing (RNA-seq) (Online Supplementary Figure S2). Pathway enrichment and gene set enrichment analyses (GSEA) associated Ikaros with cell migration, adhesion, proliferation, and B-cell differentiation and maturation (Figure 2C and D). As Ikaros antagonizes STAT5 binding to DNA in pre-B2 cells, we asked if Ikaros has a similar

Figure 2. Ikaros deficiency in B1 progenitors induces a differentiation block at fraction C' and deregulation of genes involved in B-cell differentiation. (A) VDJ rearrangement as indicated, following published protocols.4 Spleen and kidney cells were used as positive or negative controls, respectively. Germ: germline configuration. Numbers correspond to kilobases. (B) Bone marrow (BM) cells were analyzed for different B-cell stages by flow cytometry: B1 progenitors (IgMCD93+CD19+B220-/lo) were further gated on CD24 and BP-1 to give fraction (Fr.) A (CD24-BP-1-), Fr. B (CD24+BP-1-), Fr. C (CD24+BP-1+), Fr. C' (CD24+BP-1++). Graphs show frequencies (left) and absolute numbers (right) of B progenitor fractions as indicated. Data from 3 independent experiments with 5 mice for wildtype (WT) and 7 for conditional knockout (cKO). ns: not significant; **P<0.01; ****P<0.0001 (unpaired Student's t-test). (C) RNA sequencing (RNA-Seq) data of BM B1 progenitors from 3 WT and 3 cKO mice (4 weeks old) were evaluated by pathway enrichment analysis (http://metascape.org)15 for genes upregulated in cKO (left) or WT B1 progenitors (right). (D) Gene set enrichment analysis (GSEA) of 3 gene sets as indicated, compared with genes upregulated (red) or downregulated (blue) in cKO vs. WT B1 progenitors. The STAT5 gene set corresponds to STAT5 genes antagonized by Ikaros.8 NES: normalized enrichment score; P: P-value; q: false discovery rate (FDR) q-value.

haematologica | 2022; 107(1)

317

Letters to the Editor

function in B1 progenitors.8 GSEA of STAT5-induced genes antagonized by Ikaros in pre-B2 cells revealed their enrichment among the upregulated genes in Ikaros deficient B1 cells, suggesting that Ikaros also antagonizes STAT5 gene activation in B1 progenitors. Thus, Ikaros appears to affect similar biological pathways in B1 and B2 progenitors. That Ikaros deficiency correlates with increased STAT5-activated gene expression suggests that Ikaros may act as a tumor suppressor in B1 cells. We transduced BM WT and cKO B1 progenitors to express the BCRABL1 oncoprotein, or an empty vector (pMITo), carrying a TdTomato (TdTo) reporter. Transduction efficiency was assessed and BCR-ABL1 expression was confirmed by western blot (Online Supplementary Figure S3A and B). TdTo+ cells were injected into NSG mice and the health status was monitored over time. All mice that received cKO B1 BCR-ABL1+ cells developed BCP-ALL, as characterized by BM failure (anemia, thrombocytopenia, splenomegaly) (Online Supplementary Figure S3C), which was sometimes associated with hepatomegaly and neurological impairment. Blast cell (CD19+TdTo+) infiltration was observed in the BM and spleen (Figure 3A). In contrast, only 25% of mice transplanted with WT B1 BCR-ABL1+ cells developed BCP-ALL, with slower kinetics and fewer symptoms (Figure 3B). We also analyzed

symptom-free mice that received WT B1 BCR-ABL1+ or cKO B1 pMITo+ cells, but did not detect CD19+TdTo+ cells in the BM or spleen (Figure 3A). The fact that not all mice that received WT B1 BCR-ABL1+ cells developed leukemia, as previously described, may be due to the limiting numbers of cells injected here.9 Our results therefore indicated that Ikaros limits the ability of BM B1 progenitors to develop BCR-ABL1-induced BCP-ALL. In order to determine if Ikaros also functions as a tumor suppressor in fetal-derived B1 cells, we transduced WT and cKO FL B1 progenitors (from E17-E18 organs) to express BCR-ABL1 and injected them into NSG mice. Fetal B1 progenitors gave rise to BCP-ALL similar to BM B1 cells. Ikaros deficiency was associated with a higher tumor burden and decreased survival (Figure 3B). These results demonstrated that Ikaros reduces BCP-ALL development in B1 progenitors. Lastly, we asked if Ikaros loss is implicated in human B1-like BCP-ALL. Because human B1 cells are ill-defined, we first compared a published set of genes, differentially expressed between human B1- and non-B1-like BCPALL, with our available human BCP-ALL RNA-seq dataset, but this did not identify clear subgroups among samples.10 We then separated a collection of 370 pediatric and 102 adult BCP-ALL samples into B1-like and non-B1-like groups according to Vh usage, and further

Figure 3. Continued on following page.

318

haematologica | 2022; 107(1)

Letters to the Editor

Figure 3. Loss of Ikaros in BCR-ABL1 transformed B1 progenitors of adult and fetal origin results in aggressive leukemia. (A) Bone marrow (BM) B1 progenitor cells were retrovirally transduced with BCR-ABL1-IRES-TdTomato (BCR-ABL) or the empty vector (pMITo). NSG mice were intraveously injected with an equivalent of 104 TdTo+ cells as indicated. Recipient BM and spleen cells were analyzed by flow cytometry for CD19+TdTo+ cells when the mice showed signs of sickness ("sick" conditional knockout [cKO] BCR-ABL=29 days, sick wild-type [WT] BCR-ABL=56 days after transplantation) or at the corresponding "healthy" timepoints. The graph on the right shows the survival of NSG mice transplanted with cKO B1 progenitors pMITo+ (control), WT B1 progenitors BCR-ABL1+ (WT) or cKO B1 progenitors BCR-ABL1+ (cKO). The graph represents the results of 4 independent experiments with 16 WT, 14 cKO and 3 control mice. MS: median survival. (B) FL B1 progenitors were transduced with a BCR-ABL1-IRES-TdTomato construct. NSG mice were injected with an equivalent of 104 TdTo+ cells. BM and spleen cells from sick mice were analyzed for CD19+TdTo+ cells. The survival curve of 2 independent experiments with 8 WT and 7 cKO mice is shown on the right. MS: median survival. Ikf/fMb-1-Cre and NSG mice were housed in specific pathogen-free (SPF) conditions or kept in single-ventilated cages. All experiments were approved (APAFIS reference #9742-2017042718317841 v6T). (C) Vh usage of pediatric (upper graphs) and adult (lower graphs) B-cell precursor acute lymphoblastic leukemia (BCP-ALL) samples were analyzed according to Griffin et al11 and classified into B1-like (B1) and non-B1-like (nonB1), and further divided into samples with normal or deleted IKZF1. Numbers in the bars represent the % patients with the indicated IKZF1 status. Numbers at the bottom of the bars indicate the numbers of patients analyzed. (D) Pediatric BCP-ALL samples with KMT2A rearrangements were grouped according to IKZF1 status and displayed as a function of patient age. Pediatric data were obtained from the Genetics Department of the Robert Debré Hospital (Paris, France), and adult data from the Genome and Cancer Department of the Saint Louis Hospital (Paris, France). Informed consent was obtained according to the Helsinki Declaration.

analyzed them for IKZF1 deletions (gross chromosomal deletions of chromosome 7, -7p, intragenic deletions) (Figure 3C). In both children and adult samples, the ratio of mutant to functional IKZF1 samples was similar between B1- and non-B1-like BCP-ALL.11 Since IKZF1 alterations are frequent in BCR-ABL1+ BCP-ALL, we checked if BCR-ABL1+ leukemias were enriched in B1like cases. Interestingly, there was a significant enrichment of BCR-ABL1+ cases within the B1-like group when compared with the non-B1 group in both pediatric (21% vs. 13.4%) and adult (67% vs. 45%) patients (P<0.05 in both cases; hypergeometric test). Among the pediatric samples, 35 came from infants <1 year of age, and most (34/35) contained KMT2A rearrangements, considered to occur in utero;12 four of these samples exhibited an IKZF1 deletion (Figure 3D). Thus, our results in mice and humans showed that Ikaros loss is associated with BCP-ALL development in B cells of both fetal and adult origin. In conclusion, our study indicates that murine B1-cell development can be divided into phenotypically discrete stages that resemble the Hardy fractions A-C' currently used to evaluate B2-cell differentiation.6 These newly identified populations will allow further investigation into the requirements of B1-cell development and the factors involved. We show that Ikaros is required at the fraction C' stage of B1-cell differentiation as observed for B2 progenitors, suggesting similar regulation of gene expression, particularly in antagonizing genes regulated by IL-7/STAT5.8 Our results also suggest that Ikaros functions as a tumor suppressor by repressing STAT5 activity in B-cell progenitors of fetal and adult origin. Whether the B1-cell equivalent exists in humans is still unclear. However, human fetal B cells and murine B1

haematologica | 2022; 107(1)

progenitors are alike in some ways. Infant and pediatric BCP-ALL cells often co-express lymphoid and myeloid markers;13 similarly, murine B1 cells have been reported to be bipotent and can develop into both B cells and macrophages.14 While our results do not address the validity of human B1 cells, we found IKZF1 alterations in both B1- and non-B1-like leukemias of children and adults, including those with fetus-associated ETV6RUNX1 translocations and KMT2A rearrangements. Thus, Ikaros loss-of-function is associated with BCP-ALL development in patients of all ages. It will be interesting in future studies to determine if B1- and non-B1-like cases are associated with different outcomes, and if this parameter could improve the risk classification of BCPALL patients. Célestine Simand,1,2,3,4,5 Céline Keime,1,2,3,4 Aurélie Caye,6,7 Chloé Arfeuille,6,7Marie Passet,8 Rathana Kim,8 Hélène Cavé,6,7 Emmanuelle Clappier,8 Philippe Kastner,1,2,3,4,9 Susan Chan1,2,3,4# and Beate Heizmann1,2,3,4# 1 Institut de Génétique et de Biologie Moléculaire et Cellulaire (IGBMC), Illkirch; 2Institut National de la Santé et de la Recherche Médicale, INSERM U1258, Illkirch; 3Centre National de la Recherche Scientifique, CNRS UMR7104, Illkirch; 4Université de Strasbourg, Illkirch; 5Service d’Hématologie, Institut de Cancérologie Strasbourg Europe, ICANS, Strasbourg; 6Département de Génétique, Assistance Publique des Hôpitaux de Paris (AP-HP), Hôpital Robert Debré, Paris; 7Institut de Recherche Saint-Louis, INSERM UMR_S1131, Université de Paris, Paris; 8Université de Paris, Laboratory of Hematology, AP-HP, Hôpital Saint-Louis, Paris and 9 Faculté de Médecine, Université de Strasbourg, Strasbourg, France. # SC and BH contributed equally as co-senior authors

319

Letters to the Editor

Corresponding author: PHILIPPE KASTNER - scpk@igbmc.fr doi:10.3324/haematol.2021.279125 Received: June 10, 2021. Accepted: September 17, 2021. Pre-published: September 30, 2021. Disclosures: no conflicts of interest to disclose. Contributions: CS performed experiments, interpreted the data and wrote the manuscript; CK analyzed data; AC, CA, RK and MP prepared and analyzed patients’ samples; HC and EC provided clinical data;. SC and PK designed the research, interpreted the data and wrote the manuscript; BH performed experiments, designed and supervised the research, interpreted the data and wrote the manuscript. Acknowledgments: we thank members of the Chan-Kastner lab for scientific discussion and help, P. Marchal for technical support, the Group for Research in Adult ALL (GRAALL) for providing data of patients’ analysis, the IGBMC GenomEast sequencing platform, the IGBMC flow cytometry facility (C. Ebel, M. Philipps), the IGBMC animal facility (M. Gendron, S. Falcone, W. Magnant, A. Vincent), and the IGBMC cell culture facility. Funding: CS received a predoctoral fellowship from la Ligue Nationale Contre le Cancer (LNCC). This study was supported by grants from Fondation ARC (PJA20171206372 to BH), Agence Nationale de la Recherche (ANR-20-CE15-0011 to BH), LNCC (Equipe Labelissée 2015-2017 to SC), Institut National du Cancer (PLBIO-2015-114 to PK and HC), Agence Nationale de la Recherche (ANR-17-CE15-0023-01 to SC), Fondation pour la Recherche Médicale (Equipe FRM 2019, EQU201903007812), an equipment grant from LNCC Grand Est/Bourgogne Franche Comté (001K.2016 to SC), institute funds from INSERM, CNRS, Université de Strasbourg, and the institute grant ANR-10-LABX-0030-INRT, a French State fund managed by the Agence Nationale de la Recherche under the frame program Investissements d’Avenir ANR-10-IDEX0002-02. The GenomEast platform is a member of the France Génomique consortium (ANR-10-INBS-0009).

320

References 1. Montecino-Rodriguez E, Dorshkind K. B-1 B cell development in the fetus and adult. Immunity. 2012;36(1):13-21. 2. Baumgarth N. A Hard(y) Look at B-1 cell development and function. J Immunol. 2017;199(10):3387-3394. 3. Heizmann B, Kastner P, Chan S. The Ikaros family in lymphocyte development. Curr Opin Immunol. 2018;51:14-23. 4. Heizmann B, Kastner P, Chan S. Ikaros is absolutely required for preB cell differentiation by attenuating IL-7 signals. J Exp Med. 2013;210(13):2823-2832. 5. Marke R, Van Leeuwen FN, Scheijen B. The many faces of IKZF1 in B-cell precursor acute lymphoblastic leukemia. Haematologica. 2018;103(4):565-574. 6. Hardy RR. Resolution and characterization of pro-B and pre-pro-B cell stages in normal mouse bone marrow. J Exp Med. 1991;173(5):1213-1225. 7. Montecino-Rodriguez E, Leathers H, Dorshkind K. Identification of a B-1 B cell-specified progenitor. Nat Immunol. 2006;7(3):293-301. 8. Heizmann B, Le Gras S, Simand C, Marchal P, Chan S, Kastner P. Ikaros antagonizes DNA binding by STAT5 in pre-B cells. PLoS One. 2020;15(11):1-15. 9. Montecino-Rodriguez E, Li K, Fice M, Dorshkind K. Murine B-1 B cell progenitors initiate B-acute lymphoblastic leukemia with features of high-risk disease. J Immunol. 2014;192(11):5171-5178. 10. Fitch B, Roy R, Geng H, et al. Human pediatric B-cell acute lymphoblastic leukemias can be classified as B-1 or B-2-like based on a minimal transcriptional signature. Exp Hematol. 2020;90:65-71. 11. Griffin DO, Holodick NE, Rothstein TL. Human B1 cells in umbilical cord and adult peripheral blood express the novel phenotype CD20 + CD27 + CD43 + CD70 −. J Exp Med. 2011;208(1):67-80. 12. Gale KB, Ford AM, Repp R, et al. Backtracking leukemia to birth: Identification of clonotypic gene fusion sequences in neonatal blood spots. Proc Natl Acad Sci U S A. 1997;94(25):13950-13954. 13. Abdelhaleem M. Frequent but nonrandom expression of myeloid markers on de novo childhood acute lymphoblastic leukemia. Exp Mol Pathol. 2007;83(1):138-141. 14. Montecino-Rodriguez E. Identification of B/macrophage progenitors in adult bone marrow. Semin Immunol. 2002;14(6):371-376. 15. Zhou Y, Zhou B, Pache L, et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun. 2019;10(1):1523.

haematologica | 2022; 107(1)

Letters to the Editor

Planned withdrawal of dexamethasone after pomalidomide low-dose dexamethasone induction for lenalidomide-refractory multiple myeloma (ALLG MM14) Immune dysfunction, a key feature of myeloma (MM), plays an important role in promoting tumor growth and therapy resistance1 with multiple mechanisms of immune evasion described. Pomalidomide (POM) is an immunomodulatory (IMiD) compound2 that mediates direct anti-proliferative effects on tumor cells, as well as immune-modulatory effects on T cells, natural killer (NK) cells and monocytes.3 POM plus low-dose dexamethasone (LoDEX) is a standard treatment option for patients with relapsed/refractory MM (RRMM), however, dexamethasone can antagonize the immunostimulatory capacity of ImiD.3,4 Consequently, the immunostimulatory effects of IMiD may be better exploited in the longer term without concomitant DEX, particularly be relevant in the minimal disease burden setting (i.e., maintenance) when some inherent immune recovery has occurred. To our knowledge, our study is the first to evaluate this in a prospective, randomized manner, demonstrating (i) regulatory T- cell (Treg) depletion following POM-LoDEX induction was partially abrogated following withdrawal of dexamethasone in maintenance, and (ii) enrichment of heterogenous neutrophil populations and an increase in activated NK cells with commensurate decrease in inhibited NK cells following POM-LoDEX induction. ALLG MM14 was a prospective, randomized, multicenter, open-label parallel-group phase II trial comparing POM maintenance to POM-LoDEX maintenance following induction with POM-LoDEX. Eligible patients with RRMM, who had failed at least two prior therapies (including a history of lenalidomide failure [Table 1]) were enrolled. The study was conducted according to the Alfred Hospital Institutional Ethics Review Board, in accordance with the Declaration of Helsinki (ACTRN12615000447550). Patients received four cycles of induction (1 cycle: 28 days): POM (4 mg orally days 1-21) plus LoDEX (40 mg orally days 1, 8, 15, and 22). Patients who achieved stable disease (SD) or better (“responders”) were then ran-

domized (1:1) to continue on one of two arms of maintenance: POM or POM-LoDEX. Accrual continued until 80 patients were randomized. Correlative peripheral blood (PB) samples for immune studies were collected at baseline (pre-induction) and maintenance (C1D1, C3D1, C6D1 and C10D1). The primary objective was to determine whether coadministration of DEX with POM in maintenance significantly impacted NK-cell numbers, by comparing the change in PB NK-cell quantification from baseline to maintenance (C6D1) time points utilizing mass cytometry (CyTOF) (powered to detect an increase of 30% in NK-cell numbers in POM compared to POM-LoDEX). (ALLG MM14 was not powered to detect differences in secondary exploratory/clinical endpoints so conclusions on the clinical impact of one strategy over the other cannot be drawn). Exploratory CyTOF studies analyzed sequential PB samples to define differences in immune cell profiles in: (i) (all patients) responders versus nonresponders; and (ii) (randomized patients) POM versus POM-LoDEX maintenance. Secondary clinical objectives were to compare (following randomization to POM or POM-LoDEX maintenance): (i) survival (progression-free survival/ overall survival [PFS/OS]), (ii) safety/toxicity and (iii) response/survival following initiation of postprogression therapy. For CyTOF analysis cells were stained with sub-set defining antibodies (myeloid, B, T and NK cells) (Online Supplementary Table S1). Supervised analysis was performed to determine differences in canonical immune cell populations (NK cells and Treg), reported as a proportion of population (%).5,6 CD3-CD19-CD56+ NK cells were predefined from patient datasets. Boolean gating was then performed using seven NK-cell activation/inhibitory markers (CD158a/CD158b/CD159a/CD314/CD335/CD336/CD 337). Boolean populations that comprised ≥3% of the total NK-cell population (median) were then compared. A Mann-Whitney test was used to determine statistical significance for each of the defined populations between clinical groups. Analyses of the primary NK endpoints was confined to patients who had assessments at both baseline and maintenance C6D1. Treg (CD3+CD4+CD127loCD25hiCD45RO+) were defined by manual gating and assessed in all patient samples at all time points: a one-way ANOVA with a Kruskal Wallis

Table 1. Characteristics of 154 enrolled patients.

Characteristic Male sex, n (%) Age in years, median (range) ISS Stage Not Known Stage 1 Stage 2 Stage 3 Prior lines of therapy, median (range) Lenalidomide failure* Bortezomib refractory Prior autologous stem cell transplant Prior allograft Prior anti-CD38 therapy Time in years from diagnosis to study enrollment, median (range)

All Patients n=154

POM n=40

POM-LoDEX n=38

79 (51.3%) 67.4 (36.0-88.6)

20 (50.0%) 68.4 (50.3-85.4)

17 (44.7%) 66.2 (36.0-81.1)

66 (42.9%) 35 (22.7%) 36 (23.4%) 17 (11.0%) 4.5 (2-14) 154 (100%) 128 (83.1%) 96 (62.3%) 1 (0.7%) 0 (0.0%)

17 (42.5%) 9 (22.5%) 9 (22.5%) 5 (12.5%) 5 (3-9) 40 (100%) 29 (72.5%) 24 (60.0%) 0 (0.0%) 0 (0.0%)

16 (42.1%) 9 (23.7%) 10 (26.3%) 3 (7.9%) 5 (3-14) 38 (100%) 33 (86.8%) 31 (81.6%) 1 (2.6%) 0 (0.0%)

5.5 (1.2-17.8)

5.9 (2.4-12.8)

6.4 (1.9-17.8)

*Lenalidomide (LEN) failure defined as failing to respond: (1) disease progression during treatment or within 60 days of completing a LEN containing regimen or (2) failure to achieve at least a minimal response (MR) (after 2 cycles). POM: pomalidomide; LoDEX: low-dose dexamethasone. ISS: international staging system.

haematologica | 2022; 107(1)

321

Letters to the Editor

post hoc test for multiple comparisons was used to determine statistical significance. Unsupervised analysis was performed to identify immune cell populations: data were clustered in the Vortex package7 using the x-shift algorithm. Elbow-point validation was used to affirm the correct cluster number. Differences in cluster frequency between groups were assessed by Mann-Whitney test for statistical significance. Cluster phenotypes were determined and validated via multiple visualization approaches; individualized clusters were visualized using brick plots.8 Comparisons of the maintenance arms were restricted to a modified ITT (mITT) set which excluded patients randomized in error. Secondary clinical time-to-event outcomes (PFS/OS) were compared between randomized treatments using log-rank tests and estimates of the survival distributions were calculated using the Kaplan-

Meier method. Two-tailed P-values were used for all comparisons, and, unless otherwise stated, were performed using a significance level of 5%. Toxicity was assessed according to CTCAE version 4.0. 154 patients were enrolled (baseline characteristics listed in Table 1). The estimated median potential follow-up (by reverse Kaplan-Meier) for all registered patients for OS was 27.8 month (mo). Eighty-one patients were randomized, however, three were randomized in error, therefore a mITT analysis included 78 (51%) patients who achieved SD or better with POMLoDEX induction: POM n=40, POM-LoDEX n=38. Median PFS (from time of randomization) was 2.6 mo (95% confidence interval [CI]: 1.8-3.0) for POM versus 5.7 mo (95% CI: 4.5-7.5) for POM-LoDEX (log-rank P=0.051; hazard ration [HR]: 0.63, 95% CI: 0.40-1.00) (Figure 1A). Median OS (from time of randomization)

B Figure 1. Kaplan-Meier survivor functions for modified Intention to treat population (from time of randomization) (mITT: pomalidomide [POM] n=40; LoDEX: pomalidomide low-dose dexamethasone [POM-LoDEX] n=38). In anticipation of early or late differences between the maintenance treatment arms in their time-to-event outcomes, 6 comparisons between the arms were planned at 3, 6, 9, 12, 15 and 18 months (mo) from randomization. To account for multiplicity of comparisons, a Bonferroni adjustment to the alpha-level of each test was implemented, namely a comparison between the treatment arms at one of these time points was judged to be statistically significant if the associated P-value was ≤0.0083. The test was based on the complementary loglog transformation of the survival function. (a) Progression free survival: POM arm 2.6 mo (95% confidence interval [CI]: 1.8-3.0) vs. 5.7 mo (95% CI: 4.5-7.5) for POM-LoDEX (log-rank P=0.051; hazard ratio [HR]: 0.63, 95%CI: 0.40-1.00), early PFS favored POM-LoDEX, however late survival favored POM: a comparison of PFS at 6 3-monthly intervals favored POMLoDEX (3-12 mo, P<0.001) however, at 18 mo, POM was favored (P=0.018). (B) Overall survival: POM arm 25.7 mo (95% CI: 16.7-42.2) vs. 17.4 mo (95% CI: 12.5-NA) for POMLoDEX (P=0.356; HR: 1.36, 95%CI: 0.70-2.64). Like the progression-free survival (PFS) analysis, comparisons of overall survival (OS) at 6 3-monthly intervals demonstrated no difference between the arms at 3-12 mo, however at 15 mo and 18 mo, OS favored POM (P=0.006, P=0.021 respectively).

322

haematologica | 2022; 107(1)

Letters to the Editor

was 25.7 mo (95% CI: 16.7-42.2) for POM versus 17.4 mo (95% CI: 12.5-NA) for POM-LoDEX (log-rank P=0.356; HR: 1.36, 95%CI: 0.70-2.64) (Figure 1B). There was no difference in NK populations observed between responders and non-responders at baseline. However, in responders, (i) inhibited NK cells (CD3CD19-CD56+CD159a+CD158a+) were enriched at baseline and significantly decreased following induction (pooled maintenance timpepoints) (P<0.0001), and (ii) activated NK cells (CD3-CD19-CD56+CD337+CD336+, no inhibitory receptors) were significantly increased following induction (pooled maintenance time points) (P<0.0001) (Figure 2A). Following commencement of maintenance, there was no emergent difference in NK populations observed between treatment arms. There was no difference observed in NK-cell populations according to maintenance arm at baseline and at maintenance (C6D1) (primary objective). There was no difference in Treg percentage (Treg%) between responders and non-responders at baseline. After induction and prior to commencing maintenance (C1D1 timepoint), responders demonstrated a depletion of Treg% (P<0.0001). Following commencement of maintenance, Treg depletion was maintained in patients who continued on POM-LoDEX, whereas POM patients who had LoDEX withdrawn demonstrated a partial

recovery in Treg% (P<0.05). (Figure 2B). Unsupervised analysis (all patients) at baseline defined 131 immune cell populations (Figure 2C): there were no significant differences identified between responders and non-responders. At maintenance (responders), there was enrichment of heterogenous neutrophil populations (pooled maintenance time points). Of the 131 clusters identified at baseline, five of the eight large clusters (each at least 3% [median] of total nucleated cells evaluated) that were significantly enriched (P<0.0001) following POM-LoDEX induction were activated neutrophil populations (all expressed CD66b but with variable expression of CD24/CD16/CD11c/CD11b/CD45RO) (Figure 2D). Online Supplementary Table S2 lists all grade adverse events (AE). When comparing the mITT population, the incidence of AE was generally similar, including hematologic toxicity. Significant differences were observed in the incidence of lung infections (higher in POM-LoDEX, P=0.003) and peripheral sensory neuropathy (higher in POM-LoDEX, P=0.041). Median durations of exposure to maintenance POM were 2.5 mo and 6.2 mo in the POM and POM-LoDEX arms respectively. Dose intensity interquartile ranges were similar in both arms from maintenance C1D1 through to C6D1. Online Supplementary Figure S1 shows results for survival post-

Figure 2. Continued on following page.

haematologica | 2022; 107(1)

323

Letters to the Editor

Figure 2. Mass cytometry: supervised analysis (A and B), unsupervised analysis (C and D). (A) In responders (comparing baseline [pre-induction] time point to pooled maintenance time points), activated natural killer (NK) cells were enriched whereas inhibited NK cells were reduced. (****P<0.0001). (B) In responders, regulatory T cells (Treg) were depleted by pomalidomide low-dose dexamethasone (POM-LoDEX) induction (comparing baseline [pre-induction] time point with maintenance C1D1 timepoint). Following withdrawal of DEX in maintenance, some Treg recovery seen in patients on the POM only arm (*P<0.05, ****P<0.0001) (*P<0.05, **** P<0.0001). (C) Cluster analysis at baseline identified 131 immune cell populations. Plot shows a single cell force directed representation of peripheral blood. Individual clusters are indicated by colours. (D) In responders, neutrophil populations were enriched at maintenance time points (pooled). Plots show frequency of unique neutrophil clusters #1-5 out of total patients’ cells for induction and pooled maintenance samples. Example brick plot phenotype (indicative of all neutrophil populations) showing expression of CD66b, CD24, CD16, CD11c, CD11b and CD45RO. Large bricks indicate high relative expression, small bricks indicate low relative expression. Absent bricks indicate no expression of the given marker. (****P<0.0001)

progression therapy, which favored patients randomized to the POM arm. Immune dysfunction is a key feature of MM. In MM, the number and function of NK cells have been shown by several groups to affect clinical outcome, and influence disease progression.9 In responders to induction, we demonstrated an increase in activated NK cells and commensurate decrease in inhibited NK cells from baseline to C1D1 of maintenance, similar to that reported by Sehgal et al.3 The lack of difference in NK populations observed between maintenance arms may be explained by a shorter duration of POM exposure in the POM arm despite the planned withdrawal of DEX. Whilst we observed dynamic changes in Treg according to maintenance arm, the exact role of Treg in MM is yet to be determined. Muthu et al.10 have reported elevated levels of functionally active Treg in MM patients which are associated with adverse clinical features and a higher risk of progression, however there remains conflicting data11,12 regarding their role in the pathogenesis of MM and their alterations in response to therapy with IMiD, potentially due to location (PB vs. tumor), concomitant DEX, patient selection and the Treg definition used.13 Treg modulation is likely an important component of the immunomodulatory mechanisms of IMiD. Functional studies would be important to further explore our observations. We demonstrated a relative enrichment of several activated neutrophil populations in responders at all mainte324

nance time points compared to baseline. Peripheral neutrophil expansion and activation has been demonstrated in a vast array of cancers. It is thought to be driven by tumor factors that modulate bone marrow hemopoietic processes to drive neutrophil and granulocyte expansion.14 In MM, it has been shown that neutrophils potentially function in an immunosuppressive manner via arginase-1, and therefore could contribute to both disease progression and sepsis.15 Our findings provide the baseline for future studies to identify predictive markers to allow identification of patients more likely to benefit from withdrawal of DEX. Novel observations of neutrophil populations may also provide new insights into the mechanisms of action of POM in MM. Anna Kalff,1,2,3 Tiffany Khong,1,2 Malarmathy Ramachandran,1,2 P Joy Ho,4 Peter Mollee,5 James D’Rozario,6 Kerry Taylor,7 Jane Estell,8 Sam Norton,9 10 Roslyn Kemp,10 Andrew J. Mitchell,11 John Reynolds,12 Nola Kennedy,1 Hang Quach13 and Andrew Spencer1,2,3 1 Malignant Hematology and Stem Cell Transplantation, Alfred Hospital, Melbourne, Victoria, Australia; 2Myeloma Research Group, Australian Center for Blood Diseases, Alfred Hospital Monash University, Melbourne, Victoria, Australia; 3Department of Clinical Hematology, Monash University, Clayton, Victoria, Australia; 4Royal Prince Alfred Hospital, Sydney, New South Wales, Australia; 5Princess Alexandra Hospital and University of Queensland, Brisbane, Queensland, Australia; 6The Canberra Hospital, Canberra, New haematologica | 2022; 107(1)

Letters to the Editor

South Wales, Australia; 7Icon Cancer Center, Brisbane, Queensland, Australia; 8Concord Repatriation General Hospital, University of Sydney, Sydney, New South Wales, Australia; 9Nanix Ltd., Dunedin, New Zealand; 10Department of Microbiology and Immunology, University of Otago, Dunedin, New Zealand; 11Materials Characterization and Fabrication Platform, Department of Chemical Engineering, University of Melbourne, Melbourne, Victoria, Australia; 12 Department of Epidemiology and Preventive Medicine, Alfred Health – Monash University, Melbourne, Victoria, Australia and 13Faculty of Medicine, University of Melbourne, St Vincent’s Hospital Melbourne, Melbourne, Vitoria, Australia (on behalf of The Australasian Leukemia and Lymphoma Group [ALLG]) Correspondence: ANDREW SPENCER - Andrew.spencer@monash.edu doi:10.3324/haematol.2021.278655 Received: February 24, 2021. Accepted: September 21, 2021. Pre-published: September 30, 2021. Disclosures: AK received honoraria from Amgen, Celgene/BMS, Pfizer, Janssen and Roche. CSL received honoraria from Sandoz. AS consults for Takeda, Secura Bio, Servier, Clegene, Amgen, Abbvie, Specialized Therapeutics Australia; received honoraria from Servier, Celgene, Amgen, Abbvie, Specialized Therapeutics Australia, is part of the Speakers Bureau of Takeka, Jansen and Celgene; received research funding from Celgene and Amgen. HQ consults for Amgen, Celgene, Sanofi Genzyme, and Jansen; received research funding from Amgen, Celgene and Sanofi Genzyme; is part of the scientific steering committee of Amgen, Karyopharm and GSK. JDR sits on the advisory board of Celgene and Abbvie; received honoraria from Celgene and Abbvie. JE sits on the advisory board of Janssen and Celgene. JR received honoraria from Novartis Australia, is employed by Alfred Health; acts as a biostatistician for trials funded by the Australian government and Abbvie, Amgen, Celgene, GSK, Janssen-Cilag, Merck, Novartis, Takeda, but sponsored by Alfred Health.; consults for the Australasian Leukemia & Lymphoma Group (ALLG); owns equity of Novartis AG. PM sits on the advisory board of Janssen, BMS/Celgene, Amgen, Takeka, Pfizer and Caelum (no personal fees received from any of them); received research funding from Janssen. All other authors have no conflicts of interest to disclose. Contributions: AS conceived the study; AK and AS designed the work that led to the submission; AK, PJH, PM, JDR, KT, JE, HQ and NK were involved in the conduct of the study; AK, TK, MR, AM performed experiments/acquired data; AK, AS, JR, SN, RK analyzed/interpreted the data; AK wrote the manuscript; AS, AK and SN drafted the manuscript. All authors reviewed and provided revisions for the manuscript, approved the final version and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

haematologica | 2022; 107(1)

Acknowledgements: We would like to thank all the patients and their families for participating in the study and our colleagues at participating hospitals who contributed to the study. Funding: this work was supported by grants from Celgene Corporation and The Merrin Foundation.

References 1. Pratt G, Goodyear O, Moss P. Immunodeficiency and immunotherapy in multiple myeloma. Br J Haematol. 2007;138(5):563-579. 2. Lacy MQ, McCurdy AR. Pomalidomide. Blood. 2013;122(14):23052309. 3. Sehgal K, Das R, Zhang L, et al. Clinical and pharmacodynamic analysis of pomalidomide dosing strategies in myeloma: impact of immune activation and cereblon targets. Blood. 2015;125(26):40424051. 4. Gandhi AK, Kang J, Capone L, et al. Dexamethasone synergizes with lenalidomide to inhibit multiple myeloma tumor growth, but reduces lenalidomide-induced immunomodulation of T and NK cell function. Curr Cancer Drug Targets. 2010;10(2):155-167. 5. Zelle-Rieser C, Thangavadivel S, Biedermann R, et al. T cells in multiple myeloma display features of exhaustion and senescence at the tumor site. J Hematol Oncol. 2016;9(1):116. 6. de Vries NL, van Unen V, Ijsselsteijn ME, et al. High-dimensional cytometric analysis of colorectal cancer reveals novel mediators of antitumour immunity. Gut. 2020;69(4):691-703. 7. Samusik N, Good Z, Spitzer MH, Davis KL, Nolan GP. Automated mapping of phenotype space with single-cell data. Nat Methods. 2016;13(6):493-496. 8. Norton SE, Leman JKH, Khong T, et al. Brick plots: an intuitive platform for visualizing multiparametric immunophenotyped cell clusters. BMC Bioinformatics. 2020;21(1):145. 9. Frohn C, Hoppner M, Schlenke P, Kirchner H, Koritke P, Luhm J. Anti-myeloma activity of natural killer lymphocytes. Br J Hameatol. 2002;119(3):660-664. 10. Muthu Raja KR, Rihova L, Zahradova L, Klincova M, Penka M, Hajek R. Increased T regulatory cells are associated with adverse clinical features and predict progression in multiple myeloma. PLoS One. 2012;7(10):e47077. 11. Hadjiaggelidou C, Mandala E, Terpos E, et al. Evaluation of regulatory T cells (Tregs) alterations in patients with multiple myeloma treated with bortezomib or lenalidomide plus dexamethasone: correlations with treatment outcome. Ann Hematol. 2019;98(6):14571466. 12. Quach H, Ritchie D, Neeson P, et al. Regulatory T cells (Treg) are depressed in patients with relapsed/refractory multiple myeloma (MM) and increases towards normal range in responding patients treated with lenalidomide (LEN). Blood. 2008;112(11):1696-1696. 13. Joshua D, Suen H, Brown R, et al. The T cell in myeloma. Clin Lymphoma Myeloma Leuk. 2016;16(10):537-542. 14. Singel KL, Segal BH. Neutrophils in the tumor microenvironment: trying to heal the wound that cannot heal. Immunol Rev. 2016;273(1):329-343. 15. Romano A, Parrinello NL, Simeon V, et al. High-density neutrophils in MGUS and multiple myeloma are dysfunctional and immunesuppressive due to increased STAT3 downstream signaling. Sci Rep. 2020;10(1):1983.

325

Letters to the Editor

Signal transduction pathway involved in platelet activation in immune thrombotic thrombocytopenia after COVID-19 vaccination The severe acute respiratory syndrome-coronarovirus2 (SARS-CoV-2) infection is currently advancing and is exponentially developing the COVID-19 pandemic,1 especially in the USA, Europe, South America, Russia, and India, recording more than 3,000,000 deaths. The outbreak of diseases has significantly impacted the lives of millions of people and within a year, several vaccines have been developed to control the pandemia. The European Medicines Agency (EMA), on the basis of randomized, blinded, controlled trials, validated four different vaccines, among them ChAdOx1 nCov-19 (AstraZeneca), a recombinant chimpanzee adenoviral vector encoding the spike glycoprotein of SARS-CoV-2. The condition of great world emergency has imposed very tight times for the experimentation and controlled trials of these vaccines. Since many people are vaccinated and follow-up is extensive, it might even be possible that new vaccine-related adverse events will arise. Recently, four studies published by New England Journal of Medicine2-5 have described a syndrome characterized by thrombosis and thrombocytopenia that came up 5 to 24 days after initial vaccination with ChAdOx1 nCoV-19 (AstraZeneca). Patients described in the studies were healthy or in medically-stable condition without any previous history of thromboses. Notably, a high percentage of the patients had thrombosis at unusual sites in particular at cerebral venous sinus with a median platelet count at diagnosis of approximately 20,000 to 30,000 per cubic millimeter.4-6 In addition to the signs and symptoms of the syndrome and the post-vaccination concomitance, the interesting fact that correlates the individual cases with each other is the high level of antibodies to platelet factor 4 (PF4)polyanion complexes. These are platelet-activating antibodies which clinically mimic autoimmune heparininduced thrombocytopenia (HIT). Indeed, in some heparin-treated patients, the drug combines with a protein produced by platelets, PF4, to form a complex. The binding of the anti-heparin/PF4 antibody to the heparinPF4 complex activates the platelets, leading to their aggregation and thrombocytopenia.7 However, unlike the usual situation in HIT, these vaccinated patients did not receive any heparin to explain the later onset of thrombosis and thrombocytopenia. So far there are very few hypotheses on the pathophysiological mechanisms of post-vaccination PF4-polyanion antibodies.5 Some studies described the occurrence of antiphospholipid antibodies (aPL) in some patients vaccinated with ChAdOx1 nCoV-19,3,6 suggesting that a wider spectrum of antibodies may play a role in the pathogenesis of vaccine-induced immune thrombotic thrombocytopenia (VITT). Some of the authors have described two cases of malignant middle cerebral artery (MCA) infarction with a concomitant thrombocytopenia within 10 days after vaccination with ChAdOx1 nCoV-19.8 Thus, in the present study we analyze in these two patients antibody specificity and the signaling transduction pathway involved in platelet activation in detail. At the laboratory of the Autoimmunity Unit of the Umberto I Polyclinic of Rome (UTN Unit), we processed sera from two patients (female, age 57 and 55 years, respectively) with VITT, admitted to the UTN Unit, and 326

a serum from a healthy donor vaccinated with ChAdOx1 nCoV-19 (non-hospitalized female subject, age 52 years). We obtained immunoglobulin G (IgG) fractions by using protein G-Sepharose columns. Patient 1 revealed right middle cerebral artery occlusion and severe thrombocytopenia (44,000/mm3); patient 2 showed occlusion of the right internal carotid artery terminus and of the left middle cerebral artery, with mild thrombocytopenia (133,000/mm3). Both patients had extensive lung and portal vein thrombosis. The study was conducted in compliance with the Declaration of Helsinki and the local Ethical Committee approved this study (clinicaltrials gov. Identifier: NCT04844632). For blood samples, healthy donors and relatives of both patients gave written informed consent. IgG fractions were used to stimulate platelets isolated from healthy donors (females, age range, 36-40 years), who gave written informed consent. Blood samples, in the presence of sodium citrate as anticoagulant, were centrifuged at 150 g for 15 minutes (min) at 20°C to obtain platelet-rich plasma (PRP). Two-thirds of the PRP were drawn, without disturbing the buffy coat layer, in order to prevent contamination. PRP was then mixed with ACD to avoid platelet activation, and centrifuged at 900 g for 10 min at 20°C. Platelet-poor plasma (PPP) was discarded and platelet pellets were resuspended with calcium-free Tyrode’s buffer, containing 10% (v:v) ACD and washed as above. Then, platelets were resuspended in calcium-free Tyrode’s buffer with the addition of bovine serum albumin (BSA, 3 mg/mL). The purity of the isolated platelets was verified by staining with a fluorescein isothiocyanate (FITC)-conjugated anti-CD61 monoclonal antibody (mAb) (Beckman Coulter, Hialeah, FL, USA) and analyzed by flow cytometry (CytoFLEX, Beckman Coulter), as shown in the Online Supplementary Figure S1A. Since the p38 mitogen-activated protein kinase (MAPK) pathway is an important intracellular signaling pathway in platelets which can be activated by various stimuli and may be an integral component of arterial and venous thrombosis,9 we analyzed by western blot the effect of IgG fractions from these patients on ERK and p38 phosphorylation in platelet lysates. For this purpose, human platelets, untreated or treated with healthy donor IgG or patient IgG fractions, for 10 min at 37°C, were resuspended in lysis buffer containing 20 mM HEPES, pH 7.2, 1% Nonidet P-40, 10% glycerol, 50 mM NaF and 1 mM Na3VO4, including protease inhibitors. Then, whole extracts proteins (40 µg/sample) were separated in 10% SDS-PAGE. Proteins were electrophoretically transferred to PVDF membranes (Bio-Rad Laboratories, Richmond, CA, USA) and then, after blocking with Tris-buffered saline Tween 20 (TBS-T) 3% BSA, incubated with polyclonal rabbit anti-phospho-ERK1/2 (Cell Signaling, Inc., Danvers, MA, USA), polyclonal rabbit anti-phospho-p38 antibodies (Cell Signaling, Inc.) Antibody reactions were visualized by horseradish peroxidase (HRP)-conjugated anti-rabbit IgG (Sigma-Aldrich, Milan, Italy), and then by the chemiluminescence reaction, using ehnanced chemiluminescence western blot system (Amersham Pharmacia Biotech, Buckinghamshire, UK). In parallel experiments, human platelets were treated with healthy donor IgG or patient IgG fractions for 4 hours at 37°C, pretreated with ERK inhibitor PD98059 (10 µM, 3 min at 37°C). Then, platelet samples were lysed and analyzed in western blot as above using rabbit anti-tissue factor (TF) mAb (ab228968,Abcam, Cambridge, UK). Data obtained are expressed as means ± standard deviation (SD) of at least haematologica | 2022; 107(1)

Letters to the Editor

Figure 1. Immunoglobulin G fractions from patients induce ERK and p38 phosphorylation with tissue factor expression. Human platelets from healthy donors were treated with immunoglobulin G (IgG) fractions from the 2 patients and from a healthy donor vaccinated with ChAdOx1 nCoV-19 (10 minutes for ERK and p38 activation and 4 hours for tissue factor [TF] analysis). Protein extracts were separated by SDSPAGE and analyzed by western blot to investigate phosphorylated ERK (A), p38 MAPK (B), and TF (C), using anti-phospho-ERK1/2, anti-phospho-p38 and anti-TF antibodies, respectively. Data are reported as mean ± standard deviation from 3 independent experiments. Statistical analysis indicated: **P<0.001 vs. untreated; §§P<0.001 vs. healthy donor IgG fraction; § P=0.01 vs. healthy donor IgG fraction; °° P<0.001 vs. pretreated with PD98059.

haematologica | 2022; 107(1)

327

Letters to the Editor

Table 1. Serological characteristics of the patients aPF4 aCL IgG (GPL/mL) aCL IgM (MPL/mL) a 2-GPI IgG (UA/mL) a 2-GPI IgM (UA/mL) LA aCL (TLC Immunostaining) aVim/CL

Healthy donor

Patient 1

Patient 2

Negative 0 0 0 0 0.95

Positive (1,680 AU) 10.4 0 2.9 3.2 1.8

Positive (1,290 AU) 0 0 0 0.7 1.4

Negative Negative

Positive Positive

aPF4: anti-platelet factor 4 antibodies; aCL: anti-cardiolipin antibodies; a 2-GPI: anti- 2 glycoprotein I antibodies; LA: Lupus anticoagulant; TLC: thin-layer chromatography; aVim/CL: anti-Vimentin/Cardiolipin complex antibodies.

three independent experiments. Statistical analysis was performed by the paired Student’s t-test. Statistical significance was set up at P≤0.01. Results show that the treatment with both IgG fractions from patients with VITT, induced a significant increase of both phospho-ERK (Figure 1A) and phosphop38 expression (Figure 1B) compared to untreated platelets or treated with healthy donor IgG obtained from a representative healthy donor vaccinated with ChAdOx1 nCoV-19. As a consequence of ERK activation, we also observed a significant increase of TF levels in treated platelets. Moreover, pretreatment with ERK inhibitor PD98059 partially prevented TF expression in samples stimulated with patient IgG fractions (Figure 1C), indicating a functional link between ERK activation and TF expression. Surface TF expression was also verified by flow cytometry analysis, which revealed an increase of TF in platelets treated with IgG fractions from the two patients as compared to healthy donor IgG fraction (Online Supplementary Figure S1B). Sera of patients and healthy donor were tested for antibodies to platelet factor 4 (PF4)/polyanion by using a commercial enzyme immunoassay (Immucor, Lifecodes, Waukesha, WI). Sera with an optical density >500 arbitrary units (AU) were considered as positive.3 In addition, anticardiolipin (aCL, IgG, IgM) and anti-b2-glycoprotein I (b2-GPI, IgG, IgM) were detected by chemiluminescence assay, using Zenit RA Immunoanalyzer (A. Menarini Diagnostics, Florence, Italy). Finally, aCL were also detected by TLC immunostaining and anti-vimentin/cardiolipin (aVim/CL) antibodies by ezyme-linked immunosorbant assay, as previously described;10 lupus anticoagulant (LA) was analyzed by two coagulation systems, a dilute sensitized activated partial thromboplastin time (aPTT) and a dilute Russell’s viper venom time (dRVVT), also performing a confirm test (Hemoliance Instrumentation Laboratory, Lexington, MA, USA). Serological characteristics of the two patients and healthy donor are summarized in Table 1. Previous studies described that a pathogenic PF4dependent syndrome, unrelated to the use of heparin therapy, can occur after the administration of the ChAdOx1 nCoV-19 vaccine.2,3,4-6 All subjects displayed anti-PF4 antibodies which were able to activate platelets obtained from healthy donors.2 A subset of these antibodies may activate platelets after binding to PF4/heparin complexes, causing the prothrombotic adverse drug reaction HIT. In autoimmune-HIT, anti-PF4/P-antibodies activate platelets in the absence of heparin. In the present study we observed for the first time that IgG fractions of 328

these patients are able to trigger a signal transduction pathway involving ERK and p38 MAPK, which may lead to TF expression increase with a consequent amplification of platelet activation and coagulation cascade.11 We cannot exclude the possibility that other additional signal transduction pathways may be involved. However, other autoantibody specificities have been described in sera of patients affected by VITT, including aPL and/or LA.2,4 This finding may be relevant, since antib2GPI/b2GPI complexes were shown to induce signal transduction pathway(s) leading to platelet activation,12 particularly promoting thrombosis via p38 MAPK13 and consequent involvement of TF as major initiator of the clotting cascade. We confirmed and extended the analysis of aPL, also showing the presence of “unconventional” (“non-criteria”) aPL in serum of our patients.14 How the pathogenic mechanism may be consequent to ChAdOx1 nCov-19 vaccination remains unclear;15 however, a possible pathogenic role of the adenoviral viral vector cannot be excluded. Roberta Misasi,1 Antonella Capozzi,1 Gloria Riitano,1 Serena Recalchi,1 Valeria Manganelli,1 Vincenzo Mattei,2 Agostina Longo,1 Manuela De Michele,3 Tina Garofalo,1 Fabio M. Pulcinelli1 and Maurizio Sorice1 1 Department of Experimental Medicine, "Sapienza" University of Rome, Rome; 2Biomedicine and Advanced Technologies Rieti Center, Sabina Universitas, Rieti; 3Emergency Department, "Sapienza" University of Rome, Rome, Italy Correspondence: MAURIZIO SORICE- maurizio.sorice@uniroma1.it doi:10.3324/haematol.2021.279729 Received: August 3, 2021. Accepted: September 29, 2021. Pre-published: October 7, 2021. Disclosures: no conflicts of interest to disclose. Contributions:FMP and AC designed and performed the research; MDM selected the patients; GR, SR and VM performed experiments; AL and VM provided and analyzed the data; RM and TG wrote the paper; MS designed, organized, and supervised the research and edited the paper. All authors read, edited, participated in the revision, and approved the manuscript.

References 1. Hu B, Guo H, Zhou P, Shi ZL. Characteristics of SARS-CoV-2 and COVID-19. Nat Rev Microbiol. 2021;19(3):141-154. 2. Scully M, Singh D, Lown R, et al. Pathologic antibodies to platelet

haematologica | 2022; 107(1)

Letters to the Editor

factor 4 after ChAdOx1 nCoV-19 vaccination. N Engl J Med. 2021;384(23):2202-2211. 3. Greinacher A, Thiele T, Warkentin TE, et al. Thrombotic thrombocytopenia after ChAdOx1 nCov-19 vaccination. N Engl J Med. 2021;384(22):2092-2101. 4. Schultz NH, Sørvoll IH, Michelsen AE, et al. Thrombosis and thrombocytopenia after ChAdOx1 nCoV-19 vaccination. N Engl J Med. 2021;384(22):2124-2130. 5. Cines DB, Bussel JB. SARS-CoV-2 Vaccine-induced immune thrombotic thrombocytopenia. N Engl J Med. 2021;384(23):22542256. 6. Bayas A, Menacher M, Christ M, et al. Bilateral superior ophthalmic vein thrombosis, ischaemic stroke, and immune thrombocytopenia after ChAdOx1 nCoV-19 vaccination. Lancet. 2021;397(10285):e11. 7. Ahmed I, Majeed A, Powell R. Heparin induced thrombocytopenia: diagnosis and management update. Postgrad Med J. 2007;83(983):575-582. 8. De Michele M, Iacobucci M, Nicolini E, et al. Malignant cerebral infarction, systemic venous thrombosis and thrombocytopenia after ChAdOx1 nCov vaccination: a possible catastrophic variant of vaccine induced thrombotic thrombocytopenia. Nat Commun. 2021;12(1):4663. 9. Adam F, Kauskot A, Rosa JP, et al. Mitogen-activated protein kinas-

haematologica | 2022; 107(1)

es in hemostasis and thrombosis. J Thromb Haemost. 2008;6(12): 2007-2016. 10. Truglia S, Mancuso S, Capozzi A, et al. "Non-criteria antiphospholipid antibodies": bridging the gap between seropositive and seronegative Antiphospholipid Syndrome. Rheumatology. 2021 May 10. [Epub ahead of print] 11. Li Z, Zhang G, Feil R, et al. Sequential activation of p38 and ERK pathways by cGMP-dependent protein kinase leading to activation of the platelet integrin alphaIIb beta3. Blood. 2006;107(3):965972. 12. Capozzi A, Manganelli V, Riitano G, et al. Tissue factor overexpression in platelets of patients with anti-phospholipid syndrome: induction role of anti-b2-GPI antibodies. Clin Exp Immunol. 2019;196(1):59-66. 13. Zhang W, Zha C, Lu X, et al. Anti-b2GPI/b2GPI complexes induce platelet activation and promote thrombosis via p38MAPK: a pathway to targeted therapies. Front Med. 2019;13(6):680-689. 14. Zohoury N, Bertolaccini ML, Rodriguez-Garcia JL, et al. Closing the serological gap in the antiphospholipid syndrome: the value of "non-criteria" antiphospholipid antibodies. J Rheumatol. 2017;44(11):1597-1602. 15. Ledford H. How could a COVID vaccine cause blood clots? Scientists race to investigate. Nature. 2021;592(7854):334-335.

329

Letters to the Editor

First report of inherited protein S deficiency caused by paternal PROS1 mosaicism Inherited protein S (PS) deficiency is a thrombophilic disorder with an autosomal dominant mode of inheritance.1 In plasma, 60% of PS exists in a complex with C4b binding protein and the remaining 40% in free form; only free PS functions as a cofactor for the activated protein C. Inherited PS deficiency is classified into types I, II, and III according to the level of PS activity or free and total PS antigen.2 Types I and III have different phenotypes but they are considered to be caused by the same genetic abnormality. According to the Human Gene Mutation Database Pro (https://my. qiagendigitalinsights. com/bbp/view/hgmd/pro/all/php), 453 variants localized in the protein S gene (PROS1 gene) have been reported worldwide as of January 2021. Among these, missense/nonsense variants account for 61% (277 variants). However, genetic variants have been reported as undetectable in approximately half of the families with PS deficiency.3 Based on the Suita study, which examined the prevalence of PS deficiency in the Japanese general population, the prevalence is 1.12%;4 about 5-10 times higher than that in Europe and America (0.16–0.21%).5 PS deficiency is the most common inherited thrombophilia in Japan.6,7 In this study, we identified a novel PROS1 variant and paternal mosaicism in a family with inherited PS deficiency in which both parents are asymptomatic. This is the first report of inherited thrombophilia due to parental mosaicism and highlights the importance

of including information about potential mosaicism in genetic counseling. The proband was a 34-year-old Japanese woman who was hospitalized with histiocytic necrotizing lymphadenitis. On admission to the hospital, total PS antigen (NSauto total protein S; NASCA Co., Ltd., Yokohama, Japan), free PS antigen (NSauto free protein S; NASCA Co., Ltd.), and PS activity levels (HemosIL PS-clot; I.L.Japan Co.,Ltd., Tokyo, Japan) were 55%, 25%, and <10%, respectively (Figure 1A). In the proband, acute inflammation caused by histiocytic necrotizing lymphadenitis may have increased vascular permeability, resulting in a decrease in free PS and a significant decrease in PS activity. The D-dimer concentration was significantly increased to 25.3 mg/mL (reference range, <1 mg/mL), but contrast-enhanced computed tomography showed no obvious thrombosis. In addition, the proband’s sister had already been diagnosed with PS deficiency after developing deep vein thrombosis in her leg at 25 years old and is on warfarin therapy (Figure 1A). Based on the above, the proband was diagnosed with type I PS deficiency. Both parents have normal levels of PS activity and antigen (Figure 1A), and the father had suffered a stroke at 62 years old and was taking edoxaban but the mother had no history of thrombosis. This study was approved by the Ethics Committee for Human Genome and Gene Analysis Research at Kanazawa University (approval no. 450-5) and by the National Cerebral and Cardiovascular Center (approval no. M14-026). After obtaining informed consent from the proband and her family members, genomic DNA was

Figure 1. Protein S antigen and activity levels of a protein S-deficient family and the results of Sanger sequencing. (A) Levels of total protein S (PS) antigen, free PS antigen, and PS activity in PS-deficient family members. The type of PS deficiency of the proband and sister were shown. (B) Pedigree and sequencing results. Arrow indicates the proband, and filled symbols indicate affected patients. The proband and her sister were heterozygous for c.246T>G (p.Tyr82*) whereas the parents seemed to have no causative variants. Data from the father showed a very small peak of G (black) with a normal peak of T (red). Abbreviations: RR: reference range; NA: not applicable; *, warfarin therapy.

330

haematologica | 2022; 107(1)

Letters to the Editor

Figure 2. Quantification of the variant allele ratio of blood-derived DNA using digital PCR. (A) Sequences of the primers and probes. (B) Blue indicates FAM fluorescence for the target variant allele; red indicates VIC fluorescence for the normal allele; green indicates FAM+VIC; and yellow indicates no amplification.

extracted from peripheral blood leukocytes using a GENERATION Capture Column kit (QIAGEN, Tokyo, Japan). All exons and the exon-intron boundaries of PROS1 were polymerase chain reaction (PCR)-amplified using GoTaq G2 Hot Start Master Mixes (Promega, Tokyo, Japan). PCR products were sequenced using a BigDye Terminator v3.1 Cycle Sequencing kit (Thermo Fisher Scientific, Tokyo, Japan) and an Applied Biosystems 3500xL Genetic Analyzer (Thermo Fisher Scientific). Sanger sequencing indicated a novel heterozygous nonsense variant in exon 3 (NM_000313.4: c.246T>G; p.Tyr82*) in the proband and her sister. However, neither parent showed this variant by Sanger sequencing (Figure 1B). Since the probability of the same de novo variant occurring in the proband and her sister was considered virtually zero, this case was attributed to another cause such as the parents’ genetic mosaicism. We therefore attempted to quantify the variant allele ratio. First, digital PCR (dPCR) was performed using QuantStudio 3D Digital PCR Master Mix v2 (Thermo Fisher Scientific) and the QuantStudio 3D Digital PCR system (Thermo Fisher Scientific) (Figure 2). The variant allele ratio was 48.5% for the proband, 52.9% for the sister, 15.2% for the haematologica | 2022; 107(1)

father, and 0.1% for the mother (Figure 2B). Second, we quantified the variant allele ratios using pyrosequencing with PyroMark Q24 (QIAGEN) to confirm the reproducibility of paternal mosaicism using a different principle of measurement (Figure 3). The results of pyrosequencing indicated that the variant allele ratios were 50% for the proband, 49% for the sister, 15% for the father, and 1% for the mother (Figure 3B). The variant allele ratios were consistent between dPCR and pyrosequencing for all family members, suggesting the presence of somatic mosaicism in the father. Based on this result, a careful review of the Sanger sequencing data of the father revealed the presence of a very small G peak with a normal T peak (Figure 1B). Third, we evaluated the ratio of variant alleles in saliva-derived DNA to confirm the reproducibility of the somatic mosaicism presented in the blood-derived DNA of the father. The saliva-derived DNA was extracted and purified using Oragene Purifier (Kyodo International). The results of pyrosequencing showed that the variant allele ratios of the saliva-derived DNA were 50% for the proband, 50% for the sister, 17% for the father, and 2% for the mother (Figure 3C), confirming the reproducibility 331

Letters to the Editor

of the father’s mosaicism. Generally, the detection limits of dPCR and pyrosequencing are said to be more sensitive than that of Sanger sequencing: Sanger sequencing 15-20%;8,9 pyrosequencing 5%;8,10 dPCR 0.1-1%.11,12 We performed pyrosequencing of five control subjects and an average variant allele ratio was 1% (data not shown). In addition, the detection limit of our pyrosequencing assay system calculated at 3d was 2% (data not shown). However, we could not examine the control subjects with our dPCR assay system. Based on the above, the variant allele ratios detected in the mother, 0.1% for dPCR and 1-2% for pyrosequencing, were below the

detection limits of their respective assay system and were, therefore, considered to be background. In this case, the paternal mosaicism was observed in the blood cells (mesoderm origin) and salivary gland cells (ectoderm origin). This may suggest that the genetic variant might have occurred early in the fetal development process. However, the variant allele ratio of the two types of DNA collected in this study happened to be about 15%-17%, but the variant allele ratios in other organs have not yet been verified; this makes it difficult to determine when the variant occurred from the variant allele ratios. It is also assumed that the father's germ cells

Figure 3. Quantification of variant allele ratios in blood- and saliva-derived DNA using pyrosequencing. (A) Sequences of the primers. The results of variant allele ratios quantified by pyrosequencing of blood-derived DNA (B) and saliva-derived DNA (C) from the proband and family members. The gray area shows the allele ratio of A and C at the variant locus. Since pyrosequencing was performed using reverse primers, the allele ratio of C represents the variant allele ratio.

332

haematologica | 2022; 107(1)

Letters to the Editor

carry the variant allele, but the semen could not be collected and analyzed due to lack of consent. For a definitive diagnosis of gonadal mosaicism, it is necessary to prove the presence of mosaicism in the father's semen, but in this case, since the two children are heterozygous for the same variant, it is probable that the father has somatogonadal mosaicism.13 In recent years, some cases of various diseases that were originally considered to be caused by de novo variants have actually been attributed to parental somatic or gonadal mosaicism.13-15 Since the sensitivity of variation detection in Sanger sequencing, considered the standard for evaluating PROS1 variants, is limited to approximately 15-20%,8,9 heterozygous variants can be detected but mosaicisms are likely to be undetectable. Thus, another more sensitive method, such as dPCR or pyrosequencing, is needed for the genetic evaluation of parental mosaicism. In this study, we detected a mosaic variant in the father of two heterozygous sisters with PS deficiency. This is the first case of parental mosaicism in an inherited PS-deficient family, suggesting that parental mosaicism may exist among some patients who have been thought to have de novo variants of inherited thrombophilia. It has been reported that approximately 20% of sporadic cases of hemophilia A, as determined by traditional Sanger sequencing methods, have an asymptomatic mosaic mother.15 The percentage of parental mosaicism present in sporadic cases of inherited PS deficiency needs to be determined by accumulating more cases in the future. In conclusion, in cases of autosomal-dominant inherited thrombophilia, when the causative variant is detected only in the proband and a de novo variant is suspected, it is necessary to investigate the presence of parental mosaicism. For this purpose, it is very informative to quantify the variant allele ratio using multiple readily accessible specimens, such as blood, saliva, hair follicles, and urine to confirm the parental mosaicism. Furthermore, semen analysis should be performed as genetic screening for germline mosaicism in fathers of patients with de novo variants, and this analysis could be useful for prenatal counseling. In sporadic cases of inherited diseases, even if neither parent has the same variant using traditional sequencing methods, confirming the mosaicism of the parents is very important regarding the recurrence risk for the patient’s siblings and is essential information for genetic counseling. Satomi Nagaya,1,* Keiko Maruyama,2,* Atsushi Watanabe,3 Makiko Meguro-Horike,4 Yuta Imai,1 Yuki Hiroshima,5 Shin-Ichi Horike,4 Koichi Kokame2 and Eriko Morishita1,6 1 Department of Clinical Laboratory Science, Division of Health Sciences, Graduate School of Medical Science, Kanazawa University, Kanazawa, Ishikawa; 2Department of Molecular Pathogenesis, National Cerebral and Cardiovascular Center, Suita, Osaka; 3 Division of Clinical Genetics, Kanazawa University Hospital, Kanazawa, Ishikawa; 4Advanced Science Research Center, Kanazawa University, Kanazawa, Ishikawa; 5Department of Hematology, Nagano Red Cross Hospital, Nagano, Nagano and 6Department of Hematology, Kanazawa University Hospital, Kanazawa, Ishikawa, Japan

haematologica | 2022; 107(1)

*SN and KM contributed equally as co-first authors. Correspondence: ERIKO MORISHITA - eriko86@staff.kanazawa-u.ac.jp doi:10.3324/haematol.2021.278527 Received: June 23, 2021. Accepted: October 1, 2021. Pre-published: October 14, 2021. Disclosures: no conflicts of interest to disclose. Contributions: SN: investigation, visualization and writing, original draft; KM: investigation, visualization and writing, original draft; SN and KM are co-first authors. AW: methodology and supervision; MH: investigation; YI: investigation; YH: resources; SH: methodology, investigation, formal analysis; KK: investigation, formal analysis, funding acquisition, supervision; EM: conceptualization, funding acquisition, supervision, writing-review and editing. Funding: this work was partially supported by Grants-in-Aid from the Ministry of Health, Labour and Welfare of Japan to EM (grant number 20FC0201) and KK (grant number 20FC1024).

References 1. Dahlbäck B. The protein C anticoagulant system: inherited defects as

basis for venous thrombosis. Thromb Res. 1995;77(1):1-43. 2. Mannucci PM, Valsecchi C, Krachmalnicoff A, Faioni EM, Tripodi A. Familial dysfunction of protein S. Thromb Haemost. 1989;62(2):763766. 3. Johansson AM, Hillarp A, Säll T, et al. Large deletions of the PROS1 gene in a large fraction of mutation-negative patients with protein S deficiency. Thromb Haemost. 2005;94(5):951-957. 4. Sakata T, Okamoto A, Mannami T, Tomoike H, Miyata T. Prevalence of protein S deficiency in the Japanese general population: the Suita Study. J Thromb Haemost. 2004;2(6):1012-1013. 5. Beauchamp NJ, Dykes AC, Parikh N, Campbell Tait R, Daly ME. The prevalence of, and molecular defects underlying, inherited protein S deficiency in the general population. Br J Haematol. 2004; 125(5):647-654. 6. Miyata T, Kimura R, Kokubo Y, Sakata T. Genetic risk factors for deep vein thrombosis among Japanese: importance of protein S K196E mutation. Int J Hematol. 2006;83(3):217-223. 7. Kinoshita S, Iida H, Inoue S, et al. Protein S and protein C gene mutations in Japanese deep vein thrombosis patients. Clin Biochem. 2005; 38(10):908-915. 8. Harrington CT, Lin EI, Olson MT, Eshleman JR. Fundamentals of pyrosequencing. Arch Pathol Lab Med. 2013;137(9):1296-1303. 9. Cheng LY, Haydu LE, Song P, et al. High sensitivity sanger sequencing detection of BRAF mutations in metastatic melanoma FFPE tissue specimens. Sci Rep. 2021;11(1):9043. 10. Tsiatis AC, Norris-Kirby A, Rich RG, et al. Comparison of Sanger sequencing, pyrosequencing, and melting curve analysis for the detection of KRAS mutations: diagnostic and clinical implications. J Mol Diagn. 2010;12(4):425-432. 11. Carvalho TM, Dourado RM, Nakatani SM, et al. Digital PCR detection of EGFR somatic mutations in non-small-cell lung cancer formalin fixed paraffin embedded samples. Mol Cell Probes. 2021;58:101745. 12. Oshiro C, Kagara N, Naoi Y, et al. PIK3CA mutations in serum DNA are predictive of recurrence in primary breast cancer patients. Breast Cancer Res Treat. 2015;150(2):299-307. 13. Yamamoto K, Kubota T, Takeyari S, et al. Parental somatogonadal COL2A1 mosaicism contributes to intrafamilial recurrence in a family with type 2 collagenopathy. Am J Med Genet A. 2020;182(3):454460. 14. Chesneau B, Plancke A, Rolland G, et al. Parental mosaicism in Marfan and Ehlers-Danlos syndromes and related disorders. Eur J Hum Genet. 2021;29(5):771-779. 15. Lannoy N, Hermans C. Genetic mosaicism in haemophilia: A practical review to help evaluate the risk of transmitting the disease. Haemophilia. 2020;26(3):375-383.

333

Letters to the Editor

Treatment with ibrutinib does not induce a TP53 clonal evolution in chronic lymphocytic leukemia Ibrutinib is active both in treatment-naïve (TN) and relapsed/refractory (R/R) chronic lymphocytic leukemia (CLL) patients, including those with unmutated immunoglobulin heavy chain variable region (IGHV) genes and TP53 disruption.1-3 The acquisition of BTK or PLCg2 gene mutations conferring resistance, supports the existence of clonal evolution also under ibrutinib treatment.4-8 Whilst it is well described that subclonal TP53 mutations undergo a positive clonal selection following chemoimmunotherapy (CIT),9-11 being the main driver of treatment failure, recent studies have suggested that this might not be the case under ibrutinib.7,12-14 In order to investigate the dynamics of major and minor TP53 mutations under ibrutinib treatment, we performed longitudinal TP53 monitoring by deep-sequencing in CLL-treated patients. Two cohorts were included: 44 TN and 14 R/R patients. A total of 216 peripheral blood (PB) samples in TN and 52 in R/R were collected at baseline and at subsequent time points during therapy. Among TN patients, 28 were males and 16 females, with a median age of 72 years (range, 54-87); they received ibrutinib plus rituximab (GIMEMA trial LLC1114), with a median ibrutinib exposure of 2.7 years (range, 1.2-3.7) and were evaluated at 6-month intervals for a median number of 5 time points (range, 3-8). R/R patients, nine males and five females, with a median age of 71 years (range, 55-80), received ibrutinib single agent after a median of 1.5 (range, 1-4) CIT lines; they were evaluated at disease progression (PD) before each line of CIT and after ibrutinib treatment (median: 4 time points; range, 26), with 2.5 years ibrutinib exposure (range, 2.1-3.3). Amplicon libraries, covering the entire coding region and splice sites of TP53 gene (exon 2 to 11), were prepared according to the TruSeq Custom Amplicon Low Input Library Prep kit protocol dual strand (Illumina, San Diego, CA, USA) and paired-end sequenced on a Miseq Sequencer (Illumina). A mean coverage of 8,956x was obtained; across the target region a coverage >5,000x was obtained in >80% of the sequence in 70% of the samples. Bioinformatic analysis was performed by MiSeq Reporter (Illumina) and an in-house bioinformatics pipeline. A total concordance was observed for the variants with variant allele frequency (VAF) ≥3%; variants with 1%≤VAF<3%, only identified from in-house pipeline, were manually checked on alignment files resulting from MiSeq Reporter analysis of two DNA strands, using Integrative Genomics Viewer. Variants were manually curated according to the International Agency for Research on Cancer TP53 database (http://p53.iarc.fr/). Validate polymorphism, synonymous

variants and variants mapping >2 bp outside coding exons were filtered out. Mutations were defined major if VAF was >10% and minor if ≤10%; the latter were confirmed in an independent deep-sequencing run. VAF was corrected to cancer cell fraction (CCF) by the proportion of CD19+/CD5+ cells in each sample, assessed by flow cytometry. The limit of detection (LOD) was 1% (Online Supplementary Figure S1). A mutation was considered stable when the log2 fold-change (Log2FC) of CCF values before and after ibrutinib treatment was included between -0.5 and 0.5, decreasing or increasing if the Log2FC was <-0.5 or >0.5, respectively. Among the 44 TN patients at baseline (T0), 27 cases (61%) resulted TP53 wild-type (WT) and 17 (39%) mutated by deep-sequencing analysis. Nine of 43 (21%) carried the del(17p) and 30 (68%) showed unmutated IGHV. Twenty-three TP53 mutations (1.4 mutation/patient; range, 1-5) were identified: 17 (74%) were major (mean VAF 58.8%; range, 18-94.8) and six (26%) minor (mean VAF 5.3%; range, 2.1-9.2). Thirteen patients carried a sole major TP53 mutation; two cases (cases #10025, #10875) showed co-existing major and minor mutations; two cases (#9915, #10671) showed one minor mutation each and 27 none (Online Supplementary Table S1). According to the CCF, after 2.7 years from ibrutinib treatment, nine of 23 (39%) major mutations decreased, eight of 23 (35%) major mutations persisted stable, four of 23 (17%) (1 major and 3 minor) were undetectable and two of 23 (9%) minor mutations increased (Figure 1A and B; Table 1; Online Supplementary Table S1). Novel TP53 mutations emerged during treatment neither in TP53 mutated nor WT patients. Thus, TP53 mutated TN patients followed two main patterns: i) major TP53 mutations persisting major from T0 with a stable CCF (6 patients); ii) major TP53 mutations persisting major from T0 with a decreasing CCF (7 patients). In addition, in one patient (#10671) a minor TP53 mutation at T0 showed an increasing CCF, becoming major and in another patient a minor TP53 mutation resulted undetectable; in the two cases with a complex mutational architecture, TP53 mutations dynamics is shown in Figure 1C. No TP53 mutation was detected in 27 patients over time. Most patients are still on therapy, including those who showed an increase of TP53 mutation CCF (#10671, #10875) at month +20 (T20) who are still on therapy at T36 and T30, respectively. Two discontinued due to PD (#9795, #10239). Among TN patients, seven TP53 mutated and 11 WT with a measurable PB residual disease (CD19+/CD5+ >20%) at T20 (n=3), T26 (n=2), T32 (n=7), T38 (n=4) and T44 (n=2), were analyzed by next-generation sequencing for BTK and PLCg2 mutations and resulted WT for both

Table 1. Dynamics of TP53 mutations in treatment-naïve under ibrutinib and relapsed/refractory patients under ibrutinib and chemoimmunotherapy. TN patients R/R patients, ibrutinib phase R/R patients, CIT phase*

Undetectable

Decreasing

Stable

Increasing

Novel

4/23 (17.5%) 11/31 (35%) 0/29

9/23 (39%) 8/31 (26%) 3/29 (10%)

8/23 (35%) 8/31 (26%) 6/29 (21%)

2/23 (9%) 4/31 (13%) 7/29 (24%)

0 2 13/29 (45%)

*13 evaluable cases. TN: treatment-naïve; R/R: relapsed/refractory; CIT: chemoimmunotherapy.

334

haematologica | 2022; 107(1)

Letters to the Editor

Figure 1. Dynamics of TP53 mutations under ibrutinib therapy in treatment-naïve patients. (A) Each line corresponds to the cancer cell fraction (CCF) of each mutation. Dashed lines represent the median CCF. (B) Heatmap with the CCF log2-fold-change (Log2FC) for each mutation. Color code indicates the trend of TP53 mutation over time: decrease (<-0.5), increase (>1), stable (-0.5 and 1). (C) Dynamics of TP53 mutations in the 2 treatment-naïve cases with co-existing major and minor TP53 mutations at time zero (T0).

haematologica | 2022; 107(1)

335

Letters to the Editor

genes, including patients with PD (Online Supplementary Table S1). Among the 14 R/R patients, prior to ibrutinib treatment four patients (29%) resulted TP53 WT and ten (71%) mutated by deep-sequencing analysis. Four of 11 (37%) carried the del(17p) and seven (50%) showed unmutated IGHV. Thirty-one TP53 mutations (3.1 mutation/patient; range, 1-11) were identified: 11 (35.5%) were major (mean VAF 31.9%; range, 10.5-78.8) and 20 (64.5%) minor (mean VAF 2.9%; range 1-6.8). Five patients carried one or two major mutations (#8271, #5708, #3547, #9225, #7458), three cases showed a complex mutational architecture (#5717, #8353 and #3425); two patients showed only minor mutations (3 in #3546 and 11 in #8540, respectively); four patients had no mutations (Online Supplementary Table S2). As expected, before ibrutinib, the TP53 mutational load of R/R patients was higher and more complex than that of TN patients. In the R/R patients, eight of 31 (26%) TP53 mutations (4 major and 4 minor) decreased, eight of 31 (26%) (4 major and 4 minor) persisted stable, 11 of 31 (35%) (2 major and 9 minor) were undetectable, four of 31 (13%) (2 major in #7458 and #8353; 2 minor in #3546) increased in three patients, and two novel minor mutations emerged in two cases already TP53-mutated prior to ibrutinib treatment (cases #5717, #3425) (Table 1; Online Supplementary Table S2). No novel TP53 mutations arose in TP53 WT patients over time under ibrutinib. In R/R patients, because of the more complex mutational profile of each patient we could not identify patient-related patterns, rather we documented different dynamics of different TP53 mutations within the same patient (Online Supplementary Table S2), possibly due to a different sensitivity to the drug, as suggested.12,13 Among R/R patients, four TP53-mutated and two WT continued ibrutinib (#8540, #6856, #3547, #9225, #5717, #6123), four discontinued ibrutinib for adverse events (#5708, #8353, #7458, #3380), one shifted to venetoclax (#3425), three died (#3546, #8991, #8271). No significant changes were observed in the mean CCF of TP53 mutations before and after ibrutinib: 60.7% versus 43.1% and 20.5% versus 20.2%, in TN and R/R patients, respectively. On the contrary, the lymphocyte count decreased significantly after ibrutinib treatment in TP53-mutated patients from both cohorts: from 40.7x109/L (range: 4.9-132.2x109/L) to 11.2x109/L (range, 1.2-135.7x109/L) (P=0.018, Mann-Whitney test) and from 39.7x109/L (range, 1.5-99.0x109/L) to 7.1x109/L (range, 1.4-18.9x109/L) (P=0.034), respectively. The decrease in lymphocytosis in the presence of a stable TP53 mutations CCF proves the effectiveness of ibrutinib both on TP53-mutated and WT CLL cells, regardless of previous therapies, at least during the first years of treatment. In 13 of the 14 R/R patients, TP53 mutations were retrospectively evaluated by deep sequencing also before each line of CIT. At the first evaluated time point, three patients were mutated (2 with minor and 1 with one major mutation) and ten resulted WT; of the latter, six acquired major or minor TP53 mutations over time. Overall, among the nine mutated cases, 29 mutations were identified with the following dynamics: 13 (45%) (3 major and 10 minor) novel mutations emerged, seven (24%) minor mutations increased, six (21%) (3 major and 3 minor) persisted stable and three (10%) (1 major and 2 minor) decreased. The decrease in CCF for the latter was from 84.44% to 46%, from 3.3% to 2.01% and from 4.69% to 1.83%, respectively. No mutation was undetectable. While the increased and novel mutations were 336

significantly more common during the CIT phase (20/29 vs. 6/33, during CIT and ibrutinib, respectively; P<0.0001, Fisher’s exact test), the decreased and undetectable mutations were more frequent under ibrutinib treatment (3/29 vs. 19/33, during CIT and ibrutinib, respectively; P<0.0001) (Table 1). In the present study, in TP53-mutated TN and R/R CLL patients, ibrutinib appears to decrease the major and minor mutations’ numerosity and complexity, since most mutations decreased (39% and 24%) or were undetectable (17% and 34%) and one third of mutations remained stable. On the other hand, a small proportion of TP53 mutations (9%, 2 minor, in TN; 13%, 2 major and 2 minor, in R/R cases) increased in CCF under ibrutinib treatment, although without clear clinical consequences with the current follow-up. We observed no association between the dynamics of TP53 mutations and the type of mutation, or the exon involved, neither the type of karyotype (data not shown) nor the presence of del(17p) (8 TN with vs. 9 without del(17p), decreased/undetectable vs. increased/novel mutations, P=0.46; 4 R/R with vs. 3 without del(17p), P=1 at Fisher’s exact test). With a prolonged follow-up of more than 2 years, up to 44 months, our data add to the initial findings of a general stability of TP53 subclones over the early treatment period and support the notion that there is no specific positive selection of TP53 mutations under ibrutinib.7,14 Emergence of novel mutations proved exceptional events, mainly limited to R/R patients that display from the beginning a more complex mutational architecture, suggesting a potential influence of the previous CIT.13 However, in the long-term TP53 disrupted CLL patients tend to experience an inferior outcome.12,13 This can be due to an intrinsic genomic instability and a greater possibility of acquiring additional mutations conferring drug resistance and more frequent relapses in the long-term.7 Moreover, in vitro apoptosis and inhibition of proliferation are inferior in TP53 mutated than in WT cells exposed to ibrutinib, pointing to different mechanisms of cell fitness control in addition to the BCR pathway,15 that can make the difference over time. In conclusion, in TP53-mutated CLL patients ibrutinib in any line of therapy decreases the TP53 complexity at least within the first years of treatment and it does not exert a positive selective pressure on pre-existing TP53 mutated clones, unlike CIT. In TP53 WT patients, ibrutinib never induced the emergence of novel TP53 mutations after >2 years of exposure. These findings reinforce a broader use of a BTK inhibitor rather than CIT in the management of CLL, particularly for patients with an unfavorable genetic profile or with R/R disease. Luciana Cafforio,1 Sara Raponi,1 Luca Vincenzo Cappelli,1 Caterina Ilari,1 Roberta Soscia,1 Maria Stefania De Propris,1 Paola Mariglia,1 Gian Matteo Rigolin,2 Antonella Bardi,2 Nadia Peragine,1 Alfonso Piciocchi,3 Valentina Arena,3 Francesca Romana Mauro,1 Antonio Cuneo,2 Anna Guarini,4 Robin Foà1 and Ilaria Del Giudice1 1 Hematology Section, Department of Translational and Precision Medicine, Sapienza University, Rome; 2Hematology Section, Department of Medical Science, Azienda Ospedaliero-Universitaria Arcispedale S. Anna, University of Ferrara, Ferrara; 3GIMEMA Data Center, GIMEMA Foundation, Rome and 4Department of Molecular Medicine, Sapienza University, Rome, Italy Correspondence: ILARIA DEL GIUDICE - delgiudice@bce.uniroma1.it haematologica | 2022; 107(1)

Letters to the Editor

doi:10.3324/haematol.2020.263715 Received: June 18, 2020. Accepted: October 7, 2021. Pre-published: October 14, 2021. Disclosures: GMR received honoraria from Abbvie, Gilead and Janssen and research funding from Gilead; AC received honoraria from Abbvie, Gilead, Janssen, and is part of the advisory board/speakers bureau of AstraZeneca; RF received honoraria from Abbvie, Janssen, Shire/Takeda, is part of the advisory board/speakers bureau of Gilead; IDG received honoraria from Toledo and is part of the advisory board of AstraZeneca. Contributions: LC performed NGS experiments, the bioinformatic analysis and wrote the manuscript; SR performed NGS experiments and wrote the manuscript; LVC performed NGS experiments and contributed to the results’ analysis; CI performed and analyzed Sanger sequencing; RS performed NGS experiments; MSDP performed flow cytometry; PM and NP managed samples collection/processing and the biologic database; GMR and AB performed karyotype analysis; AP and VA performed the statistical analysis; FRM led the clinical trial and managed patients; AC, AG and RF discussed the results and critically revised the manuscript; IDG designed and supervised the study and wrote the manuscript. Funding: the authors wish to thank the Associazione Italiana per la Ricerca sul Cancro (AIRC), Metastases Special Program, N° 21198, Milan, Italy (RF); Progetti di Rilevante Interesse Nazionale (PRIN) 2015ZMRFEA (RF, AC).

References 1. O'Brien S, Furman RR, Coutre S, et al. Single-agent ibrutinib in treatment-naïve and relapsed/refractory chronic lymphocytic leukemia: a 5year experience. Blood. 2018;131(17):1910-1919. 2. Munir T, Brown JR, O'Brien S, et al. Final analysis from RESONATE: up

haematologica | 2022; 107(1)

to six years of follow-up on ibrutinib in patients with previously treated chronic lymphocytic leukemia or small lymphocytic lymphoma. Am J Hematol. 2019;94(12):1353-1363. 3. Burger JA, Barr PM, Robak T, et al. Long-term efficacy and safety of firstline ibrutinib treatment for patients with CLL/SLL: 5 years of follow-up from the phase 3 RESONATE-2 study. Leukemia. 2020;34(3):787-798. 4. Woyach JA, Ruppert AS, Guinn D, et al. BTKC481S- mediated resistance to ibrutinib in chronic lymphocytic leukemia. J Clin Oncol. 2017; 35(13):1437-1443. 5. Burger JA, Landau DA, Taylor-Weiner A, et al. Clonal evolution in patients with chronic lymphocytic leukaemia developing resistance to BTK inhibition. Nat. Commun. 2016;7:11589. 6. Ahn IE, Underbayev C, Albitar A, et al. Clonal evolution leading to ibrutinib resistance in chronic lymphocytic leukemia. Blood. 2017; 129(11):1469-1479. 7. Landau DA, Sun C, Rosebrock D, et al. The evolutionary landscape of chronic lymphocytic leukemia treated with ibrutinib targeted therapy. Nat Commun. 2017;8(1):2185. 8. Kanagal-Shamanna R, Jain P, Patel KP, et al. Targeted multigene deep sequencing of Bruton tyrosine kinase inhibitor-resistant chronic lymphocytic leukemia with disease progression and Richter transformation. Cancer. 2019;125(4):559-574. 9. Rossi D, Khiabanian H, Spina V, et al. Clinical impact of small TP53 mutated subclones in chronic lymphocytic leukemia. Blood. 2014;123(14):2139-2147. 10. Malcikova J, Stano-Kozubik K, Tichy B, et al. Detailed analysis of therapy-driven clonal evolution of TP53 mutations in chronic lymphocytic leukemia. Leukemia. 2015;29(4):877-885. 11. Nadeu F, Delgado J, Royo C, et al. Clinical impact of clonal and subclonal TP53, SF3B1, BIRC3, NOTCH1, and ATM mutations in chronic lymphocytic leukemia. Blood. 2016;127(17):2122-2130. 12. Kadri S, Lee J, Fitzpatrick C, et al. Clonal evolution underlying leukemia progression and Richter transformation in patients with ibrutinibrelapsed CLL. Blood Adv. 2017;1(12):715-727. 13. Gángó A, Alpár D, Galik B, et al. Dissection of subclonal evolution by temporal mutation profiling in chronic lymphocytic leukemia patients treated with ibrutinib. Int J Cancer. 2020;146(1):85-93. 14. Malcikova J, Pavlova S, Kunt Vonkova B, et al. Low-burden TP53 mutations in CLL: Clinical impact and clonal evolution within the context of different treatment options. Blood. 2021 May 4. [Epub ahead of print] 15. Guarini A, Peragine N, Messina M, et al. Unravelling the suboptimal response of TP53-mutated chronic lymphocytic leukaemia to ibrutinib. Br J Haematol. 2019;184(3):392-396.

337

Letters to the Editor

A novel index to evaluate ineffective erythropoiesis in hematological diseases offers insights into sickle cell disease Ineffective erythropoiesis (IE) is a significant pathological factor in many types of anemia including b-thalassemia and myelodysplasia and has more recently been described in sickle cell disease (SCD). We have evaluated a novel index of ineffective erythropoiesis (IoIE) as a way of quantitating IE and facilitating its comparison between patients, conditions, and the effects of treatments. We calculated IoIE by dividing the plasma concentration of soluble transferrin receptor (sTfR, nmol/L) (proportionate to the volume of erythroid tissue) by the absolute reticulocyte count (ARC, x109/L) (effective erythroid output from the bone marrow). The upper limit of normal IoIE was calculated at 0.28, using the established normal ranges of sTfR and ARC, and confirmed using control samples. We studied 414 SCD patients and show that IE is a feature of patients with HbSS (median IoIE 0.37), but not HbSC (median IoIE 0.27). We validated IoIE as a measure of IE in a cohort of 44 patients with HbE-beta thalassemia, a condition in which IE is known to play a major part and, as expected, found high levels (median IoIE 1.46). Within the HbSS cohort, we find higher HbF levels associate with reduced IE and that transfusion reduces IE, whereas hydroxyurea (HU) treatment appears to lead to increased IE. This index is a simple and meaningful measure of IE, showing that it is clinically important in SCD, and suggesting novel therapeutic approaches. IE is the abnormal differentiation of erythroid progenitors, with an expanded progenitor compartment,

increased erythroblast destruction and a relative paucity of reticulocytes produced compared to the volume of the erythron.1 IE is well described in beta-thalassemia but less so in SCD, where anemia is generally attributed to hemolysis.2 There is, however, evidence of dysfunctional erythroid differentiation in SCD. Erythroblasts isolated from the bone marrow of SCD patients sickle under hypoxic conditions3 whilst analysis of chimeric hematopoiesis in non-myeloablative transplanted SCD patients have demonstrated a survival advantage of the donor erythroid progenitor cells.4 More recently, high levels of apoptosis between the polychromatic and orthochromatic stages were identified, with fetal hemoglobin (HbF) being a key protective factor against this.5 Quantitation of IE in patients would facilitate its study in SCD and other anemias and allow an assessment of novel treatments. Transferrin receptor (CD71) is a membrane protein expressed during erythropoiesis. It is also released into the circulation, with its serum concentration shown to be proportional to the mass of erythropoietic tissue.6,7 Soluble transferrin receptor (sTfR) levels are elevated in SCD, reflecting an increased erythropoietic drive, but show no correlation with disease severity.8 We propose a meaningful representation of IE that can be determined by measuring the ratio of the mass of erythropoietic tissue (sTfR) to the erythroid output from the marrow (ARC), analogous to the way the reticulocyte percent correlates with the rate of hemolysis.9 Where IE is occurring, the erythron mass will be large, with higher sTfR levels, whilst the relative output of reticulocytes (ARC) will be lower, and the ratio will be higher. We refer to this ratio as the Index of Ineffective Erythropoiesis (IoIE), and explore its significance in patients with SCD.

Table 1. Biological parameters of controls, SCD and HbE/b-thalassaemia patients.

Genotype Treatment Number of patients Mean age in years (min - max) Gender G6PD gHbF a-thalassemic

Soluble Transferrin Receptor (nmol/L) Absolute reticulocyte count (100*10x9/L) Hemoglobin (g/dL)

AA Control 22 42.77 (16-93) 8 Female, 14 Male 0 N/A N/A

20.66 (10.6-23.8) 112.8 (48.6-214.4) N/A

HbF (%)

N/A

Erythropoietin

N/A

Non- treated

SS Transfusion

SC Non- treated

S/b + Non- treated

182 31.7 (17.3-63.58) 110 Female, 72 Male 11 2.718 (1.89-2.62) 131 aa /aa, 38 aa/ a-, 13 a -/ a123.9 (41.4-241.1)

58 50 29.17 28.71 (16.94-51.75) (18.1-57.46) 26 Female, 28 Female, 32 Male 22 Male 4 0 2.135 2.137 (1.89-2.69) (1.89-2.49) 39 aa /aa, 31 aa /aa, 17 aa/ a-, 15 aa/a -, 2 a -/ -a 4 a-/ a82.95 103.9 (27.8-187.5)**** (33.26-202)****

87 39.15 (17.9-74.9) 59 Female, 28 Male 2 2.1777 (1.89-2.49) 62 aa /aa, 23 aa/a -, 2 a-/ a (23.1-99.6)

12 42.48 (18.93-68.28) 8 Female, 4 male 1 2.144 (1.89-2.32) 8 aa /aa, 3 aa/ -a, 1 a -/ a51.72 (27.93-90.38)

366.6 (152.6-695) 84.16 (42-124.3) 6.8 (0.2-25.6) 104.8 (10.6-800)

384.5 286.5 208.1 (148.4-888.9)**** (140.3-684.8)**** (91.53-442.1) 95.96 88.44 115.4 (61.8-121)**** (41.6-113.8)* (80-147.5) 3.03 10.38 1.836 (0.2-11.1)**** (1.25-29.02)**** (0.2-11.1) 59.45 112.3 41.34 (21.3-187)**** (24.3-354.3) (14-143)

174.4 (67.97-350.8) 122.1 5.3 (0.7-11.6) 29.84 (9.6-81.15)

HbE/b-thalassemia Non- treated Transfusion 23 11 (1-39) 11 Female, 12 Male 0 N/A

22 18.5 (4-53) 7 Female, 15 Male 0 N/A

N/A

289.2 (121.5-600.3)

132.6 (18.8-404.8)****

262 (100.6-1320) 71.57 (10-105) 45.36 (8.9-66) N/A

90.37 (3.4-267.9)**** 94.5 (11.9-132)*** 35 N/A

***P<0.001 ****P<0.0001.

338

haematologica | 2022; 107(1)

Letters to the Editor

Figure 1. Index of ineffective erythropoeisis and other biological parameters in SCD. A graph representing the (A) absolute reticulocyte count (ARC) (B) Soluble Transferrin Receptor (sTfr) and (C) Index of Ineffective Erythropoiesis (IoIE) in control (n= 22), HbSS (n= 182), HbSC (n= 87) and HbS/b+ (n= 12) patients; the red box represents the normal range of IoIE. A Spearman correlation plot between (D) ARC and sTfr, (E) Hemoglobin and IoIE and (F) Fetal hemoglobin and IoIE in the HbSS group (n=182). A graph showing the (G) ARC, (H) sTfR and (I) IoIE in non-treated (n=182), transfusion (n=58) and HU (n=50) HbSS groups. *P<0.05, **P<0.01, ***P<0.001 ****P<0.0001. (J) Effect of HU therapy on patient IoIE measurements. Each line represents the change in IoIE of an individual patient from before to during HU therapy (n=20). (K) IoIE in non-treated (n= 23) and transfused (n= 22) HbE/b-thalassemia patients (L) Effect of transfusion therapy on patient IoIE measurements (n=17).

haematologica | 2022; 107(1)

339

Letters to the Editor

We also validate this index in a cohort of patients with non-transfusion dependent HbE/b-thalassemia, in which IE is known to predominate and in a control group with no iron deficiency, no hematological disease or other condition which might impact erythropoiesis. All patients with SCD were recruited from King’s College Hospital, London, UK. Written informed consent was obtained through three approved study protocols (LREC 01-083, 07/H0606/165, and 12/LO/1610). Patient electronic records were reviewed from 2008 to the present for all measurements of sTfR and ARC taken on the same day in the outpatient setting. Contemporaneous Hemoglobin (Hb) and HbF levels were also recorded and a note made of disease modifying therapy: ongoing regular blood transfusion within 90 days, HU use, or neither. Measurements were excluded if there was evidence of iron deficiency (serum ferritin <30 micrograms/L), or if the patient was younger than 16 years of age at time of sampling.10 Data from a previous study of patients with HbE/b thalassaemia11 were collected and analyzed to validate the index. Data from 22 control samples was also analyzed; these were samples from adult patients taken for clinical reasons. They were without iron deficiency or any other condition known to affect erythropoiesis or red cell survival. Determination of a thalassemia, G6PD and g(HbF), a composite measure of the major genetic determinants of HbF, in this cohort are described elsewhere.12 sTfR was measured using an ELISA (R&D Systems, Minneapolis, USA) and reticulocytes were measured using automated counting based on RNA staining (Siemens Healthcare, Erlangen, Germany). Index of ineffective erythropoiesis (IoIE) was calculated by dividing sTfR (nmol/L) by the ARC (x109/L). We defined the limit of the normal range by applying this formula to the well-established upper limit of the normal ranges of sTfR (28.1 nmol/L) and ARC (100x109/L), giving a value of 0.28. We validated this with control samples. Where multiple measurements were available, averages were taken. Statistical analyses were performed with GraphPad Prism (version 9). The data were analyzed using the Mann-Whitney unpaired test and the Wilcoxon paired test, as indicated. Correlations were performed using the Spearman correlation. There were 22 controls, 182 patients with HbSS, 87 with HbSC and 12 with HbS/b-thalassemia+ without disease modifying therapy, with a comparable age range and sex ratio (Table 1). As expected, HbSS patients had significantly lower Hb levels and higher ARC, compared to patients with either HbSC or HbS/b+-thalassemia (Figure 1A), and sTfR levels are significantly higher in the HbSS genotype (Figure 1B). There was a strong correlation between ARC and sTfR levels (r=0.3773, P<0.0001, Figure 1D). The IoIE in the non-treated HbSS group was elevated, at 0.37, compared to the control group (0.209), suggesting some degree of IE. This was significantly less than in the validation cohort of 23 non-transfused patients with HbE/b-thalassemia, with a median IoIE of 1.46 (Table 1, Figure 1C). By contrast, in the HbSC group, the IoIE (0.27) was in the normal range though it was slightly higher than in the control group (P=0.0174). The difference between the HbSS and HbSC groups was statistically significant (P<0.0001). Although the numbers were small, the IoIE in the HbS/b+-thalassemia group was higher (0.35) than that of HbSC and significantly higher than the control (P=0.0015), despite similar levels of anemia and disease severity compared to HbSC (Figure 1C). We assessed factors known to influence disease sever340

ity in HbSS with the IoIE. We found a significant negative correlation with Hb (r=-0.32, P=6x10-6), (Figure 1E), and HbF (r=-0.20, P=0.02) (Figure 1F), but no association with G6PD status, (P=0.062), deletional a-thalassemia (P=0.26), or g(HbF) (r=0.02, P=0.82). These findings, together with previous work implicating HbF levels in IE,5 may suggest that increased g globin synthesis in erythroblasts reduces IE in SCD and contributes to higher Hb levels by improving erythropoiesis, as well as prolonging red cell survival. HU and blood transfusion therapy are the two main treatment options in patients with SCD. We evaluated the IoIE in 58 regularly transfused and 50 HU-treated patients. The transfused patients were on long-term regular transfusions, mostly because of cerebrovascular disease to keep the HbS level below 30%. The patients on HU had all been taking it for at least six months and were treated according to clinical response rather than at maximum tolerated dose. In each of these groups, the measurements of Hb, HbF, and ARC differed from baseline in accordance with previously published data (Table 1),13,14 whilst sTfR levels decreased in both groups (Figure 1H). By contrast, the IoIE was significantly lower with transfusions (IoIE=0. 26, P=7.2x10-6), but significantly higher with HU treatment (IoIE=0.42, P=0.042) (Figure 1I). We validated these findings in 17 patients for whom both baseline and post-transfusion measurements were available (paired t-test, P=0.019) (Figure 1L) and 20 patients for whom both baseline and HU treated measurements were available (paired t-test, P=0.004) and confirmed that IoIE actively increased with HU (Figure 1J) and decreased with transfusion (Figure 1K). There was no correlation between HbF and IoIE in patients on HU (P=0.072), although numbers were small. The improvement in IoIE seen in patients on blood transfusions contrasts to that seen with regular blood transfusions to treat HbE/b-thalassemia (n=21), where IoIE remained high at 1.51 (Figure 1K). We suggest this reflects the inherent difference in the causes of IE in thalassemia compared to HbSS. In thalassemia there is intrinsic IE due to the imbalance in globin chain synthesis, which will not be significantly altered by blood transfusion whereas the IE seen in HbSS may be, in part, extrinsically induced by the dysfunctional bone marrow microenvironment, one shown to improve with recurrent blood transfusion in a recent study of a mouse model.15 The relationship between HbF levels and IE in SCD is complex, as demonstrated by the different effects of naturally occurring high HbF levels (reduced IE) and HU (increased IE). This suggests two opposing mechanisms: inherited high HbF levels reduce erythrocyte HbS concentrations and so reduce intramedullary HbS polymerization with improved erythroid survival; conversely, high levels of IE lead to increased selective advantage in favor of cells expressing more HbF,16 resulting in higher circulating HbF levels; this may be one mechanism through which HU mediates its therapeutic increase in HbF, and explains why there is a relative weak correlation between HbF and ineffective erythropoiesis. In summary, we propose the IoIE as a simple, meaningful and useful measure of ineffective erythropoiesis. We use it to demonstrate that IE exists in patients with HbSS, but not HbSC, and that the genetic ability to synthesize more HbF is associated with less IE. Furthermore, we make the clinically important observation that HU increases IE whilst blood transfusion reduces IE. Further investigations are required to understand this action of HU in increasing IE in SCD.17 HU has a direct effect on haematologica | 2022; 107(1)

Letters to the Editor

cell division which would be expected to increase IE. However, it may also be acting in other ways, such as increasing stress erythropoiesis in the bone marrow niche of SCD patients. IoIE may be useful to monitor the effects of transfusion and HU in SCD, and to evaluate the effects of novel therapies. John Brewin,1,2* Sara El Hoss,1* John Strouboulis1 and David Rees1,2 1 Molecular Haematology, Comprehensive Cancer Centre, School of Cancer and Pharmaceutical Sciences, King’s College London and 2 Department of Haematological Medicine, King’s College Hospital, London, UK *JB and SEH contributed equally as co-first authors. Correspondence: SARA EL HOSS - sara.el_hoss@kcl.ac.uk doi:10.3324/haematol.2021.279623 Received: July 12, 2021. Accepted: October 11, 2021. Pre-published: October 21, 2021. Disclosures: no conflicts of interest to disclose. Contributions: JB collected data, analysed, wrote and edited the manuscript; SEH analysed data, wrote and edited the manuscript; JS contributed to the design of the study, interpretation of data and wrote and edited the manuscript; DR designed the study, wrote and edited the manuscript. Acknowledgments: we would like to thank the patients involved in this study and the medical staff of KCH for taking care of the patients. We also thank the members of the Molecular Haematology research lab at King’s College London for their support during this work.

References 1. Orkin SH. Diversification of haematopoietic stem cells to specific lineages. Nat Rev Genet. 2000:1(1):57-64. 2. Ribeil JA, Arlet JB, Dussiot M, et al. Ineffective erythropoiesis in beta-thalassemia. ScientificWorldJournal. 2013;2013:394295. 3. Hasegawa S, Rodgers GP, Dwyer N, et al. Sickling of nucleated ery-

haematologica | 2022; 107(1)

throid precursors from patients with sickle cell anemia. Exp Hematol. 1998;26(4):314-319. 4. Wu CJ, Krishnamurti L, Kutok JL, et al. Evidence for ineffective erythropoiesis in severe sickle cell disease. Blood. 2005;106(10):36393645. 5. El Hoss S, Cochet S, Godard A, et al. Fetal hemoglobin rescues ineffective erythropoiesis in sickle cell disease. Haematologica. 2021; 106(10):2707-2719. 6. Huebers HA, Beguin Y, Pootrakul P, Einspahr D, Finch CA. Intact transferrin receptors in human plasma and their relation to erythropoiesis. Blood. 1990;75(1):102-107. 7. Lulla RR, Thompson AA, Liem RI. Elevated soluble transferrin receptor levels reflect increased erythropoietic drive rather than iron deficiency in pediatric sickle cell disease. Pediatr Blood Cancer. 2010; 55(1):141-144. 8. Al-Saqladi AW, Bin-Gadeem HA, Brabin BJ. Utility of plasma transferrin receptor, ferritin and inflammatory markers in children with sickle cell disease. Paediatr Int Child Health. 2012;32(1):27-34. 9. Hebbel RP. Reconstructing sickle cell disease: a data-based analysis of the "hyperhemolysis paradigm" for pulmonary hypertension from the perspective of evidence-based medicine. Am J Hematol. 2011;86(2):123-154. 10. Allen J, Backstrom KR, Cooper JA, et al. Measurement of soluble transferrin receptor in serum of healthy adults. Clin Chem. 1998; 44(1):35-39. 11. Rees DC, Porter JB, Clegg JB, Weatherall DJ. Why are hemoglobin F levels increased in HbE/beta thalassemia? Blood. 1999;94(9):31993204 12. Brewin JN, Rooks H, Gardner K, et al. Genome wide association study of silent cerebral infarction in sickle cell disease (HbSS and HbSC). Haematologica. 2021;106(6):1770-1773. 13. Rana S, Houston PE, Wang WC, et al. Hydroxyurea and growth in young children with sickle cell disease. Pediatrics. 2014;134(3):465472. 14. Schuchard SB, Lissick JR, Nickel A, et al. Hydroxyurea use in young infants with sickle cell disease. Pediatr Blood Cancer. 2019; 66(7):e27650. 15. Park SY, Matte A, Jung Y, et al. Pathologic angiogenesis in the bone marrow of humanized sickle cell mice is reversed by blood transfusion. Blood. 2020;135(23):2071-2084. 16. Ballas SK. The Evolving Pharmacotherapeutic Landscape for the Treatment of Sickle Cell Disease. Mediterr J Hematol Infect Dis. 2020;12(1):e2020010. 17. Pule GD, Mowla S, Novitzky N, Wiysonge CS, Wonkam A. A systematic review of known mechanisms of hydroxyurea-induced fetal hemoglobin for treatment of sickle cell disease. Expert Rev Hematol. 2015;8(5):669-679.

341

Letters to the Editor

Tumor suppressor function of WT1 in acute promyelocytic leukemia Originally identified as a cancer susceptibility gene, Wilms’ Tumor 1 gene (WT1) is overexpressed or mutated in a wide variety of malignancies, including acute myeloid leukemia (AML). WT1 is a zinc-finger transcription factor comprised of C-terminal zinc-finger DNA binding domains and an N-terminal transactivation domain thought to regulate interactions with partner proteins. Germline WT1 mutations consist primarily of nonsense mutations that truncate the C-terminal domains, or missense mutations that disrupt DNA binding, and these mutations result in both developmental abnormalities and predisposition to Wilms’ tumor.1 In normal human CD34+ hematopoietic stem/progenitor cells (HSPC), wild-type WT1 is expressed at a low level, but it is highly expressed in nearly all cases of AML. Among all AML subtypes, WT1 expression is generally highest in acute promyelocytic leukemia (APL/M3 AML), the AML subtype initiated by the PML-RARA fusion gene (Online Supplementary Figure S1A and D).2,3 In addition, we and others4 have identified recurrent WT1 mutations in APL cases (11/42 [26%] in this study) (Online Supplementary Figure S1C). In seven of 11 of these cases, WT1 mutations occurred at a significantly lower variant allele frequency (VAF) than PML-RARA, suggesting they are co-operating events in subclones (data not shown). WT1 mutations have been associated with worse prognosis in non-M3 AML, although no such association has been shown in APL. The spectrum of WT1 mutations is similar in APL versus other AML cases (Online Supplementary Figure S1B), suggesting that WT1 mutations may have similar biologic activities across all AML subtypes. These observations frame a well-known - but unsolved - paradox that we attempt to address here: does a high level of wild type WT1 expression contribute to the initiation or progression of AML/APL, or conversely, does it reflect a tumor suppressor activity, since inactivating mutations appear to contribute to disease progression? In order to explore these questions, we first tested the ability of Wt1 mutations to co-operate with PML-RARA in a well-characterized murine APL model. Ctsg-PMLRARA mice express the PML-RARA fusion cDNA in immature hematopoietic progenitor cells, and succumb to an APL-like disease with a latency of about 1 year in C57Bl/6J mice.5 In order to test whether Wt1 mutations can co-operate with PML-RARA in this model, we used CRISPR/Cas9 to generate indels in Wt1 exon 8 or, as a control, the Rosa26 locus. Since murine Wt1 is highly homologous to the human protein, these mutations should mimic those commonly found in APL patients. Despite efficient mutation generation in Ctsg-PML-RARA bone marrow cells (Online Supplementary Figure S2A), there was no survival difference between mice transplanted with Ctsg-PML-RARA cells with Wt1 mutations versus Rosa26 mutations (Online Supplementary Figure S2C). We therefore evaluated APL tumors arising in these mice, and observed that tumors could arise either from wild-type or mutant Wt1/Rosa26 progenitors (Online Supplementary Figure S2B). Surprisingly, we did not detect Wt1 protein in these tumors by western blotting (data not shown). Similarly, in a panel of 16 previously banked murine APL tumors from Ctsg-PML-RARA mice, RNA sequencing revealed virtually undetectable levels of Wt1 mRNA (Online Supplementary Figure S2D), 342

in contrast to the high expression seen in human APL samples. Together, these data suggest that important interspecies differences in Wt1 regulation and function are important for the lack of a phenotype in this mouse model. We next evaluated WT1 expression in human CD34+ cells by transducing umbilical cord blood-derived CD34+ cells with retroviruses encoding PML-RARA, RUNX1-RUNX1T1, or MYC; RUNX1-RUNX1T1 and MYC have previously been shown to confer self-renewal and expansion of human HSPCs in vitro and in xenograft models, and therefore act as positive controls.6,7 Consistent with previously reported results, we found that human HSPC transduced with RUNX1-RUNX1T1 and MYC expand robustly over 2 weeks in culture, while HSPC transduced with PML-RARA expand more slowly (data not shown). In order to test whether WT1 expression is affected by transduction with these retroviral constructs, GFP+ cells were sorted 7 days after transduction, RNA was isolated, and WT1 expression was measured by quantitative reverse transcription polymerase chain reaction (RT-PCR). Figure 1A shows upregulation of WT1 mRNA in sorted human HSPC transduced with PML-RARA, RUNX1-RUNX1T1, and MYC (6 to 18-fold increase, P<0.05 for PML-RARA and MYC). WT1 protein abundance also increased dramatically in the same cells during this timeframe (Figure 1B). In order to identify other genes dysregulated by PML-RARA transduction, we transduced both mouse and human HSPC with GFPtagged retroviruses encoding PML-RARA or an empty vector, as has been previously reported (n=2 and n=3 separate experiments for human and mouse cells respectively).8,9 Seven days after transduction, GFP+ cells were sorted and bulk RNA sequencing was performed to identify differentially expressed genes (DEG, 5,347 identified for mouse samples, and 1,885 identified for human samples, Figure 1C). There was significant overlap between orthologous mouse and human DEG after transduction with PML-RARA (Figure 1D, P=9.6x10-118 based on the hypergeometric test); further, of 867 overlapping orthologues, 82% were coordinately regulated. However, while WT1 was ~13-fold upregulated in human cells transduced with PML-RARA, Wt1 expression was extremely low and did not increase in murine cells transduced with PML-RARA (Figure 1E), validating the interspecies difference in WT1/Wt1 regulation noted above. GFP+ PML-RARA-expressing human cord blood cells expanded modestly during the first weeks of culture, but increased dramatically 3-4 weeks after transduction (Online Supplementary Figure S3A). After 6 weeks in culture, they resembled primary APL cells morphologically and immunophenotypically (Online Supplementary Figure S3B and C). In addition, they were sensitive to treatment with all-trans retinoic acid (ATRA), a hallmark of APL cells (Online Supplementary Figure S3D). Transduced cells were not immortalized, as they stopped proliferating around 8-9 weeks after initiation, and they failed to engraft immunodeficient mice (data not shown). Given its reported role as a tumor suppressor in other cancer types, we hypothesized that this upregulation may reflect an attempt of WT1 to suppress the proliferative response induced by PML-RARA, similar to the increased TP53 activity observed in cells responding to genotoxic stressors.10 However, as noted above, it is also possible that high WT1 expression actively promotes the growth or survival of APL cells. In order to distinguish between these possibilities, we performed WT1 overexpression versus loss-of-function experiments in PML-RARA-transduced cord blood cells. haematologica | 2022; 107(1)

Letters to the Editor

Figure 1. WT1 expression is induced in human CD34+ cells transduced with RUNX1-RUNX1T1, PML-RARA, or MYC. Umbilical cord blood-derived CD34+ cells were cultured in cytokines after transduction with GFP-tagged retroviruses expressing RUNX1-RUNX1T1, MYC, PML-RARA, or an empty vector control. (A) WT1 expression is induced in CD34+ cells transduced with RUNX1-RUNX1T1, MYC, or PML-RARA compared to controls transduced with empty vector (GFP, green) or untransduced cells (CD34, pink). CD34+ cells transduced with each vector (n=2-6 separate experiments) were cultured for 7 days, RNA was isolated from flow-sorted GFP+ cells, and WT1 mRNA was quantified by real time polymerase chain reaction. P-values were calculated using Student’s t-test. (B) Western blot showing expression of WT1 in CD34+ cord blood cells 7 days after transduction with RUNX1-RUNX1T1 (AE), PML-RARA (PR), MYC, empty vector (GFP), or untransduced (CTRL). Lysates were made from sorted GFP+ cells except control (CTRL), which was made from equivalent cell numbers of untransduced cells cultured in parallel. Blot represents one of 3 representative experiments. (C) Heatmaps showing differentially expressed genes (DEG) in human or mouse cells transduced with a PML-RARA-expressing MSCV vector. Human CD34+ cells or mouse lineage-depleted bone marrow cells were transduced with IRES-GFPtagged retroviruses containing a PML-RARA cDNA, or no insert (empty vector). After 7 days in culture, GFP+ cells were flow sorted and RNA was isolated for RNA sequencing. DEG were identified using a false discovery rate (FDR) cutoff of <0.05 after filtering out genes with low expression across all samples (see the Online Supplementary Appendix). Heatmaps show DEG in PML-RARA vs. empty vector-transduced human (n=2 separate experiments) and mouse (n=3 separate experiments) progenitor cells. (D) Venn diagram showing overlap in orthologous mouse and human DEG from (C). Of 4,915 mouse DEG having human orthologues, 867 are DEG in the analysis of human genes (P=9.6x10-118 using the hypergeometric test). (E) WT1 expression is increased by PML-RARA transduction in human CD34+ cells (left panel), but not in mouse bone marrow-derived cells (right panel). WT1/Wt1 expression values from the RNA sequencing experiment described above are shown. P-values were calculated using Student’s t-test. TPM: transcripts per million.

haematologica | 2022; 107(1)

343

Letters to the Editor

First, we used lentiviral vectors to overexpress the two most common isoforms of WT1 (KTS+ and KTS-) in human CD34 cells in culture. In the absence of other cooperating oncogenes, WT1 overexpression led to the rapid disappearance of transduced cells (Figure 2A), consistent with previous reports that WT1 overexpression

causes differentiation and death of CD34+ cells.11 Next, we asked whether inactivating mutations in WT1 could enhance expansion of CD34+ cells expressing PMLRARA. We transduced CD34+ cells with PML-RARA or empty vector, and 2 days after transduction used CRISPR/Cas9 to generate inactivating indels in WT1, or

Figure 2. Legend on following page.

344

haematologica | 2022; 107(1)

Letters to the Editor

Figure 2. Inactivating mutations in WT1 provide a growth advantage for PML-RARA-transduced CD34+ cells. (A) Umbilical cord blood-derived CD34+ cells were transduced with GFP-tagged lentiviruses encoding the two most common WT1 isoforms (KTS+ and KTS-), or an empty vector. Cells were maintained in culture with cytokines, and GFP+ cells were quantified at different time points. Shown are percent GFP+ cells over time in WT1 (right) or empty vector (left) transduced cultures normalized for transduction efficiency at beginning of the culture period (n=4 individual experiments). Black dotted lines show line of best fit calculated by linear regression. Transduction with WT1 isoforms (KTS+ and KTS-) leads to loss of GFP+ cells (slope b=-1.18 per day, P<0.001), while empty vector-transduced cells have GFP+ cells throughout the culture period (slope b=-0.54, P=0.32). P-values were calculated using a linear regression model, and represent the probability that the slope of the best fit line equals zero. (B) Human CD34+ cord blood cells were transduced with PML-RARA-expressing retrovirus or empty vector, and 48 hours later CRISPR/Cas9 was used to generate mutations in WT1 (exon 1 or exon 8) or AAVS1 (a negative control locus). GFP+ cells were sorted at different time points from cultures that had been transduced with a vector containing PML-RARA (right panels) or no insert (empty vector, left panels). DNA was isolated and polymerase chain reaction products containing the guide RNA target sites were digitally sequenced to determine the precise variant allele frequencies of mutations in WT1 (bottom panels) or AAVS1 (top panels). Shown are change in variant allele frequency (VAF) of AAVS1 mutations or WT1 mutations over time (n=3-6 separate experiments). Mutations in WT1 exon1 are shown in green, or WT1 exon 8 in red. Black dotted lines show line of best fit calculated by linear regression. Cells containing mutations in WT1 show a trend toward expansion in empty vector-transduced CD34+ cells (slope b=0.003 increase per day, P=0.20), and a statistically significant expansion in PML-RARA-transduced cells (slope b=0.006 increase per day, P=0.007). In contrast, cells with mutations in AAVS1 do not expand over time. P-values were calculated using a linear regression model, and represent the probability that the slope of the best fit line equals zero. (C) Increase in overall cell numbers in cultures transduced with GFP (left) or PML-RARA (right). P-values were calculated using Student’s t-test.

as a control, AAVS1. After 4-8 weeks in culture, GFP+ cells were sorted and the frequency of WT1 or AAVS1 indels in GFP+ cells at the end of the culture period was compared to the frequency at the beginning. WT1 mutations were selected for over time, and were significantly increased in PML-RARA-transduced cells 4-8 weeks after the WT1 mutations were introduced (Figure 2B, bottom panels). In contrast, cells containing mutations in AAVS1 did not increase in frequency (Figure 2B, top panels). Overall cell numbers significantly increased in cells bearing both PML-RARA and WT1 mutations, compared to cells with PML-RARA and AAVS1 mutations, or cells transduced with an empty vector with WT1 mutations (Figure 2C). Together, these findings suggest that WT1 inactivation enhances the growth of PML-RARAexpressing hematopoietic cells, strongly suggesting that WT1 acts as a tumor suppressor in this context. Based on the above findings, we propose a simple model to explain these paradoxical observations: WT1 expression in HSPC is normally activated as an adaptive and inhibitory response to oncogenic mutations that cause proliferation, a response that is intended to slow their growth. The subsequent development of inactivating WT1 mutations in some cases would then provide a further growth advantage by removing that normal inhibitory response. Supporting this hypothesis, we found that i) retroviral transduction of CD34+ cells with PML-RARA, RUNX1-RUNX1T1, or MYC all led to a robust induction of WT1 expression; ii) forced expression of wild-type WT1 by itself does not promote CD34 cell expansion; and iii) inactivation of WT1 in PMLRARA-expressing CD34+ cells leads to an additional growth advantage. Although the mechanism of WT1 gene activation by oncogenes is not yet clear, high levels of WT1 expression are found in nearly all AML cases, regardless of subtype or mutational landscape. In addition, since the majority of AML/APL cases do not have WT1 mutations, a corollary of this hypothesis is that leukemias with wild-type WT1 must have developed alternative means to circumvent the inhibitory pressure that WT1 induction may exert. Finally, the downstream mechanisms by which WT1 mutations lead to a growth advantage in AML cells are currently unclear, and may depend on the context of the co-operating mutations. In addition to its welldescribed function as a locus-specific transcription factor,12 recent studies have suggested that WT1 mutations may cause epigenetic changes via effects on DNA methylation and interactions with TET family methylcytosine deoxygenases.13-16 A better understanding of how WT1 mutations activate these pathways in AML

haematologica | 2022; 107(1)

cells will be required to fully exploit their potential therapeutic value. Matthew J. Christopher,1 Casey D. S. Katerndahl,1 Hayley R. LeBlanc,1 Tyler T. Elmendorf,1 Vaishali Basu,1 Margery Gang,1 Andrew J. Menssen,1 David H. Spencer,1 Eric J. Duncavage,2 Shamika Ketkar,1° Lukas D. Wartman,1 Sai Mukund Ramakrishnan,1 Christopher A. Miller1 and Timothy J. Ley1 1 Section of Stem Cell Biology, Division of Oncology, Department of Internal Medicine, Washington University in St. Louis, St. Louis, MO and 2Department of Pathology and Immunology, Washington University in St. Louis, St. Louis, MO, USA °Current address: Department of Molecular and Human Genetics, Baylor College of Medicine, Houston, TX, USA Correspondence: MATTHEW J. CHRISTOPHER- christopherm@wustl.edu doi:10.3324/haematol.2021.279601 Received: July 9, 2021. Accepted: October 12, 2021. Pre-published: October 21, 2021. Disclosures: no conflicts of interest to disclose. Contributions: MJC performed data analysis, supervised the study, and wrote the manuscript; DHS performed data analysis; EJD, SK, LDW, SRM and CAM performed data analysis; CDSK, HRL, TTE, VB, MG and AJM performed experiments; TJL supervised the study and wrote the manuscript. Acknowledgments: technical assistance was provided by the Siteman Cancer Center Tissue Procurement and Cell Sorting cores (NCI Cancer Center Support Grant P30CA91842). Umbilical cord blood units were provided by the St. Louis Cord Blood Bank. Funding: this work was supported by the National Cancer Institute to MJC (K12 program, CA167540, R. Govindan, PI; SPORE Career Enhancement Award, P50 CA171963, D. Link, PI; and K08CA222630). Additional support was provided by NIH grants CA101937 and CA197561 (TJL), CA211782 (CAM), T32 HL007088 (CDSK), K08CA190815 (DS), and the Barnes Jewish Hospital Foundation (TJL).

References 1. Royer-Pokora B, Beier M Fau - Henzler M, et al. Twenty-four new cases of WT1 germline mutations and review of the literature: genotype/phenotype correlations for Wilms tumor development. Am J Med Genet A. 2014;127A(3):249-257. 2. Ley TJ, Miller C, Ding L, et al. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med. 2013;368(22):2059-2074. 3. Cilloni D, Gottardi E, De Micheli D, et al. Quantitative assessment

345

Letters to the Editor

of WT1 expression by real time quantitative PCR may be a useful tool for monitoring minimal residual disease in acute leukemia patients. Leukemia. 2002;16(10):2115-2121. 4. Madan V, Shyamsunder P, Han L, et al. Comprehensive mutational analysis of primary and relapse acute promyelocytic leukemia. Leukemia. 2016;30(12):2430. 5. Westervelt P, Lane AA, Pollock JL, et al. High-penetrance mouse model of acute promyelocytic leukemia with very low levels of PML-RARα expression. Blood. 2003;102(5):1857-1865. 6. Bulaeva E, Pellacani D, Nakamichi N, et al. MYC-induced human acute myeloid leukemia requires a continuing IL3/GM-CSF co-stimulus. Blood. 2020;136(24):2764-2773. 7. Mulloy JC, Cammenga J, MacKenzie KL, Berguido FJ, Moore MA, Nimer SD. The AML1-ETO fusion protein promotes the expansion of human hematopoietic stem cells. Blood. 2002;99(1):15-23. 8. Minucci S, Monestiroli S, Giavara S, et al. PML-RAR induces promyelocytic leukemias with high efficiency following retroviral gene transfer into purified murine hematopoietic progenitors. Blood. 2002;100(8):2989-2995. 9. Sternsdorf T, Phan VT, Maunakea ML, et al. Forced retinoic acid receptor alpha homodimers prime mice for APL-like leukemia. Cancer Cell. 2006;9(2):81-94.

346

10. Karpinich NO, Tafani M, Rothman RJ, Russo MA, Farber JL. The course of etoposide-induced apoptosis from damage to DNA and p53 activation to mitochondrial release of cytochrome c. J Biol Chem. 2002;277(19):16547-16552. 11. Ellisen LW, Carlesso N, Cheng T, Scadden DT, Haber DA. The Wilms tumor suppressor WT1 directs stage-specific quiescence and differentiation of human hematopoietic progenitor cells. EMBO J. 2001;20(8):1897-1909. 12. Toska E, Roberts SGE. Mechanisms of transcriptional regulation by WT1 (Wilms' tumour 1). Biochem J. 2014;461(1):15-32. 13. Pronier E, Bowman RL, Ahn J, et al. Genetic and epigenetic evolution as a contributor to WT1-mutant leukemogenesis. Blood. 2018;132(12):1265-1278. 14. Rampal R, Alkalin A, Madzo J, et al. DNA hydroxymethylation profiling reveals that WT1 mutations result in loss of TET2 function in acute myeloid leukemia. Cell Rep. 2014;9(5):1841-1855. 15. Sinha S, Thomas D, Yu L, et al. Mutant WT1 is associated with DNA hypermethylation of PRC2 targets in AML and responds to EZH2 inhibition. Blood. 2015;125(2):316-326. 16. Wang Y, Xiao M, Chen X, et al. WT1 recruits TET2 to regulate its target gene expression and suppress leukemia cell proliferation. Mol Cell. 2015;57(4):662-673.

haematologica | 2022; 107(1)

Letters to the Editor

Increased double-negative ab+ T cells reveal adultonset autoimmune lymphoproliferative syndrome in a patient with IgG4-related disease Autoimmune lymphoproliferative syndrome (ALPS) is a rare genetic disorder of defective lymphocyte apoptosis characterized by non-malignant expansion of CD4 and CD8 double negative T-cell receptor (TCR) ab+ T cells (ab+DNT) leading to chronic lymphadenopathy, splenomegaly, autoimmune cytopenias and increased susceptibility to malignancy, particularly Hodgkin and non-Hodgkin lymphoma.1 ALPS is driven by mutations in the FS-7 fibroblast cell line-associated surface antigen (FAS)/CD95 signaling pathway, with deleterious hemizygous mutations in the FAS gene representing approximately 70% of cases. Other less commonly involved genes are FASL, FADD and CASP10.1 Affected patients are typically diagnosed in early childhood, however due to incomplete penetrance and variable expressivity, some patients are asymptomatic or may present in adulthood.2 Genotype-phenotype correlations have also been described, with a more severe disease course associated with dominant negative FAS mutations involving the intracellular death domain, while FAS mutations in the extracellular domain may lead to haploinsufficiency and a milder phenotype.1 IgG4-related disease (IgG4-RD) is a systemic immunemediated fibroinflammatory disease characterized by infiltration of lymphocytes, eosinophils and IgG4-positive plasma cells in various organs with associated fibrosis.3 Onset of IgG4-RD typically occurs between 50 and 70 years of age and the symptoms vary depending on the affected organ.3 The most commonly involved organs include the salivary and lacrimal glands, pancreas and biliary tract, kidneys and lymph nodes. Laboratory findings may include elevated serum IgG4, eosinophilia, increased serum IgE and increased serum plasmablasts. Tissue biopsy is the diagnostic gold standard for IgG4RD, which classically demonstrates a dense lymphoplasmacytic infiltrate, storiform fibrosis, obliterative phlebitis, and an elevation in IgG4+ plasma cells. Various cutoffs for the IgG4/IgG ratio and absolute number of IgG4+ cells per high-power field (hpf) have been proposed, and this metric has been incorporated into the 2019 American College of Rheumatology/European League Against Rheumatism (ACR/EULAR) Classification Criteria.4 The co-occurrence of ALPS and IgG4-RD is extremely rare with only two cases currently reported in the English literature.5-6 In both prior reports, the patients were initially diagnosed with ALPS with subsequent development of IgG4-RD several years later. Herein we report the first case of an adult patient with IgG4-RD, in which expansion of ab+DNT by flow cytometry and subsequent genetic testing ultimately uncovered an underlying diagnosis of ALPS with a pathogenic FAS mutation. A 63-year-old female with a history of peripheral Tcell lymphoma, not otherwise specified (PTCL, NOS) and invasive ductal carcinoma (IDC) of the breast presented with left axillary lymphadenopathy. At 54 years of age, she was diagnosed with stage IIIA PTCL and treated on a clinical trial with six cycles of cyclophosphamide, etoposide, vincristine, and prednisone alternating with pralatrexate with a complete remission followed by consolidative autologous stem cell transplant. At 59 years of age, she was diagnosed with stage IA (pT1bN0) triple negative IDC of the left breast and treated with lumpectomy, adjuvant carboplatin/paclitaxel haematologica | 2022; 107(1)

and radiotherapy. One-year following treatment for breast cancer, she developed bilateral cervical lymphadenopathy with a waxing and waning course. She was observed for approximately 1 year until an excisional biopsy of a right cervical lymph node was performed. The biopsy showed no evidence of carcinoma or lymphoma, but rather increased IgG4+ cells (>50/HPF, and >40% IgG4:IgG4 ratio) and multicentric Castleman disease-like features, including interfollicular plasmacytosis, follicles showing multiple germinal centers (“twinning”), concentric layering of mantle zone lymphocytes around follicles (“onion skinning”) and germinal center lymphocyte depletion (“regression”) (Figure 1). Serologic examination for human immunodeficiency viruse (HIV) and human gamma herpesvirus 8 (HHV8) were negative, with the differential diagnosis including IgG4-RD and HHV8-negative idiopathic multicentric Castleman disease (iMCD). The patient continued to experience waxing and waning cervical and axillary adenopathy, ultimately leading to a positron emission tomography (PET) scan, which showed extensive hypermetabolic adenopathy above and below the diaphragm, along with PETavid lesions in the pancreas and spleen, concerning for recurrent lymphoma (Figure 2A). A fine needle aspiration and core biopsy performed on the pancreatic lesion showed an increase of IgG4+ cells (>25/HPF [high power field of microscope] and an IgG4:IgG ratio >50%) and storiform fibrosis (Figure 2B to D). Serum IgG4 and IgE were markedly elevated at 18.33 g/L (normal 0.11-1.57 g/L) and 24,98 kU/L (normal <100 kU/L), respectively. Peripheral blood B-cell phenotyping showed an increase in serum CD19+CD38+CD27+ plasmablasts, enumerated at 63.9% of B cells (normal 0.7-6%). Overall, the clinical, pathologic, and serologic findings were diagnostic of IgG4-RD per the 2019 ACR/EULAR classification criteria. Prior to starting treatment for IgG4-RD, excisional biopsy of a PET-avid right level V cervical lymph node was performed to rule out recurrent lymphoma. This biopsy showed similar findings of increased IgG4+ cells (>50/HPF, and >80% IgG4:IgG4 ratio) and multicentric Castleman disease-like features (Figure 3A to D). Flow cytometry demonstrated an increase in ab+DNT (7.1% of lymphocytes and 14.2% of CD3+ T cells), and immunohistochemistry showed paracortical localization of these T-cells around reactive germinal centers raising suspicion for ALPS (Figure 3E to H). Peripheral blood flow cytometry confirmed an increase in circulating ab+DNT (4.3% of lymphocytes, 6.1% of CD3+ T cells, 27 cells/uL). Additional laboratory testing revealed elevated soluble FAS ligand (824 pg/mL) and markedly elevated serum vitamin B12 level above the measured range for our laboratory instrument (>1,000 pg/mL). T-cell receptor b and g chain rearrangements by next-generation sequencing (NGS) were negative, while targeted NGS mutational analysis (Stanford Actionable Mutation Panel for Hematopoietic and Lymphoid Malignancies) detected a pathogenic FAS c.841T>A (p.W281R) mutation, that resulted in an amino acid substitution likely to be damaging in a region essential for formation of the death-inducing signaling complex involved in apoptosis induction.1 This variant had an allele frequency of 56% suggesting a germline mutation and confirming the diagnosis of ALPS-FAS.1 ALPS caused by FAS mutation (ALPS-FAS) usually manifests in early childhood at a median age of 2-3 years.1 However, due to incomplete penetrance, variable expressivity, and variation in phenotype according to genotype, a subset of patients may be asymptomatic or present in adulthood.1,2 In these cases acquisition of fam347

Letters to the Editor

C C

D D

Figure 1. Increased immunoglobulin G4 postive plasma cells in a lymph node with multicentric Castleman disease-like features. Right cervical lymph node excision stained with hemoxylin and eosin (H&E). (A) 2x objective, 20x total magnification and (B) 20x objective, 200x total magnification). (C) Immunoglobulin G4 (IgG4) 20x objective, 200x total magnification and (D) IgG 20x objective, 200x total magnification.

348

Figure 2. Radiographic and pathologic features of immunoglobulin G4-related disease. (A) Positron emission tomography maximum intensity projection image showing hypermetabolic lymphadenopathy above and below the diaphragm along with lesions in the pancreas and spleen . (B) Pancreatic core needle biopsy stained with hemoxylin and eosin (H&E), (C) immunoglobulin G4 (IgG4) and (D) IgG. For all micrographs, a 20x objective was used and 200x total magnification is presented.

haematologica | 2022; 107(1)

Letters to the Editor

were masked by IgG4-RD. Thus, as previously suggested by van de Ven and colleagues,6 screening for increased ab+DNT by flow cytometry or ALPS-associated mutations by NGS should be considered in patients with IgG4-RD, particularly in those with other clinical, pathologic, or laboratory features characteristic of ALPS. Identification of ALPS-FAS patients with concurrent IgG4-RD may have significant therapeutic implications, as patients may require chronic therapy or become intolerant to standard immunosuppressive therapy, and thus may benefit from targeted, steroid-sparing therapy such as rituximab or sirolimus.8 A mechanistic link between the pathogenesis of ALPS and IgG4-RD is currently unknown. B lymphocytes are involved in the pathogenesis of IgG4-RD as evidenced by marked clinical responses to B-cell-directed therapy with rituximab. It is hypothesized that the oligoclonal

ily and past medical history of malignancy may be informative, as ALPS-FAS and other germline diseases are known to predispose to lymphoma and solid tumors. However, in cases of adult-onset ALPS-FAS, the diagnosis may still be challenging as other hematologic diseases may have overlapping clinical and pathologic features, including angioimmunoblastic T-cell lymphoma, RosaiDorfman disease, iMCD, and IgG4-RD. Rarely, ALPS may co-occur with one of these disease entities and as such mask the typical morphologic features of ALPS.5,6 While certain disease working groups, including that for iMCD, currently recommend evaluation for ALPS, the current consensus diagnostic criteria for IgG4-RD do not.7 In our reported case, if flow cytometry had not been performed to further investigate the increased ab+DNT, the diagnosis of ALPS would likely have been missed as the lymph node histopathological features

Figure 3. Expansion of double-negative αβ+ T cells in a background of immunoglobulin G4-related disease. (A) Right level V cervical lymph node excision stained with hemoxylin and eosin (H&E), (B) CD20, (C) immunoglobulin G4 (IgG4) and (D) IgG. (E, left) Flow cytometry analysis of lymph node gated on CD3+ lymphocytes showing CD4 vs. CD8 and (E, right) gated on CD3+CD4-CD8- showing TCRab vs. TCR𝛾𝛿. (F) Immunostaining for CD3, (G) CD4 and (H) CD8. For all micrographs, a 20x objective was used and 200x total magnification is presented.

haematologica | 2022; 107(1)

349

Letters to the Editor

expansion of B cells and plasmablasts leads to IgG4 production, which may contribute to IgG4-RD pathogenesis9. Plasmablasts from patients with IgG4-RD show extensive immunoglobulin somatic hypermutation, upregulation of FAS/CD95, and active proliferation and secretion of IgG4.10 Conceivably, dysregulation of FAS signaling in plasmablasts in ALPS patients may contribute to a preponderance of plasmablasts, and if skewed toward IgG4+ to IgG4-RD. The extensive immunoglobulin somatic hypermutation in plasmablasts suggests a T-cell-dependent germinal center-derived ontology. As such, T lymphocytes have also been implicated in the pathogenesis of IgG4-RD, particularly T-follicular helper cells (TFH), T-follicular regulatory cells (Tregs), and CD4+ cytotoxic T lymphocytes (CD4 CTL). Cytokine production by these T-cell subsets appear to contribute to IgG4-RD pathogenesis. Namely, IL-4 and IL-10 produced by TFH and Tregs promote IgG4 isotype switching, whereas IL-1 and TGFb produced by CD4 CTL promote fibrosis.11-13 While CD4+ T cells are the best characterized, Carruthers et al. have shown that expansion of DNT may also occur in IgG4-RD, however it is unknown if DNT contribute to disease pathogenesis.14 A recent study by Maccari et al. has identified and characterized ab+DNT in ALPS patients using RNA sequencing, mass cytometry (CyTOF) and functional cytokine analysis. They found that ALPS-DNT show a unique surface marker profile with high expression of CD38, CD45RA, CD27, CD28, CLTA4, TIGIT and TIM3, and additionally show upregulation of IL-10 transcripts and protein levels.15 Conceivably, ALPS-DNT may participate in IL-10-mediated class switching of Bcells/plasmablasts to IgG4, however further studies are needed to test this hypothesis. In conclusion, this case demonstrates the utility of assessing for expanded ab+DNT in patients with IgG4RD, which revealed the diagnosis of ALPS-FAS in our patient. Future studies are needed to investigate a potential mechanistic link between these entities. Nivaz Brar,1* Michael A. Spinner,2* Matthew C. Baker,3 Ranjana H. Advani,2 Yasodha Natkunam,1 David B. Lewis4 and Oscar Silva1 1 Department of Pathology, Stanford University School of Medicine; 2 Division of Oncology, Department of Medicine, Stanford University School of Medicine; 3Division of Immunology and Rheumatology, Department of Medicine Stanford University School of Medicine and 4 Division of Allergy, Immunology, and Rheumatology, Department of Pediatrics, Stanford University School of Medicine, Stanford, CA, USA *NB and MAS contributed equally as co-first authors.

350

Correspondence: OSCAR SILVA MD, PhD - osilva@stanford.edu doi:10.3324/haematol.2021.279297 Received: June 9, 2021. Accepted: August 26, 2021. Pre-published: September 2, 2021. Disclosures: no conflicts of interest to disclose Contributions: NB, MAS and OS collected data, generated figures and wrote the initial draft of the manuscript. All authors contributed to the writing and/or editing of the manuscript.

References 1. Price S, Shaw PA, Seitz A, Joshi G, et al. Natural history of autoimmune lymphoproliferative syndrome associated with FAS gene mutations. Blood. 2014;123(13):1989-1999. 2. Lambotte O, Neven B, Galicier L, et al. Diagnosis of autoimmune lymphoproliferative syndrome caused by FAS deficiency in adults. Haematologica. 2013;98(3):389-392. PMID: 22983577 3. Stone JH, Zen Y, Deshpande V. IgG4-related disease. N Engl J Med. 2012;366(6):539-551. 4. Wallace ZS Naden RP, Chari S, et al. The 2019 American College of Rheumatology/European League Against Rheumatism classification criteria for IgG4-related disease. Ann Reum Dis. 2020;72(1):77-87. 5. Langan RC, Gill F, Raiji MT, et al. Autoimmune pancreatitis in the autoimmune lymphoproliferative syndrome (ALPS): a sheep in wolves' clothing? Pancreas. 2013;42(2):363-366. 6. van de Ven AAJM, Seidl M, Drendel V, et al. IgG4-related disease in autoimmune lymphoproliferative syndrome. Clin Immunol. 2017;180:97-99. 7. Fajgenbaum DC, Uldrick TS, Bagg A, et al. International, evidencebased consensus diagnostic criteria for HHV-8-negative/idiopathic multicentric Castleman disease. Blood. 2017;129(12):1646-1657. 8. Bride KL, Vincent T, Smith-Whitley K, et al. Sirolimus is effective in relapsed/refractory autoimmune cytopenias: results of a prospective multi-institutional trial. Blood. 2016;127(1):17-28. 9. Perugino CA, Stone JH. IgG4-related disease: an update on pathophysiology and implications for clinical care. Nat Rev Rheumatol. 2020;16(12):702-714. 10. Lin W, Zhang P, Chen H, et al. Circulating plasmablasts/plasma cells: a potential biomarker for IgG4-related disease. Arthritis Res Ther. 2017;19(1):25. 11. Jeannin P, Lecoanet S, Delneste Y, Gauchat JF, Bonnefoy JY. IgE versus IgG4 production can be differentially regulated by IL-10. J Immunol. 1998;160(7):3555-3561. 12. Morita R, Schmitt N, Bentebibel SE, et al. Human blood CXCR5(+)CD4(+) T cells are counterparts of T follicular cells and contain specific subsets that differentially support antibody secretion. Immunity. 2011;34(1):108-121. 13. Mattoo H, Mahajan VS, Maehara T, et al. Clonal expansion of CD4(+) cytotoxic T lymphocytes in patients with IgG4-related disease. J Allergy Clin Immunol. 2016;138(3):825-838. 14. Carruthers MN, Park S, Slack GW, et al. IgG4-related disease and lymphocyte-variant hypereosinophilic syndrome: A comparative case series. Eur J Haematol. 2017;98(4):378-387. 15. Maccari ME, Fuchs S, Kury P, et al. A distinct CD38+CD45RA+ population of CD4+, CD8+, and double-negative T cells is controlled by FAS. J Exp Med. 2021;218(2):e20192191.

haematologica | 2022; 107(1)

Hameatologica, Volume 107, Issue 1

Articles inside

Indolent lymphomas: introduction to a series highlighting progress and ongoing challenges

Adding eltrombopag to immunosuppression: the importance of predicting outcome

Fusion genes in acute myeloid leukemia: do acute myeloid leukemia diagnostics need to fuse with RNA-sequencing?

Prospects in the management of patients with follicular lymphoma beyond first-line therapy

Time for an individualized approach to first-line management of follicular lymphoma

Marginal zone lymphoma: present status and future perspectives

Toward further excellence in Haematologica

Images from the Haematologica Atlas of Hematologic Cytology: Sézary syndrome