Haematologica, Volume 106, Issue 9 by Haematologica

haematologica Journal of the Ferrata Storti Foundation

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

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

Managing Director Antonio Majocchi (Pavia)

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)

Assistant Editors Britta Dorst (English Editor), Catherine Klersy (Statistical Consultant), Rachel Stenner (English Editor)

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)

Editorial Office Simona Giri (Production & Marketing Manager), Lorella Ripari (Peer Review Manager), Paola Cariati (Senior Graphic Designer), Igor Ebuli Poletti (Senior Graphic Designer), Marta Fossati (Peer Review), Diana Serena Ravera (Peer Review)

Affiliated Scientific Societies SIE (Italian Society of Hematology, www.siematologia.it) SIES (Italian Society of Experimental Hematology, www.siesonline.it)

haematologica Journal of the Ferrata Storti Foundation

Information for readers, authors and subscribers 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, research articles, review articles, guideline articles and letters. 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 online (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. Detailed information about subscriptions is available online at www.haematologica.org. Haematologica is an open access journal. 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 International edition for the year 2021 are as following: Print edition

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 106, Issue 9: September 2021 About the Cover 2297

Images from the Haematologica Atlas of Hematologic Cytology: primary myelofibrosis, micromegakaryocytic transformation Rosangela Invernizzi https://doi.org/10.3324/haematol.2021.279315

Editorials 2298

Preventing central nervous system spread in diffuse large B-cell lymphoma – novel approaches needed Mark Roschewski https://doi.org/10.3324/haematol.2021.278559

2300

All in the family: back-to-back kinase inhibitors for the treatment of chronic lymphocytic leukemia Meghan C. Thompson et al. https://doi.org/10.3324/haematol.2021.278535

2302

Do we need more genome wide association studies? Stephan Menzel https://doi.org/10.3324/haematol.2021.278642

Review Articles 2304

Towards manufactured red blood cells for the treatment of inherited anemia Stephanie Pellegrin et al. https://doi.org/10.3324/haematol.2020.268847

2312

Targeting the tumor microenvironment in chronic lymphocytic leukemia Rebecka Svanberg et al. https://doi.org/10.3324/haematol.2020.268037

Articles Acute Myeloid Leukemia 2325 Monitoring of clonal evolution of acute myeloid leukemia identifies the leukemia subtype, clinical outcome and potential new drug targets for post-remission strategies or relapse Esther Onecha et al. https://doi.org/10.3324/haematol.2020.254623

Chronic Lymphocytic Leukemia 2334 Three-dimensional co-culture model of chronic lymphocytic leukemia bone marrow microenvironment predicts patient-specific response to mobilizing agents Federica Barbaglio et al. https://doi.org/10.3324/haematol.2020.248112

2345

2354

Zanubrutinib monotherapy for patients with treatment-naïve chronic lymphocytic leukemia and 17p deletion Constantine S. Tam et al. https://doi.org/10.3324/haematol.2020.259432

2364

Phase II study of acalabrutinib in ibrutinib-intolerant patients with relapsed/refractory chronic lymphocytic leukemia Kerry A. Rogers et al. https://doi.org/10.3324/haematol.2020.272500

Haematologica 2021; vol. 106 no. 9 - September 2021 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

Complications in Hematology 2374 Elastography improves accuracy of early hepato-biliary complications diagnosis after allogeneic stem cell transplantation Pierre-Edouard Debureaux et al. https://doi.org/10.3324/haematol.2019.245407

Myeloproliferative Disorders 2384 Altered T-cell subset repertoire affects treatment outcome of patients with myelofibrosis Ivo Veletic et al. https://doi.org/10.3324/haematol.2020.249441

2397

Dynamics of mutations in patients with essential thrombocythemia treated with imetelstat Elisabeth Oppliger Leibundgut et al. https://doi.org/10.3324/haematol.2020.252817

Non-Hodgkin Lymphoma 2405 A three-gene signature based on MYC, BCL-2 and NFKBIA improves risk stratification in diffuse large B-cell lymphoma Enrico Derenzini et al. https://doi.org/10.3324/haematol.2019.236455

2417

2427

Genetic manipulation of primary human natural killer cells to investigate the functional and oncogenic roles of PRDM1 Gehong Dong et al. https://doi.org/10.3324/haematol.2020.254276

Platelet Biology & its Disorders 2439 CAMT-MPL: congenital amegakaryocytic thrombocytopenia caused by MPL mutations - heterogeneity of a monogenic disorder a comprehensive analysis of 56 patients Manuela Germeshausen and Matthias Ballmaier https://doi.org/10.3324/haematol.2020.257972

2449

Red Cell Biology & its Disorders 2458 Effect of HBB genotype on survival in a cohort of transfusion-dependent thalassemia patients in Cyprus Petros Kountouris et al. https://doi.org/10.3324/haematol.2020.260224

2469

Interleukin-1 receptor inhibition reduces stroke size in a murine model of sickle cell disease Jessica Venugopal et al. https://doi.org/10.3324/haematol.2020.252395

2478

Oxidative stress activates red cell adhesion to laminin in sickle cell disease Maria Alejandra Lizarralde-Iragorri et al. https://doi.org/10.3324/haematol.2020.261586

Letters to the Editor 2489

Survival and causes of death in 2,033 patients with non-transfusion-dependent β-thalassemia Khaled M. Musallam et al. https://doi.org/10.3324/haematol.2021.278684

Haematologica 2021; vol. 106 no. 9 - September 2021 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation 2493

Circulating histones play a central role in COVID-19-associated coagulopathy and mortality Rebecca J. Shaw et al. https://doi.org/10.3324/haematol.2021.278492

2499

Heritability and association with distinct genetic loci of erythropoietin levels in the general population Tanguy Corre, et al. https://doi.org/10.3324/haematol.2021.278389

2502

Prognostic impact of soluble CD163 in patients with diffuse large B-cell lymphoma Heli Vajavaara et al. https://doi.org/10.3324/haematol.2020.278182

2507

Dual intracellular targeting by ruxolitinib and the Mcl-1 inhibitor S63845 in interleukin-6-dependent myeloma cells blocks in vivo tumor growth Renate Burger et al. https://doi.org/10.3324/haematol.2020.276865

2511

Clinical diagnostic value of telomere length measurement in inherited bone marrow failure syndromes Shunsuke Miwata et al. https://doi.org/10.3324/haematol.2021.278334

2516

Synergistic interaction between HDAC and MCL-1 inhibitors through downregulation of BCL-XL in multiple myeloma Anja Schneller et al. https://doi.org/10.3324/haematol.2020.277152

2522

Myeloma natural killer cells are exhausted and have impaired regulation of activation Criselle D’Souza et al. https://doi.org/10.3324/haematol.2020.277525

2527

Targeting BRD4 in acute myeloid leukemia with partial tandem duplication of the MLL gene Marius Bill et al. https://doi.org/10.3324/haematol.2020.271627

Case Report 2533

Clinical genomic profiling of novel grey zone lymphoma paired lesions with sequential central nervous system involvement in two adolescent patients Cagla Yasa-Benkli et al. https://doi.org/10.3324/haematol.2021.278936

Haematologica 2021; vol. 106 no. 9 - September 2021 http://www.haematologica.org/

ABOUT THE COVER Images from the Haematologica Atlas of Hematologic Cytology: primary myelofibrosis, micromegakaryocytic transformation Rosangela Invernizzi1,2 1

University of Pavia and 2IRCCS Policlinico San Matteo Foundation, Pavia, Italy

E-mail: ROSANGELA INVERNIZZI - rosangela.invernizzi@unipv.it doi:10.3324/haematol.2021.279315

pproximately 15% of patients with primary myelofibrosis (PMF) develop terminal blast crisis, usually either myeloblastic or myelomonocytic, whereas the presence in the blood of immature cells exclusively of the megakaryocytic type is uncommon. In this case, diagnosed as PMF for many years, a peripheral blood smear reveals a group of cells with an eccentric nucleus, condensed chromatin, no nucleoli, basophilic cytoplasm and cytoplasmic protrusions. A granulocyte precursor and an undifferentiated blast can also be seen. Platelets are often giant and sometimes agranular (top image). Immunocytochemistry using an anti-CD61 monoclonal antibody demonstrates the megakaryocytic nature of the small mononuclear cells and of the blasts suggesting a diagnosis of micromegakaryocytic leukemia as transformation of PMF. Platelets are strongly positive to the immunoalkaline-phosphatase reaction (bottom image).1 Disclosures No conflicts of interest to disclose.

Reference 1. Invernizzi R. Myeloproliferative neoplasms. Haematologica. 2020;105(Suppl 1):49-59.

haematologica | 2021; 106(9)

2297

EDITORIALS Preventing central nervous system spread in diffuse large B-cell lymphoma – novel approaches needed Mark Roschewski Lymphoid Malignancies Branch, Center for Cancer Research, National Institutes of Health, Bethesda, MD, USA E-mail: MARK ROSCHEWSKI - mark.roschewski@nih.gov doi:10.3324/haematol.2021.278559

number of retrospective datasets have addressed the controversial topic of chemotherapy as central nervous system (CNS) prophylaxis during frontline management of diffuse large B-cell lymphoma (DLBCL).1-5 Despite the fact that CNS spread is a feared and often terminal complication of DLBCL, there is not a broad consensus regarding which patients should receive CNS prophylaxis or the most effective method of delivery. Overall, the incidence of CNS relapse across all subsets of DLBCL is only about 5%, but some clinical risk factors, including the involvement of specific anatomic sites, are associated with a significantly higher rate of CNS spread. Furthermore, we are beginning to uncover the biological basis for DLBCL involving the CNS as specific genetic subtypes demonstrate an inherently higher rate of CNS tropism.6-8 The CNS International Prognostic Index (CNS-IPI) is a commonly used risk model that stratifies patients into risk categories;9 combining this model with the

cell-of-origin phenotype may improve selection of patients.10 However, even the most robust predictive models cannot overcome the fundamental problem that the chemotherapy agents most effective for the cure of systemic DLBCL do not reliably penetrate the blood-brain barrier (Figure 1).11 Conversely, methotrexate, which reliably penetrates the CNS, is not highly effective for DLBCL. The most commonly used prophylactic strategy is repeated intrathecal injections of chemotherapy such as methotrexate during frontline therapy, but since brain parenchymal sites are the commonest site of CNS relapse, some advocate the use of deeply penetrant drugs such as high-dose methotrexate.3,12 No randomized prospective study has directly addressed this specific issue and, as a result, practice patterns rely on consensus guidelines and vary widely across institutions and individual providers.13 In essence, the debate about optimal delivery methods is a “race to the bottom” that compares two strate-

Figure 1. A subset of patients with diffuse large B-cell lymphoma are at high-risk of disease spread to the central nervous system and are often treated with chemotherapy prophylaxis. A critical barrier to effective central nervous system (CNS) prophylaxis is the blood-brain barrier (1) which limits the entry of the chemotherapy agents that are most effective for systemic diffuse large B-cell lymphoma (DLBCL) (2). Current therapeutic options for CNS chemotherapy prophylaxis are systemic chemotherapy (3) or intrathecal chemotherapy (4) which are both limited in efficacy and increase toxicity. Novel small molecule inhibitors that effectively penetrate the blood-brain barrier are being tested in DLBCL involving the CNS and may improve treatment options.

2298

haematologica | 2021; 106(9)

Editorials

gies that do not adequately address the clinical problem. Although retrospective series are effective at generating hypotheses or identifying specific issues that warrant further study, it is nearly impossible to control for all the permutations of approaches to CNS prophylaxis as there is truly no standard approach. The best available data suggest that the most common approach to CNS prophylaxis involves repeated intrathecal injections of methotrexate or cytarabine during frontline therapy, while only a significant minority of patients receive high-dose methotrexate at a median dose of 3.5 g/m2 either during frontline therapy or immediately following. Notably, patients who receive high-dose methotrexate may also receive concomitant intrathecal chemotherapy which blurs this arbitrary distinction. Furthermore, most datasets demonstrate that nearly half of patients considered high-risk receive no form of CNS prophylaxis. This observation highlights that patient-related factors, such as age and perceived ability to tolerate treatment-related toxicity, greatly influence treatment decisions beyond prognostic scores and/or involvement of extranodal sites. Since all forms of CNS prophylaxis have clinically meaningful toxicities, this underscores the fact that an important limitation of all available datasets is selection bias. Finally, no form of CNS prophylaxis is universally effective and the rate of CNS relapse in patients who receive prophylaxis is typically about 5% after 2 to 3 years of follow-up. In recognition that CNS relapses may be late events, the actual risk reduction of any form of CNS prophylaxis with chemotherapy is likely modest at best and currently employed strategies may simply delay the timing of CNS recurrence.14 The risk of CNS involvement is not equally distributed across all subsets of DLBCL, however, which may allow for precision medicine strategies. In fact, DLBCL is not a single disease but comprises a spectrum of aggressive lymphomas with striking underlying genetic diversity. The current classification system recognizes both activated B-cell (ABC) DLBCL and germinal center B-cell (GCB) DLBCL as distinct molecular subtypes and introduced a new entity, high-grade B-cell lymphoma, defined by the presence of MYC and BCL2 and/or BCL6 rearrangements (HGBCL-DH/TH).15 Indeed, patients with ABC (nonGCB) DLBCL subtype have an overall higher risk of CNS relapse.10 Furthermore, recent multiplatform genomic profiling studies have identified genetic subtypes of DLBCL with shared genetic features.6,7 One genetic subtype, MCD, is characterized by frequent co-occurrence of MYD88L265P and CD79B mutations, prominent immuneediting features, and PIM1 mutations.6 These tumors occur almost exclusively within ABC DLBCL and frequently involve extranodal sites including the testes, breast and CNS.6 It is noteworthy that a separate multiplatform genomic profiling study described a very similar subtype termed Cluster 5 (C5) tumors which were characterized by MYD88L265P and CD79B mutations, gain of 18q, and PIM1 mutations and also exhibited a propensity for extranodal sites, including the CNS and testes.7 Furthermore, a recently reported series of 26 cases of secondary DLBCL of the CNS confirmed a higher prevalence of MCD subtype than that observed in a reference cohort of relapsed DLBCL without CNS spread (38% vs. 8%,

haematologica | 2021; 106(9)

P=0.003).16 In this study, the majority of other DLBCL cases with CNS spread were either HGBCL-DH/TH or associated with TP53 mutations. Another recent study investigated the genomic predictors of CNS relapse in 82 cases of primary testicular DLBCL which has a strong predilection for CNS spread.17 The authors identified BCL6 and/or PDL1 or PDL2 rearrangements as the most common genetic aberrations associated with CNS relapse after treatment for primary testicular DLBCL. Although the precise mechanisms by which various genetic aberrations cooperate to promote CNS spread remain undetermined, these results suggest that a more nuanced understanding of the molecular biology of DLBCL involving the CNS may lead to novel therapeutic targets. In order to improve clinical outcomes, however, novel therapies with demonstrable efficacy within genetically defined subtypes will be necessary. Multiple clinical studies have reported impressive clinical activity of the Bruton tyrosine kinase (BTK) inhibitor ibrutinib and ibrutinib-based regimens in DLBCL involving the CNS, including patients who are refractory to chemotherapy.18,19 Even though a randomized phase III study did not show an overall benefit from adding ibrutinib to R-CHOP as part of frontline therapy for non-GCB DLBCL, certain subsets appeared to have improved outcomes.20 Further studies of BTK inhibitors with R-CHOP are currently ongoing and should provide additional data regarding rates of CNS relapse. In addition, the immunomodulatory agent lenalidomide has demonstrated good clinical activity and favorable safety in DLBCL involving the CNS.21 Lenalidomide has also been added to R-CHOP as part of frontline therapy for DLBCL which may benefit certain subsets of DLBCL.22 The currently available data do not support the use of either ibrutinib or lenalidomide as part of frontline therapy to prevent CNS spread of DLBCL, but all clinical trials testing novel agents should report CNS-specific outcomes within genetically defined subtypes. In summary, chemotherapy as CNS prophylaxis is not universally effective no matter what the delivery method, and the prevention and treatment of CNS relapse remain unmet clinical needs in the management of DLBCL. Penetration of the blood-brain barrier is an important consideration, but improved therapies will be required to overcome intrinsic chemotherapy resistance. A nuanced mechanistic understanding of targetable pathways underpinning DLBCL involving the CNS has led to novel targeted agents and immunotherapy approaches that demonstrate promising clinical activity and good CNS penetrance. Novel agents that target oncogenic drivers based on the underlying biology of DLBCL subtypes may ultimately prove to be the most effective way to prevent and/or treat CNS recurrence. Disclosures No conflicts of interest to disclose.

References 1. Puckrin R, El Darsa H, Ghosh S, Peters A, Stewart DA. Lack of effectiveness of intravenous high-dose methotrexate for prevention of CNS relapse in patients with high-risk DLBCL: a retrospective analy-

2299

Editorials

sis from Alberta, Canada. Blood. 2020;136(Suppl 1):26-27. 2. Orellana-Noia VM, Reed DR, Sen JM, et al. CNS prophylaxis during front-line therapy in aggressive non-Hodgkin lymphomas: realworld outcomes and practice atterns from 19 US academic nstitutions. Blood. 2020;136(Suppl 1):27-28. 3. Eyre TA, Djebbari F, Kirkwood AA, Collins GP. Efficacy of central nervous system prophylaxis with stand-alone intrathecal chemotherapy in diffuse large B-cell lymphoma patients treated with anthracycline-based chemotherapy in the rituximab era: a systematic review. Haematologica. 2020;105(7):1914-1924. 4. Eyre TA, Kirkwood AA, Wolf J, et al. Stand-alone intrathecal central nervous system (CNS) prophylaxis provide unclear benefit in reducing CNS relapse risk in elderly DLBCL patients treated with RCHOP and is associated increased infection-related toxicity. Br J Haematol. 2019;187(2):185-194. 5. Cheah CY, Herbert KE, O'Rourke K, et al. A multicentre retrospective comparison of central nervous system prophylaxis strategies among patients with high-risk diffuse large B-cell lymphoma. Br J Cancer. 2014;111(6):1072-1079. 6. Schmitz R, Wright GW, Huang DW, et al. Genetics and pathogenesis of diffuse large B-cell lymphoma. N Engl J Med. 2018;378(15):13961407. 7. 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. 8. Wright GW, Huang DW, Phelan JD, et al. A probabilistic classification tool for genetic subtypes of diffuse large B cell lymphoma with therapeutic implications. Cancer Cell. 2020;37(4):551-568.e14. 9. Schmitz N, Zeynalova S, Nickelsen M, et al. CNS International Prognostic Index: a risk model for CNS relapse in patients with diffuse large B-cell lymphoma treated with R-CHOP. J Clin Oncol. 2016;34(26):3150-3156. 10. Klanova M, Sehn LH, Bence-Bruckler I, et al. Integration of cell of origin into the clinical CNS International Prognostic Index improves CNS relapse prediction in DLBCL. Blood. 2019;133(9):919-926. 11. Arvanitis CD, Ferraro GB, Jain RK. The blood-brain barrier and blood-tumour barrier in brain tumours and metastases. Nat Rev Cancer. 2020;20(1):26-41. 12. Abramson JS, Hellmann M, Barnes JA, et al. Intravenous methotrexate as central nervous system (CNS) prophylaxis is associated with a

low risk of CNS recurrence in high-risk patients with diffuse large Bcell lymphoma. Cancer. 2010;116(18):4283-4290. 13. McKay P, Wilson MR, Chaganti S, Smith J, Fox CP, Cwynarski K; British Society of Haematology. The prevention of central nervous system relapse in diffuse large B-cell lymphoma: a British Society for Haematology good practice paper. Br J Haematol. 2020;190(5):708714. 14. Ambady P, Holdhoff M, Bonekamp D, Wong F, Grossman SA. Late relapses in primary CNS lymphoma after complete remissions with high-dose methotrexate monotherapy. CNS Oncol. 2015;4(6):393398. 15. 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. 16. Ollila TA, Kurt H, Waroich J, et al. Genomic subtypes may predict the risk of central nervous system recurrence in diffuse large B-cell lymphoma. Blood. 2021;137(8):1120-1124. 17. Twa DDW, Lee DG, Tan KL, et al. Genomic predictors of central nervous system relapse in primary testicular diffuse large B-cell lymphoma (DLBCL). Blood. 2021;137(9):1256-1259. 18. Lionakis MS, Dunleavy K, Roschewski M, et al. Inhibition of B cell receptor signaling by ibrutinib in primary CNS lymphoma. Cancer Cell. 2017;31(6):833-843.e5. 19. Grommes C, Pastore A, Palaskas N, et al. Ibrutinib unmasks critical role of Bruton tyrosine kinase in primary CNS lymphoma. Cancer Discov. 2017;7(9):1018-1029. 20. Younes A, Sehn LH, Johnson P, et al. Randomized phase III trial of ibrutinib and rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone in non-germinal center B-cell diffuse large Bcell lymphoma. J Clin Oncol. 2019;37(15):1285-1295. 21. Ghesquieres H, Chevrier M, Laadhari M, et al. Lenalidomide in combination with intravenous rituximab (REVRI) in relapsed/refractory primary CNS lymphoma or primary intraocular lymphoma: a multicenter prospective 'proof of concept' phase II study of the French Oculo-Cerebral Lymphoma (LOC) Network and the Lymphoma Study Association (LYSA). Ann Oncol. 2019;30(4):621-628. 22. Nowakowski GS, Hong F, Scott DW, et al. Addition of lenalidomide to R-CHOP improves outcomes in newly diagnosed diffuse large Bcell lymphoma in a randomized phase II US Intergroup Study ECOG-ACRIN E1412. J Clin Oncol. 2021;39(12):1329-1338.

All in the family: back-to-back kinase inhibitors for the treatment of chronic lymphocytic leukemia Meghan C. Thompson, Lindsey E. Roeker and Anthony R. Mato Leukemia Service, Memorial Sloan Kettering Cancer Center, New York, NY, USA E-mail: ANTHONY R. MATO - matoa@mskcc.org doi:10.3324/haematol.2021.278535

n this issue of Haematologica, Rogers et al. address a key sequencing question in the management of chronic lymphocytic leukemia (CLL) by reporting the results of the largest prospective clinical trial evaluating acalabrutinib for the treatment of CLL following intolerance to ibrutinib.1 While the Bruton tyrosine kinase (BTK) inhibitor ibrutinib has led to a paradigmatic shift in the treatment of CLL away from chemoimmunotherapy, high rates of ibrutinib discontinuation remain a major problem. Real-world evidence and long-term follow-up from clinical trials of ibrutinib have established that drug intolerance due to toxicity, rather than progressive CLL, is the most common reason for discontinuation of ibrutinib treatment.2-4 Real-world data from 616 CLL patients treated with ibrutinib in clinical practice documented that 41% of patients discontinued ibrutinib (median follow-

2300

up 17 months), and more than half of all discontinuations were due to toxicity.2 Real-world evidence from the UK documents high rates of ibrutinib discontinuation due to reasons other than disease progression (17.5%).3 Furthermore, similar patterns have emerged with longer follow-up data from clinical trials, with more patients discontinuing ibrutinib due to toxicity than because of CLL progression. At 5 years of follow-up of the RESONATE-2 trial of ibrutinib for initial treatment of CLL, 41% of patients had discontinued ibrutinib therapy, with a 21% discontinuation rate due to adverse events including atrial fibrillation.4 Furthermore, in a pooled analysis of CLL patients treated with ibrutinib on three randomized phase III studies, 11% of patients permanently discontinued ibrutinib due to adverse events and 13% of patients required dose reductions due to adverse events, highlighting the significant impact of adverse events while on

haematologica | 2021; 106(9)

Editorials

Figure 1. A proposed sequencing algorithm for treatment of chronic lymphocytic leukemia following discontinuation of ibrutinib due to intolerance. CLL: chronic lymphocytic leukemia; iwCLL: International Workshop on CLL; BTKi: Bruton tyrosine kinase inhibitor; PI3K: phosphoinositide 3-kinase.

treatment with ibrutinib.5 These studies clearly established that intolerance to ibrutinib is a common scenario encountered in clinical practice, which may limit the clinical benefit of this drug that has been largely studied as a continuous therapy. Given the clinical efficacy of BTK inhibition in CLL, for patients who discontinue a BTK inhiibitor due to intolerance, an important question is whether treatment with an alternative kinase inhibitor is an acceptable treatment option. This is particularly relevant given the development of more selective BTK inhibitors with fewer off-target effects. Newer BTK inhibitors include approved therapies such as acalabrutinib, as well as emerging covalent and non-covalent BTK inhibitors in clinical development (zanubrutinib, LOXO-305, ARQ-351). Previously, Awan et al. addressed this key question by conducting a small cohort study of acalabrutinib treatment for patients who discontinued ibrutinib due to intolerance (defined by the investigator’s discretion).6 In this study of 33 patients, the efficacy of acalabrutinib following ibrutinib was high (overall response rate 76%) with only 9% of patients discontinuing acalabrutinib due to an adverse event.6 However, this study examined only a small number of patients and lacked an objective definition of ibrutinib intolerance. The study by Rogers et al. is the first prospectively designed study to answer this important sequencing question.1 Intolerance was defined as discontinuation of ibrutinib due to either persistent/recurrent grade 2 adverse events despite dose modification or interruption or persistent grade 3/4 adverse events. Sixty patients with relapsed and/or refractory CLL were treated with acalabrutinib (median number of prior therapies 2) with a

haematologica | 2021; 106(9)

prior median duration of ibrutinib therapy of 5.7 months. Overall, the approach was well-tolerated, with the most common adverse events being diarrhea (53%), headache (42%) and contusion (40%). Only 40% of patients had ibrutinib-related intolerance adverse events, and 67% of events were lower grade with acalabrutinib than with ibrutinib; only one adverse event (increased levels of liver enzymes) occurred at a higher grade with acalabrutinib treatment than with ibrutinib treatment. Notably, more patients discontinued acalabrutinib because of CLL progression (23%) than because of adverse events (17%). Acalabrutinib following discontinuation of ibrutinib for intolerance was efficacious, with an overall response rate of 73% and a 24-month estimated progression-free survival of 72% (median follow-up, 35 months). It should be noted that the majority (94%) of patients with available pre-treatment sequencing data did not have BTK or PLCG2 mutations prior to initiating treatment with acalabrutinib.1 In addition to the work presented by Rogers et al., two additional recent studies have also shown that treatment of CLL with an alternative kinase inhibitor following ibrutinib intolerance is safe and efficacious.7,8 A phase II study examined the phosphoinositide 3-kinase (PI3K) inhibitor umbralisib in 51 patients with relapsed/refractory CLL who were intolerant to prior BTK inhibition (n=44) or PI3K inhibition (n=7) and showed a median progression-free survival of 23.5 months, with the majority (58%) of patients remaining on umbralisib for longer than on their prior kinase inhibitor therapy.7 Additionally, LOXO-305 (pirtobrutinib), a novel, highly selective, noncovalent BTK inhibitor showed a favorable safety profile in 170 patients with CLL/small lymphocytic leukemia, of

2301

Editorials

whom 86% had received prior treatment with a BTK inhibitor, with 33% discontinuing the prior BTK inhibitor due to reasons other than progressive CLL.8 Furthermore, LOXO-305 had promising efficacy in this heavily pretreated population with an overall response rate of 62% in 121 efficacy-evaluable patients with CLL/small lymphocytic leukemia who had previously been treated with a BTK inhibitor.8 Taken together, these studies challenge the traditional sequencing paradigm of switching drug classes in the setting of CLL therapy discontinuation for intolerance. In Figure 1, we propose a sequencing algorithm incorporating the new data from Rogers et al. While venetoclax is an acceptable option in the setting of intolerance to BTK inhiitors,9 CLL remains an incurable, chronic disease and there is a strong scientific rationale for maximizing clinical benefit from each drug class prior to exposing patients to the selective pressures of another therapeutic class. In the case of the common problem of intolerance to ibrutinib it is best to keep the solution “all in the (BTK inhibitor) family.” Disclosures The authors have no disclosures to make regarding this editorial. With regard to work outside this publication, MCT has received honoraria from MJH Life Sciences, VJHemOnc, and Curio Science. LER has received research funding from the American Society of Hematology and Pfizer; has minority ownership interest in Abbott Laboratories; provides consultancy services for AbbVie, AstraZeneca, Pharmacyclics, the Vaniam group and, uncompensated, for Verastem. ARM has received grants and personal fees from and is a data safety monitoring board member for TG Therapeutics; has received grants and personal fees from Loxo Oncology (a wholly owned subsidiary of Eli Lilly), Genentech, AbbVie, AstraZeneca, Adaptive,

Pharmacyclics, and Curio Sciences; has received grants from Sunesis, Regeneron, Pfizer, Aprea, Aptose, and DTRM; has received non-financial support from the NCCN, CLL society, and Lymphoma Research Foundation; and has received grants from and sat on a steering committee for Verastem. Contributions MCT and ARM drafted the manuscript. MCT, LER and ARM provided feedback and edited the manuscript.

References 1. Rogers KA, Thompson PA, Allan JN, et al. Phase II study of acalabrutinib in ibrutinib-intolerant patients with relapsed/refractory chronic lymphocytic leukemia. Haematologica. 2021;106(9):2364-2373. 2. Mato AR, Nabhan C, Thompson MC, et al. Toxicities and outcomes of 616 ibrutinib-treated patients in the United States: a real-world analysis. Haematologica. 2018;103(5):874-879. 3. UK CLL Forum. Ibrutinib for relapsed/refractory chronic lymphocytic leukemia: a UK and Ireland analysis of outcomes in 315 patients. Haematologica. 2016;101(12):1563-1572. 4. Burger JA, Barr PM, Robak T, et al. Long-term efficacy and safety of first-line 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. 5. Coutre SE, Byrd JC, Hillmen P, et al. Long-term safety of single-agent ibrutinib in patients with chronic lymphocytic leukemia in 3 pivotal studies. Blood Adv. 2019;3(12):1799-1807. 6. Awan FT, Schuh A, Brown JR, et al. Acalabrutinib monotherapy in patients with chronic lymphocytic leukemia who are intolerant to ibrutinib. Blood Adv. 2019;3(9):1553-1562. 7. Mato AR, Ghosh N, Schuster SJ, et al. Phase 2 study of the safety and efficacy of umbralisib in patients with CLL who are intolerant to BTK or PI3Kδ inhibitor therapy. BBlood. 2021;137(20):2817-2826. 8. Mato AR, Shah NN, Jurczak W. Pirtobrutinib in relapsed or refractory B-cell malignancies (BRUIN): a phase 1/2 study. Lancet. 2021;397(10277):892-901. 9. Jones JA, Mato AR, Wierda WG, et al. Venetoclax for chronic lymphocytic leukaemia progressing after ibrutinib: an interim analysis of a multicentre, open-label, phase 2 trial. Lancet Oncol. 2018;19(1):65-75.

Do we need more genome wide association studies? Stephan Menzel Red Cell Research Unit, King's College London, London, UK E-mail: STEPHAN MENZEL - stephan.menzel@kcl.ac.uk doi:10.3324/haematol.2021.278642

uch of the individual biological traits we have, of what we look like, of our physical and mental abilities, of our risk to suffer from the noncommunicable diseases that will ultimately end our lives, is encoded in the genetic ‘background’, consisting of millions of single-nucleotide polymorphisms (SNP) and other common sequence variants that each have minute functional effects on regulatory sequences within our genome. Genome-wide association studies (GWAS) are the tool of choice to make the connection between commonvariant genotype data, collected either through genome sequencing or with genotyping arrays (‘chips’), and human phenotype. In its simplest form, GWAS compare the frequency for each of of thousands or millions of

2302

common genetic variants between groups of patients and controls, thus identifying genetic risk factors for the diseases studied this way. There are limits to what the traditional GWAS approach can achieve. Suffocating type-I error rates arising from the analysis of millions of genetic variants make it necessary to assemble very large groups of patients and controls, but even then, only the strongest genetic risk factors can be identified with meaningful certainty. Even so, finding this initial set of genetic factors has significantly enhanced our understanding of pathways leading to common disease or shaping healthrelevant physiological traits. With the majority of disease risk factors still hidden, however, it is presently impossible to assemble enough genetic information to

haematologica | 2021; 106(9)

Editorials

make clinically relevant predictions for any given person. The gambit for detecting additional disease-relevant genetic factors has been to the assemble ever larger subject cohorts, reaching hundreds of thousands of participants for some conditions. Still the majority of the disease-relevant genetic background remains untouchable,1 with thousands of weak-effect alleles still hidden, trapped in what is termed the ‘missing heritability’. General frustration with the GWAS approach is prevalent among researchers. Corre et al.2, on page 2499 of this issue, report a study of the type that offers a way out of this trap. The authors present a quantitative-trait association study, comparing circulating levels of the hormone erythropoietin with the genotype of a genome-wide SNP set. In contrast to the original case-control setup, such quantitative-trait GWAS offer crucial advantages. They allow ‘drilling down’ into the pathways underlying biological characters and disease pathogenesis, thereby reducing complexity and increasing the signal-to-noise ratio of genetic analysis. Quantitative-trait GWAS can utilise various large subject cohorts assembled for other studies, such as groups of patients or population samples, if the parameter of interest or related biological traits have been recorded. Loci and variants discovered in quantitative-trait studies can subsequently be evaluated with more complex traits, such as disease risk. Several large GWAS with red blood cell traits have been conducted and the genes identified have contributed to our understanding of anemia. This has been complemented with GWAS investigation of circulating erythropoietin levels, the main hormonal regulator of the system. Unsurprisingly, the set of genes detected overlap between the two approaches, e.g., HBS1L-MYB, which is a quantitative-trait locus (QTL) for various redblood cell traits (HbF%, MCV, MCH, RBC), has also shown strong association with erythropoietin levels in a 2018 Dutch population study with 6,777 participants (Grote Beverborg et al.3). The present study of Corre et al., while smaller, has provided confirmation of HBS1LMYB as an erythropoietin locus and the joint analysis of both cohorts has yielded a significance level of P<10-22. Heritability of erythropoietin levels was found to be higher than in another previous study (by Wang et al.4) and the set of genes detected is also somewhat different. In quantitative-trait studies, heritability estimates and the spectrum of loci detected is fluid, and specific outcomes depend on peculiarities of subject recruitment, trait assay method, and measurement routines. However, with multiple cohorts available to study a given parameter and its related traits, a network of quantitative-trait studies can be built that, together with

haematologica | 2021; 106(9)

knowledge gained from laboratory-experimental studies, paints a picture of functional and genetic architecture of the investigated tissue system and any disease risk connected with it. The most intriguing outcome of the present paper is the detection of a putative new QTL for erythropoietin levels on chromosome 15, with evidence for trait association (P=1.05x10-7) just short of the acknowledged level of genome-wide statistical significance. Corre et al. have started to harness data from GWAS performed with blood cell parameters in an attempt to confirm the validity of this preliminary result: in the UK Biobank study variants at this locus were found associated with erythroid traits, e.g., with hemoglobin concentration and reticulocyte count at P<10-5, but it is not clear why Corre and colleagues have not presented a ‘look up’ of their new locus in the erythropoietin GWAS dataset of Grote Beverborg et al.3 Confirmation of initial, ‘suggestive’, findings in a set of related studies must be integral part to any QTL GWAS, thus harnessing the full power of this approach. Obtaining data for that from colleagues in the field is usually straightforward. It will be fascinating to see how this story develops following publication in Haematologica. The possibility of uncovering a new mechanism regulating oxygen transport capacity through erythropoietin is tantalising. In general, present efforts to build large population cohorts of extensively phenotyped individuals with complementary genotype data (genome array or sequence) will generate increasingly powerful datasets allowing to decipher our genetic blueprint and help to fulfil the promise of genetics for the improvement of human health. Dislcosures No conflicts of interest to disclose. Funding SM is presently supported by an MRC project grant to investigate the genetic determination of fetal-hemoglobin levels in sickle cell disease.

References 1. Goldstein DB. Common genetic variation and human traits. N Engl J Med. 2009;360(17):1696-1698. 2. Corre T, Ponte B, Pivin E, et al. Heritability and association with distinct genetic loci of erythropoietin levels in the general population. Haematologica. 2021;106(8):2499-2501. 3. Grote Beverborg N, Verweij N, Klip IT, et al. Erythropoietin in the general population: reference ranges and clinical, biochemical and genetic correlates. PLoS One. 2015;10(4):e0125215. 4. Wang Y, Nudel R, Benros ME, et al. Genome-wide association study identifies 16 genomic regions associated with circulating cytokines at birth. PLoS Genet. 2020;16(11):e1009163.

2303

REVIEW ARTICLE Ferrata Storti Foundation

Towards manufactured red blood cells for the treatment of inherited anemia Stephanie Pellegrin,1,2 Charlotte E. Severn1,2 and Ashley M. Toye1,2,3 School of Biochemistry, Biomedical Sciences Building; 2National Institute for Health Research (NIHR) Blood and Transplant Research Unit in Red Blood Cell Products, University of Bristol and 3Bristol Institute of Transfusion Sciences, NHSBT Filton, Bristol, UK

Haematologica 2021 Volume 106(9):2304-2311

ABSTRACT

atients with inherited anemia and hemoglobinopathies (such as sickle cell disease and β-thalassemia) are treated with red blood cell (RBC) transfusions to alleviate their symptoms. Some of these patients may have rare blood group types or go on to develop alloimmune reactions, which can make it difficult to source compatible blood in the donor population. Laboratory-grown RBC represent a particularly attractive alternative which could satisfy an unmet clinical need. The challenge, however, is to produce - from a limited number of stem cells - the 2x1012 RBC required for a standard adult therapeutic dose. Encouraging progress has been made in RBC production from adult stem cells under good manufacturing practice. In 2011, the Douay group conducted a successful proof-of-principle mini-transfusion of autologous manufactured RBC in a single volunteer. In the UK, a trial is planned to assess whether manufactured RBC are equivalent to RBC produced naturally in donors, by testing an allogeneic mini-dose of laboratory-grown manufactured RBC in multiple volunteers. This review discusses recent progress in the erythroid culture field as well as opportunities for further scaling up of manufactured RBC production for transfusion practice.

Introduction Correspondence: ASHLEY TOYE ash.m.toye@bristol.ac.uk Received: February 2, 2021. Accepted: March 31, 2021. Pre-published: May 27, 2021. https://doi.org/10.3324/haematol.2020.268847

©2021 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.

2304

Red blood cell (RBC) transfusions are the mainstay treatment for anemic patients and are given routinely in hospitals around the world. Depending on the clinical presentation, blood transfusions are likely to be administered to improve the patients’ quality of life and are administered either intermittently when needed, or for extended periods, as in the case of 50-90% of patients suffering from myelodysplastic syndromes and for transfusion-dependent patients with inherited anemia and hemoglobinopathies (such as sickle cell disease and β-thalassemia). Chronic RBC transfusions introduce secondary complications which contribute to morbidity, due mainly to transfusion-induced iron overload and erythrocyte alloimmunization.1,2 It is particularly challenging for blood services to source RBC compatible for multiply alloimmunized patients with chronic transfusion-dependent anemia or rare blood types.3,4 Breakthroughs in the field of erythropoiesis research have led to the development of reproducible protocols that can yield large numbers of cultured human reticulocytes, often referred to as laboratory-grown or cultured red blood cells (cRBC). In addition to being an excellent model system for exploring human erythropoiesis in health and disease, this work has laid the foundations for the interest in producing human cRBC for transfusion purposes or as a vehicle for red cell-based therapeutics. The production of cRBC from stem cells or other cellular sources (see below) may one day fill the unmet clinical need for transfusion-dependent patients, but only if the challenge of growing enough clinical grade RBC can be met. The term “manufactured RBC” (mRBC) refers to clinical grade cRBC grown under good manufacturing practises (GMP). Both cRBC and mRBC are in fact nascent RBC, known as reticulocytes5-7 and they are referred to only as mRBC within this manuscript from now on to avoid confusion. One added benefit of mRBC, compared to standard donor-derived RBC, is that they are a homogeneous population of immature RBC that should last the normal 120-day lifetime in the circulation. This

haematologica | 2021; 106(9)

Manufactured red blood cells

regenerative medicine product is anticipated to reduce transfusion frequency and the associated iron overload.6 Support for this idea comes from studies showing that transfusion of young RBC (also called neocytes) is beneficial to patients with inherited anemias, reducing iron overload and increasing the interval between transfusions.8-12 mRBC have been tested in immunocompromised mice models5,6,13 and non-human primates.13 Importantly, a proof-of-principle mini-transfusion of autologous mRBC has been conducted in a single volunteer, illustrating that mRBC can survive in the circulation and are safe for use in humans.5 The benefit of mRBC in allogeneic transfusions across multiple recipients still needs to be demonstrated before moving to patients. The commercial company Rubius Therapeutics has a business model built around producing novel mRBC for therapeutics. To date, Rubius has conducted one injection of mRBC engineered for treatment of phenylketonuria in a single patient (Rubius press release 12th March 202014), but no further specific information was released. In the UK, the National Health Service Blood and Transplant (England’s Blood Service) is intending to conduct a single-center, randomized, allogeneic, controlled, phase I, cross-over trial denominated RESTORE (Recovery and Survival of Stem Cell Originated Red Cells (ISRCTN:42886452 and EudraCT: 2017-00217838). This healthy volunteer trial has faced significant delays, most recently due to the COVID-19 pandemic but will, it is hoped, be carried out in the near future to assess the recovery and survival of a mini-dose of mRBC derived from CD34+ cells isolated from adult blood donors versus the standard RBC from the same donor. Laboratory-grown RBC offer the greatest potential in terms of sourcing rare blood groups for sickle cell and thalassemia patients with alloimmunity. It must be acknowledged however, that these patients also present the greatest challenge in terms of requirements for blood. Adult patients require multiple units of blood per month.15 Realistically, the first therapeutic use of mRBC is likely to take place in a pediatric setting or for red cell-based therapeutics, such as enzyme replacement therapies, as both these applications require smaller numbers of mRBC. There is also a need to determine the number of mRBC that represents a therapeutic dose. For adult patients, one unit of standard RBC is estimated to consist of approximately 2x1012 RBC which raises the hemoglobin of an average adult by 1 g/dL. For pediatric patients, doses are more variable as they depend on the weight of the patient but are lower than an adult dose. It should be noted that a proportion (5-10%) of standard RBC are lost within the first 24 hours after transfusion16 and this increases to 25% or more with blood storage time. Therefore, the actual number of RBC required to treat anemia is likely to be lower if the majority of the cells are nascent. Many excellent reviews have been written on mRBC and the prospect of using mRBC for transfusion.17-25 We therefore offer below a concise overview of the progress to date, highlighting the relevant issues and opportunities for optimizing and increasing the mRBC yield to an adult therapeutic dose.

Overview of the erythroid culture process The recapitulation of erythropoiesis using primary hematopoietic stem and progenitor cells (HSPC) ex vivo haematologica | 2021; 106(9)

requires specific combinations of cytokines and growth factors in order to first expand the HSPC, and then to direct lineage specification to ensure full differentiation to the reticulocyte stage (see Figure 1). Over the last 20 years, multiple laboratories have developed two-dimensional liquid culture systems that reproduce the process and stages of human erythropoiesis to generate reticulocytes. These include two to four stages, each characterized by the inclusion or omission of specific growth factors (Table 1). The general consensus is for the inclusion of a primary stage favoring HSPC expansion with interleukin-3 and stem cell factor, a secondary erythroblast expansion stage including stem cell factor and erythropoietin, followed by a terminal differentiation stage with erythropoietin. Notably, holotransferrin is included throughout the culture period. Some laboratories further modify the initialstage culture media by including, for example, thrombopoietin and fms-like tyrosine kinase 3 (Flt-3) to enhance stem cell proliferation13 and may also include glucocorticoids to increase expansion prior to differentiation.26 The more recent culture protocols listed in Table 1 have been undertaken at considerably larger scale (i.e., at least 1 L), with some reports of successful generation of large numbers of reticulocytes. The challenge for the field is to increase the production even further to generate the equivalent of a therapeutically useful adult dose.

Starting material The studies reporting the highest yields all use HSPC specifically isolated from cord blood,13,27 mobilized5 or standard peripheral blood6,28 (Table 1). Another option is to use the whole peripheral blood mononuclear cell (PBMNC) component for production of mRBC, thereby omitting the expensive step of CD34+ isolation.7,29 As well as reducing costs, the use of PBMNC as the starting material under the correct culture conditions can contribute towards increasing the yield of mRBC. Indeed, PBMNC include all cells with erythroid lineage potential, some of which are CD34- that can enhance culture yield.29 PBMNC also include CD14+ cells that might act as helper/feeder cells that can limit the cell death of erythroid progenitors during the first few days of culture when volumes are still small and cells are kept in static tissue culture flasks or dishes.30,31 We highlight that there are other sources that are gaining traction, including immortalized pluripotent stem cells and immortalized cell lines,32,33 which can be used to differentiate to reticulocytes, but we will not discuss these here because these have not yet been grown at scale. These sustainable cellular sources have great potential for continual blood production once the technical challenges of growing them have been circumvented and are likely to comprise a second wave of blood products after stem cellderived mRBC.

Natural donor variation and yields The genetic makeup of the donor-derived cellular starting material has long been recognized to have an impact on yield, which is problematic when trying to consistently produce a high number of mRBC using random donors.34 This variation could be due to the number of HSPC pres2305

S. Pellegrin et al.

ent per volume of blood, which is highly variable between donors, or down to subtle variations in genetic factors that influence how HSPC and erythroid progenitors proliferate, enucleate and/or respond to culture media composition. More work is therefore needed to explain the differences in yields observed between donors under standardized and reproducible culture conditions. The payoff from this painstaking work would be the identification of potential genetic variations that could be pre-screened for or utilized by genetically engineering alterations the starting material to benefit production. Genome-wide association studies as well as identification of rare phenotypes linked to specific RBC traits might help to identify genetic variants suitable for reliably producing large numbers of mRBC. An example of such an approach was carried out by Sankaran’s group.35 LNK/SH2B3 is an adaptor protein that negatively regulates hematopoietic cytokine signaling. Rare lost-of-function SH2B3 alleles have been associated with JAK2-mutationnegative erythrocytosis36,37 and a hypomorphic allele of SH2B3 (single nucleotide polymorphism rs3184504) was found to be significantly associated with high hemoglobin levels, packed cell volume and RBC count in vivo.38 Using shRNA knockdown in adult, mobilized, peripheral blood and cord blood CD34+ cells, Giani et al.35 suppressed the expression of the LNK/SH2B3 protein and reported a 2- to 7-fold increase in yield of enucleated RBC in shRNA-treated cells compared to cells transduced with a control shRNA. More recently, a study of rare MAM-negative individuals by Thornton and colleagues39 showed that

peripheral blood CD34+ cells from two MAM-negative individuals had a proliferation advantage in ex-vivo erythroid cultures, resulting in an average 5-fold increase in cell number compared to four age- and gender-matched MAM-positive controls. Whether the same observations concerning loss-of-function SH2B3 and MAM-negative cells hold true for large-scale cultures and across multiple donors still needs to be determined. Beyond yield, the choice of donor can also affect the quality of the final mRBC product due to the donor’s own RBC intrinsic characteristics – not just in terms of blood group which can be selected for, but also in terms of storage characteristics or even longevity in circulation once transfused. The planned RESTORE clinical trial may provide data on this as the survival time of transfused stem cell-derived mRBC in the circulation of the recipient will be directly compared to the survival time in circulation of the same donor’s standard RBC.

Genetic manipulation and small molecules A key challenge for the field is to prevent the attrition of the self-renewal capacity of HSPC and to maintain the expansion capacity of erythroid progenitors (burst-forming and colony-forming units-erythroid) for a longer period before terminal erythroid differentiation occurs. One way to do this is through the use of glucocorticoids (see below) which can potentially improve the asynchronicity of enucleation in cultures, and may then improve reticulo-

Figure 1. Overview of the erythroid culture process. Ex vivo culture systems require the isolation of peripheral blood mononuclear cells (PBMNC) or magnetic sorting of the CD34+ cells as starting material. Culture systems then employ either flasks, spinner flasks or a bioreactor system depending on scale. The volume of the culture will increase dramatically as the cells expand and differentiate through days 7-14. Upon generation of a mixed population of reticulocytes, nucleated cells and pyrenocytes at day 21, cells then require filtration using either a syringe (small scale) or multiple leukocyte filters (large scale) depending on volume to give a pure reticulocyte population. Diagram made using biorender.com.

2306

haematologica | 2021; 106(9)

Manufactured red blood cells

cyte stability and filtration efficiency. Another way is to use culture manipulation to try and maintain stemness and proliferation potential. For example, there are small molecule inhibitors that are reported to enhance cord blood HSPC self-renewal, such as UM17140 and the aryl hydrocarbon receptor agonist SR-1.41 There are also factors such as angiopoietin-like 5 and IGBFP242 and notch ligand.43 Although there is evidence that these small molecules or factors enhance HSPC expansion, there are no data yet to suggest that they can enhance the yield of reticulocytes when cultured on a large scale. Perhaps the most exciting advancement in erythroid progenitor manipulation is the recently reported overexpression of BMI1 in human HSPC, which gave a 1012 fold increase of erythroblasts.44 Not only does the extensive expansion give the potential for higher yields (assuming the cells do not differentiate spontaneously when grown in large volumes), it also confers the opportunity for further genetic manipulation due to the extended time frame. Genetic manipulation in the form of YTHDF2 knockdown also generated a reported 14.3-fold increase in CD34+ frequency in the culture conditions used by a separate group of researchers.45 Alternatively, better biomimicry of the stem cell niche to recapitulate conditions ex vivo has the potential to maintain HSPC stemness for longer periods and therefore increase yield; however, these technologies still require further development.46-52 It will be very interesting to see if applications of these innovations can translate into higher yields for large-scale mRBC production.

ries maintain consistency with suppliers whenever possible. There are many different IMDM commercially available, some better than others in terms of supporting the proliferation and enucleation of erythroid progenitors. Studies are needed to determine exactly what nutrients are required to support the highest proliferation rates of HSPC and erythroid progenitors in culture, particularly important when culturing at high cell densities. Interestingly Heshusius et al. supplemented their IMDM with nucleosides and a range of trace elements to make a more defined GMP-compliant medium.7 Zhang and colleagues13 added folic acid and selenium to their large-scale cultures of human cord blood CD34+ cells. An experimental approach, using parallel stirred tank micro-bioreactors, is needed to identify the definitive media and supplements to use for erythroid culture. The lipid sources added to base media by different laboratories also vary significantly, with some groups favoring different amounts of plasma, serum (human or bovine) or serum-free conditions supplemented with animal, human or plant-derived lipid-rich reagents. For compliance with GMP, animal sources must eventually be substituted, which can have an impact on yields. In their recent report Heshuvius et al. also highlighted the importance of albumin purity for proliferation.7 Interestingly, Wilkinson et al.53 showed that 0.1% human serum albumin can be replaced by 0.1% polyvinyl alcohol for cultures of human umbilical cord blood-derived CD34+ cells but as yet this observation has not been tested on a large scale.

Media composition and optimization

Glucocorticoids

As well as exploiting cell-intrinsic properties, the base medium composition could be further developed and supplemented. Erythroid progenitors are generally cultured in Iscove modified Dulbecco medium (IMDM) and laborato-

The importance of glucocorticoids in promoting stress erythropoiesis was originally discovered in avian and mice studies54,55 and glucocorticoids have been used to increase the yield of human mRBC.26,29,56,57 Three of the larger-scale

Table 1. Summary of recently published, large-scale erythroid culture systems, with expansion and enucleation rates where provided as well as bioreactor and GMP/non-GMP media constituents where applicable. Exhaustive reviews of small-scale erythroid cultures can be found elsewhere.20,22,25

Source and culture period

General protocol

Expansion

Enucleation rate

Two-stage

2.3x10 by extrapolation

>90%

Peripheral blood CD34+ cells

Three-stage

68%

Peripheral blood CD34+ cells, 20 days

Three-stage

6.15x104 fold Large cultures (actual yield) >104 fold Large cultures (actual yield)

Peripheral blood and cord blood CD34+ cells 20 days

Three-stage

>105 fold Large cultures (actual yield)

Cord blood CD34+ cells

Four-stage

Peripheral blood MNC (no CD34+ cell isolation) 21 to 37 days (due to expansion stage)

Three-stage

2.9x105 fold Large cultures (actual yield) 107 fold by extrapolation

Cord blood, 33 days

55-95%

63%

> 90%

Key points

Reference

First demonstration of bioreactor use; Timmins et al., 2011 1 L cultures in wave-type bioreactor; non-GMP (use of BSA) 1 culture of 2.5 mL packed filtered Giarratana et al., 2011 mRBC under GMP conditions autologous human transfusion 5 mL packed filtered mRBC, Griffiths et al., 2012 constant batch feeding in spinner flasks (no medium changes); non GMP Large scale cultures (~25 L) Kupzig et al., 2017 10 mL packed filtered mRBC using constant batch feeding in spinner flasks under GMP conditions Large scale culture in rotating wall vessels; Zhang et al., 2017 non-GMP (use of 15% FBS in steps 2 & 3) G-Rex bioreactor Heshusius et al., 2019 GMP compliant: serum-free and plant derived lipids for expansion; 5% human plasma for differentiation

GMP: Good Manufacturing Practice; BSA: bovine serum albumin, mRBC: manufactured red blood cells, FBS: fetal bovine serum; MNC: mononuclear cells.

haematologica | 2021; 106(9)

2307

S. Pellegrin et al.

erythroid culture protocols reported to date incorporate glucocorticoids, using either dexamethasone in serum-free conditions7 or hydrocortisone in the presence of 5% plasma5 or in serum-free conditions.27 Recent studies have highlighted both the importance of lipid metabolism during terminal erythroid differentiation58 and the fact that exposure to glucocorticoids can affect the lipid metabolism in cultured erythroid cells.59,60 For erythroid cells cultured in the presence of hydrocortisone, the defect

in lipid metabolism and resulting fragility of mRBC can be counteracted by supplementing the medium with cholesterol-rich lipids.60 Interestingly though, addition of cholesterol-rich lipids slightly accelerated differentiation60 and Heshuvius et al. reported that the addition of plasma caused premature differentiation in the presence of dexamethasone.7 The interplay between glucocorticoid exposure time and lipid metabolism needs further investigation to maximize yield without compromising mRBC quality.

Table 2. Highlighted strategies to improve current culture systems for manufactured red blood cells.

Improving yield & quality of mRBC Starting material

• selection of donors with genetic markers linked to specific RBC traits (high hemoglobin, packed cell volume or RBC count) • selection of donors with consistently high HSPC counts • use of whole mononuclear cell population (PBMNC) not just CD34+ cells • genetic manipulation of starting material to enhance proliferation and compatibility

Media composition and supplements

• slow down attrition of HSPC self-renewal capacity • maximize expansion of erythroid progenitors (CFU-E) • maximize enucleation and maintain nascent reticulocyte viability

Filtration

• novel filtration technology to minimize loss of mRBC • maturation of mRBC to erythrocytes will enhance filtration efficiency

Storage

• optimal reticulocyte storage conditions to minimize loss of filtered mRBC until transfusion • maturation of mRBC to erythrocytes to enhance storage times

GMP-compliance and reproducibility Starting material

• minimize stem cell donor variability and cell loss during isolation

Media composition and supplements

• use defined reagents of known reliability and controlled provenance • multiple suppliers identified for key reagents • supplier surveys and site visits

Bioreactors and filtration

• closed systems • scalable GMP-compliant bioreactors with ease of use • close monitoring of the culture growth • GMP-compliant filtration process

Other considerations

• define release criteria of product and storage times • identification of optimum therapeutic dose of mRBC

Cost reduction Starting material

• use of PBMNC to circumvent expensive CD34+ cell isolation

Media composition and supplements

• defined in-house media constituents • in-house growth factors • replace or reduce the most expensive constituents (e.g. holotransferrin) • develop protocols/bioreactors that use less medium overall without affecting the yield and quality of mRBC obtained • enhance media to enable increases in cell density

Bioreactors and filtration

• fully automated culture processes • automated filtration • minimal footprint and labor requirement

RBC: red blood cells; mRBC: manufactured RBC; HSPC: hematopoietic stem and progenitor cells; PBMNC: peripheral blood mononuclear cells; CFU-E: colony-forming unit – erythroid; GMP: Good Manufacturing Practice.

2308

haematologica | 2021; 106(9)

Manufactured red blood cells

Holotransferrin The majority of current protocols use between 0.33 and 0.7 mg/mL holotransferrin isolated from human plasma,5-7 which represents the highest costing individual culture reagent. Holotransferrin is the natural carrier used to deliver iron to the developing erythroid cell by binding to CD71 and being internalized. It is then recycled by the cell and released back into the medium in its apo-form. A cheaper, plant-derived recombinant holotransferrin, optiferrin, is available but this is still expensive. Theoretically, the transferrin concentration could be reduced in culture media as long as iron is also supplemented to bind apo-transferrin, without causing cell toxicity or increasing the likelihood of bacterial growth. For example, Timmins and colleagues used 0.12 mg/mL holotransferrin in combination with 900 ng/mL ferrous sulfate and 90 ng/mL ferric nitrate.27 Olivier et al. used holotransferrin at a concentration of 0.05 mg/mL and 3 mM FeIII-EDTA in small scale immortalized pluripotent stem cell cultures, reportedly without affecting the yield of reticulocytes, but the potential impact on mean cell hemoglobin concentrations and viability during storage of the mRBC produced was not measured.61 Other iron supplements that could be tested include reagents that would deliver iron to erythroid cells in a CD71-independent manner. These include the small lipophilic molecule hinokitiol that can carry iron across the cell membrane into erythroid cells62,63 or alternatively, ferric carboxymaltose and iron sucrose, both already prescribed to patients suffering from iron deficiency.64

Bioreactors and growing erythroid cells at larger scale The majority of erythroid cultures described in the literature are small and rely on the use of static tissue plastic flasks. For the reported larger scale cultures, a variety of culture vessels have been utilized. The original 2.5 mL packed mRBC produced under GMP conditions and tested in a single volunteer were cultured in static plastic flasks.5 Spinner flasks (of 1.5 L and 3 L volumes) have since been used successfully from day 7 onwards for constantly batch-fed cultures reaching a volume of ~28 L to produce 10 mL of packed filtered reticulocytes.6 Zhang et al. used rotating wall vessels to grow 2x108 cells from cord blood CD34+.13 Most recently, Heshusius and colleagues used a 1 L gas-permeable, rapid expansion bioreactor (G-Rex; Wilsonwolf) which, as well as facilitating partial media replenishment, allowed 90% of the expansion medium to be removed and replaced by differentiation medium. Although the expansion was 10-fold lower in the G-Rex compared to static dishes, the enucleation rate was similar and by extrapolation the authors predicted that ~4.5 mL mRBC could be produced using this bioreactor.7 The cell numbers from these studies are encouraging but are still a long way from the prediction by Timmins et al.,27 who suggested it may be possible to produce 500 units from a single cord blood donation. There are still many types of bioreactors to choose from and explore further for erythroid culture including: (i) continuous stirred tank bioreactors or spinner flasks which contain internal impellers; (ii) fluidized bed bioreactors in which cells are kept in suspension by the culture medium moving upwards; (iii) rocking heated platforms (wave-like haematologica | 2021; 106(9)

bioreactors) onto which large disposable bags are attached; (iv) rotating wall vessel bioreactors also known as roller bottles and finally, (v) multi-layered static flasks. Whatever becomes the bioreactor of choice for erythroid cell culture, it will need to facilitate higher density culture, be scalable and incorporate automation. In the long term, this will make erythroid culture more cost-effective by: (i) reducing labor costs - cells would be cultured in a single container making the cultures easier to feed and less laborintensive, with the possibility of automated, remote feeding of media and/or specific depleted nutrients; (ii) reducing the footprint and space required for each batch production; (iii) easing scale up; and (iv) minimizing human error and batch-to-batch variation by carefully controlling different parameters (such as pH, agitation and oxygenation) for optimal culture conditions. One can eventually then imagine rooms filled with bioreactors manufacturing mRBC continuously at scale for clinical use. The challenge for producing mRBC in any of the above types of bioreactors type lies in the variety of culture conditions required during this 3-week process. All cultures are initiated from a small number of HSPC or approximately 100x106 PBMNC, seeded in a small volume of medium. The erythroid cells then proliferate reaching a fold expansion of >105 and requiring low cell densities for optimum growth (2-8x105 cells/mL) or medium replenishment (constant versus repeated batch feeding). Orthochromatic erythroblasts then enucleate and are relatively fragile during this process. Moreover, erythroid cultures are inherently asynchronous, a proportion of cells start enucleating while others are still proliferating; this is noticeable in the last week of culture. To maximize yield, culture conditions have to support cells at different stages of terminal differentiation and nascent reticulocytes must remain viable until the highest percentage enucleation is reached. A manufacturing process that uses a bioreactor efficiently, minimizing the transitions between different types of culture vessels during the 21 days of culture, producing a high yield of reticulocytes and automating feeding, still needs to be identified. Once the final product is made (i.e., mRBC have been filtered and stored), new cultures have to be reinitiated using new donor-derived HSPC. This is where immortalized pluripotent stem cells or an immortalized cell line that enucleates efficiently would be a game-changer as stocks of the same cell phenotype could be maintained.

Reticulocyte filtration and storage At the end of the procedure, large-scale erythroid cultures need to be volume reduced and filtered to separate the mRBC from the nucleated cells and expelled nuclei (also known as pyrenocytes). Currently, the filters used are dead-end leukoreduction filters routinely used by blood banks. These have been designed to filter whole blood, which typically consists of ~ 5x109 RBC and only ~5x106 nucleated cells per milliliter of blood. In comparison, the percentage enucleation for mRBC cultures is approximately 80% and the medium contains pyrenocytes as well as free DNA released by disintegrated nuclei. It is notable that very few studies on mRBC yield test the filterability of their final product or report the yield after filtration. This is a key parameter as the mRBC must be purified before clinical use and this process currently alters 2309

S. Pellegrin et al.

the yield by 30-50%. Strategies and more research are needed to improve reticulocyte recovery after filtration, which will lead to a dramatic increase in yield. There are open-ended filter technologies, such as tangential flow filtration systems or acoustic resonance cell filtration,65 which could potentially be used but are as yet untested. More experimental approaches include lab-on-a-chip microfluidic label-free reticulocyte-sorting methods.66 All these possibilities need testing and optimizing on a large culture scale which is expensive, therefore, highlighting filtration as a key area of research calling for innovation as well as commercial investment and collaboration. Once filtered, the mRBC need to be stored whilst quality control tests are carried out and until they can be transfused into a patient. As reticulocytes, mRBC are more fragile than RBC and more optimal storage conditions need to be developed. An alternative approach is to promote maturation of cultured reticulocytes into bona fide erythrocytes, which is another area of active research.

Compliance with Good Manufacturing Practice and quality control Finally, it should be remembered that the challenge for mRBC production is not only to deliver a manufacturing process that can produce enough mRBC at scale but to develop a process that is GMP-compliant. Often GMP compliance includes the use of clean rooms, highly trained staff specialized in GMP, using closed processes to minimize any risk of infection, robust batch manufacturing protocols, as well as specific manufacturing processing and quality release criteria which are not required in standard R&D laboratories. The challenges of GMP compliance on large-scale erythroid cultures for a transfusion product may have an impact on yield, add pressure to culture times and increase costs, so will need to be considered and planned for from the outset.

References 1. Tzounakas VL, Valsami SI, Kriebardis AG, Papassideri IS, Seghatchian J, Antonelou MH. Red cell transfusion in paediatric patients with thalassaemia and sickle cell disease: current status, challenges and perspectives. Transfus Apher Sci. 2018;57(3): 347-357. 2. Ware RE, de Montalembert M, Tshilolo L, Abboud MR. Sickle cell disease. Lancet. 2017;390(10091):311-323. 3. Hawksworth J, Satchwell TJ, Meinders M, et al. Enhancement of red blood cell transfusion compatibility using CRISPR-mediated erythroblast gene editing. EMBO Mol Med. 2018;10(6):1-11. 4. Zimring JC, Welniak L, Semple JW, Ness PM, Slichter SJ, Spitalnik SL. Current problems and future directions of transfusioninduced alloimmunization: summary of an NHLBI working group. Transfusion. 2011;51(2):435-441. 5. Giarratana MC, Rouard H, Dumont A, et al. Proof of principle for transfusion of in vitrogenerated red blood cells. Blood. 2011;118(19):5071-5079.

2310

Summary Technologies to revolutionize transfusion options for patients with anemia, in particular for those receiving regular life-long transfusions, are much needed. There is great potential for laboratory-grown mRBC to be used in transfusion practice once production is mastered at scale. In the meantime, these culture systems at smaller scale have proven to be brilliant tools for understanding human erythropoiesis and optimizing culture methodology. The efficient production of mRBC at scale is now essentially a biotechnological challenge that requires multidisciplinary efforts. We have highlighted some of the key areas, breakthroughs and challenges (summarized in Table 2), in which investment together with intensive research into further optimization of culture systems and use of bioreactors at scale are needed to make the clinical use of adult therapeutic doses of mRBC become a reality. Focused research and collaboration between academics, blood banks, commercial entities and new spinouts, especially around the use of RBC-based therapeutics, will no doubt help to drive the development and efficiency of mRBC production under GMP conditions into the clinic. Disclosures No conflicts of interest to disclose. Contributions SP, CES and AMT wrote the review together and all authors approved the final submitted version. Funding SP, CES and the work in AMT’s laboratory is funded in part by a National Institute for Health Research Blood and Transplant Research Unit (IS-BTU-1214-10032) in red blood cell products (University of Bristol) and NHSBT R&D grants(WP15-05; WP15-04). The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care.

6. Kupzig S, Parsons SF, Curnow E, Anstee DJ, Blair A. Superior survival of ex vivo cultured human reticulocytes following transfusion into mice. Haematologica. 2017;102 (3):476483. 7. Heshusius S, Heideveld E, Burger P, et al. Large-scale in vitro production of red blood cells from human peripheral blood mononuclear cells. Blood Adv. 2019;3(21): 33373350. 8. Collins AF, Gonçalves-Dias C, Haddad S, et al. Comparison of a transfusion preparation of newly formed red cells and standard washed red cell transfusions in patients with homozygous β- thalassemia. Transfusion. 1994;34(6):517-520. 9. Klein HG. Transfusions with young erythrocytes (neocytes) in sickle cell anemia. Am J Pediatr Hematol Oncol. 1982;4(2): 162-165. 10. Sharma DC, Rai S, Agarwal N, Sao S, Gaur A, Sapra R. Transfusion of neocytes concentrate/pooled neocytes in β-thalassemic patients. Indian J Hematol Blood Transfus. 2008;24(4):173-177. 11. Spanos T, Ladis V, Palamidou F, et al. The impact of neocyte transfusion in the management of thalassaemia. Vox Sang. 1996;70(4):217-223.

12. Triadou P, Girot R, Rebibo D, et al. Neocytopheresis: a new approach for the transfusion of patients with thalassaemia major. Eur J Pediatr. 1986;145(1-2):10-13. 13. Zhang Y, Wang C, Wang L, et al. Large-scale ex vivo generation of human red blood cells from cord blood CD34+cells. Stem Cells Transl Med. 2017;6(8):1698-1709. 14. Therapeutics R. Rubius Therapeutics Reports Fourth Quarter and Full-Year 2019 Financial Results and Announces Strategic Focus on Oncology and Autoimmunity. 2020. Available from: https://ir.rubiustx.com/news-releases/newsrelease-details/rubius-therapeutics-reportsfourth-quarter-and-full-year-2019. 15. Trompeter S, Estcourt L, Mora A, et al.The haemoglobinopathy survey: the reality of transfusion practice in sickle cell disease and thalassaemia in England. Transfus Med. 2020;30(6):456-466. 16. Luten M, Roerdinkholder-Stoelwinder B, Schaap NP, de Grip WJ, Bos HJ, Bosman GJ. Survival of red blood cells after transfusion: a comparison between red cells concentrates of different storage periods. Transfusion. 2008;48(7):1478-1485. 17. Anstee DJ, Gampel A, Toye AM. Ex-vivo

haematologica | 2021; 106(9)

Manufactured red blood cells

generation of human red cells for transfusion. Curr Opin Hematol. 2012;19(3):163169. 18. Douay L. Why industrial production of red blood cells from stem cells is essential for tomorrow's blood transfusion. Regen Med. 2018;13(6):627-632. 19. Migliaccio AR, Palis J. Blood in a dish: in vitro synthesis of red blood cells. Drug Discov Today Dis Mech. 2011; 8(1-2):e3-e8 20. Migliaccio AR, Masselli E, Varricchio L, Whitsett C. Ex-vivo expansion of red blood cells: How real for transfusion in humans? Blood Rev. 2012;26(2):81-95. 21. Rousseau GF, Mazurier C, Douay L. Culturing red blood cells from stem cells: a solution to present and future challenges of transfusion medicine? ISBT Sci Ser. 2016;11(S1):111-117. 22. Severn CE, Toye AM. The challenge of growing enough reticulocytes for transfusion. ISBT Sci Ser. 2018;13(1):80-86. 23. Timmins NE, Nielsen LK. Manufactured RBC - rivers of blood, or an oasis in the desert? Biotechnol Adv. 2011; 29(6):661-666. 24. Timmins NE, Nielsen LK. Blood cell manufacture: current methods and future challenges. Trends Biotechnol. 2009;27(7):415422. 25. Rousseau GF, Giarratana MC, Douay L. Large-scale production of red blood cells from stem cells: What are the technical challenges ahead? Biotechnol J. 2014;9(1):28-38. 26. Von Lindern M, Zauner W, Mellitzer G, et al. The glucocorticoid receptor cooperates with the erythropoietin receptor and c-Kit to enhance and sustain proliferation of erythroid progenitors in vitro. Blood. 1999;94 (2):550-559. 27. Timmins NE, Athanasas S, Günther M, Buntine P, Nielsen LK. Ultra-high-yield manufacture of red blood cells from hematopoietic stem cells. Tissue Eng Part C Methods. 2011;17(11):1131-1137. 28. Griffiths RE, Kupzig S, Cogan N, et al. Maturing reticulocytes internalize plasma membrane in glycophorin A-containing vesicles that fuse with autophagosomes before exocytosis. Blood. 2012;119(26): 6296-6306. 29. van den Akker E, Satchwell TJ, Pellegrin S, Daniels G, Toye AM. The majority of the in vitro erythroid expansion potential resides in CD34(-) cells, outweighing the contribution of CD34(+) cells and significantly increasing the erythroblast yield from peripheral blood samples. Haematologica. 2010; 95(9):1594-1598. 30. Heideveld E, Hampton-O'Neil LA, Cross SJ, et al. Glucocorticoids induce differentiation of monocytes towards macrophages that share functional and phenotypical aspects with erythroblastic island macrophages. Haematologica. 2018;103(3):395-405. 31. Heideveld E, Masiello F, Marra M, et al. CD14+ cells from peripheral blood positively regulate hematopoietic stem and progenitor cell survival resulting in increased erythroid yield. Haematologica. 2015;100(11): 1396-1406. 32. Bernecker C, Ackermann M, Lachmann N, et al. Enhanced ex vivo generation of erythroid cells from human induced pluripotent stem cells in a simplified cell culture system with low cytokine support. Stem Cells Dev. 2019;28(23):1540-1551. 33. Trakarnsanga K, Griffiths RE, Wilson MC, et al. An immortalized adult human erythroid line facilitates sustainable and scalable generation of functional red cells. Nat

haematologica | 2021; 106(9)

Commun. 2017;8:14750. 34. Migliaccio AR, Whitsett C, Migliaccio G. Erythroid cells in vitro: from developmental biology to blood transfusion products. Curr Opin Hematol. 2009;16(4):259-268. 35. Giani FC, Fiorini C, Wakabayashi A, et al. Targeted application of human genetic variation can improve red blood cell production from stem cells. Cell Stem Cell. 2016;18(1):73-78. 36. Lasho TL, Pardanani A, Tefferi A. LNK mutations in JAK2 mutation-negative erythrocytosis. N Engl J Med. 2010;363(12): 1189-1190. 37. Spolverini A, Pieri L, Guglielmelli P, et al. Infrequent occurrence of mutations in the PH domain of LNK in patients with JAK2 mutation-negative 'idiopathic' erythrocytosis. Haematologica. 2013;98(9):e101-102. 38. van der Harst P, Zhang W, Mateo Leach I, et al. Seventy-five genetic loci influencing the human red blood cell. Nature. 2012;492 (7429):369-375. 39. Thornton N, Karamatic Crew V, Tilley L, et al. Disruption of the tumour-associated EMP3 enhances erythroid proliferation and causes the MAM-negative phenotype. Nat Commun. 2020;11(1):3569. 40. Fares I, Chagraoui J, Gareau Y, et al. Cord blood expansion. Pyrimidoindole derivatives are agonists of human hematopoietic stem cell self-renewal. Science. 2014;345 (6203):1509-1512. 41. Boitano AE, Wang J, Romeo R, et al. Aryl hydrocarbon receptor antagonists promote the expansion of human hematopoietic stem cells. Science. 2010;329(5997):13451348. 42. Zhang CC, Kaba M, Iizuka S, Huynh H, Lodish HF. Angiopoietin-like 5 and IGFBP2 stimulate ex vivo expansion of human cord blood hematopoietic stem cells as assayed by NOD/SCID transplantation. Blood. 2008;111(7):3415-3423. 43. Delaney C, Heimfeld S, Brashem-Stein C, et al. Voorhies H, Manger RL, Bernstein ID. Notch-mediated expansion of human cord blood progenitor cells capable of rapid myeloid reconstitution. Nat Med. 2010;16 (2):232-236. 44. Liu S, Wu M, Lancelot M, et al. BMI1 enables extensive expansion of functional erythroblasts from human peripheral blood mononuclear cells. Mol Ther. 2021;29(5): 1918-1932. 45. Li Z, Qian P, Shao W, et al. Suppression of m(6)A reader Ythdf2 promotes hematopoietic stem cell expansion. Cell Res. 2018;28(9):904-917. 46. Severn CE, Eissa AM, Langford CR, et al. Ex vivo culture of adult CD34(+) stem cells using functional highly porous polymer scaffolds to establish biomimicry of the bone marrow niche. Biomaterials. 2019; 225:119533. 47. Severn CE, Macedo H, Eagle MJ, Rooney P, Mantalaris A, Toye AM. Polyurethane scaffolds seeded with CD34(+) cells maintain early stem cells whilst also facilitating prolonged egress of haematopoietic progenitors. Sci Rep. 2016;6:32149. 48. Raic A, Rodling L, Kalbacher H, LeeThedieck C. Biomimetic macroporous PEG hydrogels as 3D scaffolds for the multiplication of human hematopoietic stem and progenitor cells. Biomaterials. 2014;35(3): 929940. 49. Rodling L, Raic A, Lee-Thedieck C. Fabrication of biofunctionalized, cell-laden macroporous 3D PEG hydrogels as bone

marrow analogs for the cultivation of human hematopoietic stem and progenitor cells. Methods Mol Biol. 2014;1202:121-130. 50. Mortera-Blanco T, Mantalaris A, Bismarck A, Aqel N, Panoskaltsis N. Long-term cytokine-free expansion of cord blood mononuclear cells in three-dimensional scaffolds. Biomaterials. 2011;32(35):9263-9270. 51. Raic A, Naolou T, Mohra A, Chatterjee C, Lee-Thedieck C. 3D models of the bone marrow in health and disease: yesterday, today and tomorrow. MRS Commun. 2019;9(1):37-52. 52. Bello AB, Park H, Lee SH. Current approaches in biomaterial-based hematopoietic stem cell niches. Acta Biomater. 2018;72:1-15. 53. Wilkinson AC, Ishida R, Kikuchi M, et al. Long-term ex vivo haematopoietic-stemcell expansion allows nonconditioned transplantation. Nature. 2019;571(7763): 117-121. 54. Bauer A, Tronche F, Wessely O, et al. The glucocorticoid receptor is required for stress erythropoiesis. Gen Dev. 1999;13(22):29963002. 55. Wessely O, Deiner EM, Beug H, von Lindern M. The glucocorticoid receptor is a key regulator of the decision between self-renewal and differentiation in erythroid progenitors. EMBO J. 1997;16(2):267-280. 56. Flygare J, Estrada VR, Shin C, Gupta S, Lodish HF. HIF1α synergizes with glucocorticoids to promote BFU-E progenitor selfrenewal. Blood. 2011;117(12):3435-3444. 57. Narla A, Dutt S, McAuley JR, et al. Dexamethasone and lenalidomide have distinct functional effects on erythropoiesis. Blood. 2011;118(8):2296-2304. 58. Huang NJ, Lin YC, Lin CY, et al. Enhanced phosphocholine metabolism is essential for terminal erythropoiesis. Blood. 2018;131 (26):2955-2966. 59. Zingariello M, Bardelli C, Sancillo L, et al. Dexamethasone predisposes human erythroblasts toward impaired lipid metabolism and renders their ex vivo expansion highly dependent on plasma lipoproteins. Front Physiol. 2019;10:281. 60. Bernecker C, Köfeler H, Pabst G, et al. Cholesterol deficiency causes impaired osmotic stability of cultured red blood cells. Front Physiol. 2019;10:1529. 61. Olivier EN, Zhang S, Yan Z, et al. PSC-RED and MNC-RED: albumin-free and lowtransferrin robust erythroid differentiation protocols to produce human enucleated red blood cells. Exp Hematol. 2019;75:3152.e15. 62. Grillo AS, SantaMaria AM, Kafina MD, et al. Restored iron transport by a small molecule promotes absorption and hemoglobinization in animals. Science. 2017;356(6338): 608-616. 63. Aoto M, Iwashita A, Mita K, Ohkubo N, Tsujimoto Y, Mitsuda N. Transferrin receptor 1 is required for enucleation of mouse erythroblasts during terminal differentiation. FEBS Open Bio. 2019;9(2):291-303. 64. Geisser P, Burckhardt S. The pharmacokinetics and pharmacodynamics of iron preparations. Pharmaceutics. 2011;3(1):12-33. 65. Trampler F, Sonderhoff SA, Pui PW, et al. Kilburn DG, Piret JM. Acoustic cell filter for high density perfusion culture of hybridoma cells. Biotechnology (N Y). 1994;12(3):281284. 66. Zeming KK, Sato Y, Yin L, et al. Microfluidic label-free bioprocessing of human reticulocytes from erythroid culture. Lab Chip. 2020;20(18):3445-3460.

2311

REVIEW ARTICLE Ferrata Storti Foundation

Targeting the tumor microenvironment in chronic lymphocytic leukemia Rebecka Svanberg,1* Sine Janum,2* Piers E.M. Patten,3 Alan G. Ramsay3 and Carsten U. Niemann1 1 Department of Hematology, Rigshospitalet, Copenhagen, Denmark; 2Department of Clinical Haemato-oncology, Bartholomew’s Hospital, Barts Health Trust, London, UK; 3 School of Cancer and Pharmaceutical Sciences, Faculty of Life Sciences & Medicine, King's College London, London, UK

Haematologica 2021 Volume 106(9):2312-2324

*RS and SJ contributed equally as co-first authors.

ABSTRACT

Correspondence: CARSTEN UTOFT NIEMANN carsten.utoft.niemann@regionh.dk Received: August 6, 2020. Accepted: March 31, 2021. Pre-published: April 22, 2021. https://doi.org/10.3324/haematol.2020.268037

2312

he tumor microenvironment (TME) plays an essential role in the development, growth, and survival of the malignant B-cell clone in chronic lymphocytic leukemia (CLL). Within the proliferation niches of lymph nodes, bone marrow, and secondary lymphoid organs, a variety of phenotypically and functionally altered cell types, including T cells, natural killer cells, monocytes/macrophages, endothelial and mesenchymal stroma cells, provide crucial survival signals, along with CLL-cellinduced suppression of antitumor immune responses. The B-cell receptor pathway plays a pivotal role in mediating the interaction between CLL cells and the TME. However, an increasing number of additional components of the multifactorial TME are being discovered. Although the majority of therapeutic strategies employed in CLL hitherto have focused on targeting the leukemic cells, emerging evidence implies that modulation of microenvironmental cells and CLL-TME interactions by novel therapeutic agents significantly affect their clinical efficacy. Thus, improving our understanding of CLL-TME interactions and how they are affected by current therapeutic agents may improve and guide treatment strategies. Identification of novel TME interactions may also pave the road for the development of novel therapeutic strategies targeting the TME. In this review, we summarize current evidence on the effects of therapeutic agents on cells and interactions within the TME. With a growing demand for improved and personalized treatment options in CLL, this review aims at inspiring future exploration of smart drug combination strategies, translational studies, and novel therapeutic targets in clinical trials.

Introduction Chronic lymphocytic leukemia (CLL) is a B-cell malignancy characterized by the clonal expansion of CD5+/CD19+ malignant B cells, and displays a heterogeneous pathology with chromosomal aberrations, recurrent mutations, and microenvironmental involvement.1 Although characterized by an accumulation of malignant cells in peripheral blood, CLL develops in protective niches and proliferation centers within the bone marrow, lymph nodes, the spleen and, more rarely, the liver.2 These tissues allow close interactions between malignant cells and various host cells constituting the tumor microenvironment (TME). The survival and growth of CLL cells is highly dependent on support from these surrounding microenvironmental cells that include T cells, monocytes/macrophages, endothelial and mesenchymal stroma cells, and natural killer (NK) cells.2-5 The complex crosstalk between CLL cells and these essential microenvironmental components is still poorly defined but studies have revealed how these interactions support disease progression and drug resistance.6-9 For an extensive and detailed overview of the CLL-TME constituents and interactions, we refer the reader to previously published reviews,3-5 as a complete review of the CLL TME is beyond the scope of this review. However, key components and interactions relevant for the contents of this review are briefly highlighted here.

haematologica | 2021; 106(9)

Targeting the tumor microenvironment in CLL

The T-cell compartment in CLL has a complex dual role since it can exert both pro-tumor as well as anti-CLL cytotoxic activity.10 Recruited CD4+ T helper cells (T cells) within proliferation centers provide tumor support through CD40/CD40 ligand (CD40L) co-stimulation and cytokine signaling.11,12 In the peripheral blood of patients, T-cell numbers are increased with skewing towards cytotoxic CD8+ T cells and enriched effector cell subpopulations.13 Both CD4+ and CD8+ T-cell subpopulations exhibit functional defects including impaired immune synapse formation with antigen-presenting cells, impaired cytokine production, degranulation, and antitumor cytotoxicity.14-16 Furthermore, T cells in CLL show increased expression of markers of chronic activation and exhaustion, such as programmed cell death protein 1 (PD-1),13,16 contributing to inhibited effector function and impaired immunological synapse formation.15,16 Patients with CLL also have elevated numbers of regulatory T cells (T ), a subset of immunosuppressive T cells that constitute significant suppressors of antitumor T-cell responses.17 Thus, T cells play an important supportive role in CLL, whereas the accumulation of T and exhausted cytolytic T cells prevent effective anti-CLL effector functions. Similarly, myeloid cells in CLL play both tumor-supportive and immunosuppressive roles. These cells include nurse-like cells (NLC), which constitute an essential tumor-supporting component of the TME. NLC, generated in vitro, protect CLL cells from spontaneous and druginduced apoptosis, promote migration, and aid recruitment of tumor-supportive T cells.18-20 Importantly, NLC reveal a strong resemblance to tumor-associated macrophages infiltrating lymph node tissue in CLL.21 In contrast, myeloid cells with immunosuppressive properties, termed myeloid-derived suppressor cells (MDSC), accumulate in the peripheral blood of CLL patients.22 In vitro, CLL-induced MDSC suppress T-cell effector function and promote T differentiation.23 Thus, MDSC represent a significant immunosuppressive component within the CLL-TME. Co-culturing CLL cells with bone marrow-derived stromal cells or endothelial cells abrogates the spontaneous apoptosis of CLL cells in vitro, highlighting the supportive role of stromal cells in the CLL-TME.24 Stromal cells mediate lymphocyte trafficking and homing, and promote CLL survival and proliferation by inducing expression of proangiogenetic and anti-apoptotic proteins.19,24 Thus, the CLL-TME constitutes a complex cellular and molecular network that contributes to tumor survival and immune suppression. The B-cell receptor pathway is a central mechanism by which CLL cells maintain their crucial interaction with the TME. It consists of an antigen-binding transmembrane immunoglobulin connected to downstream regulators including spleen tyrosine kinase (SYK), Bruton tyrosine kinase (BTK), and phosphoinositide-3-kinase δ (PI3Kδ) (Figure 1A). B-cell receptor signaling, recently reviewed elsewhere,25 promotes proliferation, survival, and migration of the malignant clone. Stimulation of B-cell activating factor receptor (BAFF-R) by its ligand B-cell activating factor (BAFF) provided by, for example, NLC in the TME, also promotes important pro-survival and growth signals.26 Furthermore, through direct cell-cell contact by coexpressed adhesion molecules such as lymphocyte function-associated antigen-1 (LFA-1) and intercellular adhesion molecule-1, and chemokine signaling via the CXC h

reg

haematologica | 2021; 106(9)

motif chemokine receptor (CXCR)4/CXC ligand (CXCL)12 axis, TME constituents, such as NLC and stromal cells, aid migration and homing of CLL cells into protective niches.4,18,19,24 Reciprocally, CLL cells release cytokines including interleukin (IL)-6 and IL-10,27,28 chemokines such as CCL2,12 and extracellular vesicles,4,29 through which they recruit and alter microenvironmental cells, thus inducing a tumor-supportive niche. The above highlighted CLL-TME constituents and interactions are summarized in Online Supplementary Figure S1. The immune-subversive milieu preventing the host immune system from eliminating CLL cells also entails a state of clinical immune dysfunction, manifested as an increased risk of infections and autoimmune conditions in patients with CLL.30 Thus, the CLL-TME is not merely a “silent” support system for malignant cells, but contributes significantly to clinical presentation and disease aggressiveness. The majority of therapeutic strategies employed hitherto have been designed to target the survival axes of CLL cells, as exemplified by the development of inhibitor drugs targeting the B-cell receptor pathway. However, as our knowledge on the mechanisms of action is expanding, there is emerging evidence that targeted agents modulate immune TME cells and interactions, which likely profoundly influences clinical responses. These effects occur both indirectly, through elimination of CLL cells and/or disruption of critical CLL-TME interaction pathways, and directly, through inhibition of targets within the specific TME cells (Figure 1B). Furthermore, some novel treatment modalities rely directly on the engagement and activation of microenvironmental cells for their anti-CLL activity (Figure 1B). In order to improve tailored treatment options for patients with CLL, and ultimately improve the clinical course of the disease, a better understanding of how current novel therapies affect the CLL-TME is warranted. Here we review the current knowledge on how novel targeted therapies modulate CLL-TME cells and their interactions. We discuss implications for future treatment strategies and the development of combination therapy, and highlight potential novel therapeutic targets that warrant future exploration.

BTK inhibitors The introduction of small molecule inhibitors of BTK, a TEC family kinase that plays a crucial role downstream of B-cell receptor signaling, has shifted the paradigm for CLL treatment during the past decade. Ibrutinib (PCI-32765) was the first oral covalent BTK inhibitor to be approved for CLL by the Food and Drug Administration. Secondgeneration BTK inhibitors, acalabrutinib and zanubrutinib, are currently being introduced into clinical use.31,32 BTK inhibition by ibrutinib inhibits activation-induced proliferation and induces apoptosis of CLL cells.33 However, a growing number of studies describe effects of ibrutinib on several components of the TME. Changes in total T-cell numbers induced by ibrutinib are controversial, as studies have documented both increased and decreased total T-cell numbers in patients treated with ibrutinib.34-36 This discrepancy may be due to differences in treatment duration and disease status at the time of fol2313

R. Svanberg et al.

low-up, as well as differences between cohorts of patients. Increased T-cell numbers were observed during the first 6 months of treatment in one study,35 while a decrease and normalization of T-cell numbers were found in studies with longer follow-up.33,34,36 This may suggest a correlation between T-cell dynamics and CLL tumor burden during ibrutinib treatment. It was previously demonstrated that T-cell receptor repertoire diversity increased in patients upon ibrutinib treatment, which correlated with disease response and lower infection rates.34 Interestingly, an increase in clonal T cells during ibrutinib treatment, which could be linked to residual CLL disease persistence and the co-occurrence of anti-CLL T-cell clones, was reported recently,37 suggesting that residual disease may maintain certain, specific anti-CLL T-cell clones. Thus, reduced tumor burden as an indirect effect of ibrutinib likely contributes significantly to normalization of the majority of

the T-cell repertoire along with T-cell numbers. Ibrutinib exhibits off-target activity against IL-2-inducible T-cell kinase (ITK), a TEC kinase signaling downstream of the Tcell receptor, which plays a role in T-cell activation, cytokine release, and proliferation.38 The second-generation BTK inhibitors, acalabrutinib and zanubrutinib, have increased BTK selectivity but an insignificant inhibitory effect on ITK.39,40 In contrast to ibrutinib, treatment with acalabrutinib and zanubrutinib did not alter patients’ Tcell numbers; however, the follow-up time in these studies was limited to 6-7 months, when residual disease may still be present.35,40 Thus, further studies are warranted to clarify the potential contribution of direct ITK inhibition to the changes in T-cell numbers seen with ibrutinib. It was also demonstrated that ibrutinib restored T-cell proliferation and degranulation,36 enhanced T-cell lytic immune synapse function,41 and reversed the

Figure 1. Overview of targets within the chronic lymphocytic leukemia cell, and mechanisms of tumor microenvironment modulation by targeted agents. (A) The chronic lymphocytic leukemia (CLL) cell including targets within the B-cell receptor pathway and anti-apoptotic pathway. Downstream of the B-cell receptor, BTK is inhibited by ibrutinib, acalabrutinib, and zanubrutinib, and PI3Kδ is inhibited by idelalisib, duvelisib, and umbralisib. The anti-apoptotic protein BCL-2 is inhibited by venetoclax. (B) Direct versus indirect effects of targeted agents, and activation of tumor microenvironment (TME) anti-CLL activity by novel treatment modalities. Inhibition (both on- and off-target) of targets within the specific TME cells are here referred to as direct effects, exemplified by off-target inhibition of ITK in T cells by ibrutinib, and (on-target) inhibition of PI3Kδ in T cells by idelalisib. Changes occurring due to elimination of CLL cells and/or disruption of critical CLL-TME interaction pathways are here referred to as indirect effects, exemplified by CLL tumor-debulking by ibrutinib, idelalisib, or venetoclax, and disruption of protective signaling between nurse-like cells/tumor-associated macrophages and CLL cells by ibrutinib and idelalisib. Chimeric antigen receptor (CAR) T cells, bispecific antibodies, and immune checkpoint blockade immunotherapy rely directly on the engagement and activation of microenvironmental cells for antiCLL activity. Binding of CAR T cells to CD19 on CLL cells activates cytolytic antiCLL T-cell activity, bispecific antibodies redirect T cells into CLL cell proximity and engage T-cell anti-tumor activity, and immune checkpoint blockade abrogates checkpoint inhibitory signals unleashing the anti-CLL activity of tumor-infiltrating T cells. CLL: chronic lymphocytic leukemia; BTK: Bruton tyrosine kinase; SYK: spleen tyrosine kinase; PI3Kδ: phosphoinositide-3-kinase δ, BCR: B-cell receptor; TCR: T-cell receptor; ITK: interleukin-2-inducible T-cell kinase BCL-2: B-cell leukemia/lymphoma-2; TME: tumor microenvironment; NLC: nurse-like cells; TAM: tumor-associated macrophages; CAR T: chimeric antigen receptor T cells; CD: cluster of differentiation; PD-1: programmed cell death protein-1; PD-1L: programmed cell death protein-1 ligand.

2314

haematologica | 2021; 106(9)

Targeting the tumor microenvironment in CLL

exhaustion/chronic activation T-cell phenotype illustrated by PD-1 downregulation,33,35 supporting the concept that ibrutinib improves T-cell function. Similarly, treatment with acalabrutinib and zanubrutinib downregulated T-cell PD-1 expresssion.35,40 Thus, reduced exhaustion phenotypes could be due to indirect removal of tumor burden by all three BTK inhibitors. However, improved T-cell functionality may also be due to differential effects on CD4+ and CD8+ subsets, which could be linked to direct off-target activity of ibrutinib. ITK has particular importance for T 2 T-cell polarization as well as for the development of T . Ibrutinib promoted T 1 polarization in a CLL mouse model,38 but this has been more challenging to detect in patients receiving therapy.35 Furthermore ex vivo ibrutinib treatment of γδ T cells from CLL patients promoted a T 1 phenotype leading to improved antitumor effector function, indicating effects due to direct off-target ITK inhibition.42 Ibrutinib treatment also reduced the fraction of T in CLL patients,43 while treatment with acalabrutinib did not affect T numbers, further indicating direct off-target ITK inhibition by ibrutinib.35,43 Reduced numbers of CD4+ IL-17 producing T cells (T 17 cells) in ibrutinib-treated patients, as well as reduced T 17 differentiation in vitro, have also been demonstrated, recapitulating findings from ITK knockout mice.33 As for T , acalabrutinib did not affect T 17-cell numbers.35 However, contradictory findings, with increased T 17 T cells in patients with CLL receiving ibrutinib, have been reported.35 This is perhaps due to complex CD4+ subset changes which are related to time on therapy and prior treatment history in study cohorts. Additionally, although current data support ibrutinib-mediated direct ITK inhibition in both T 17 and T subsets, given their antagonizing roles,44 indirect effects on T 17 T cells due to reductions of T may “dominate” the direct effects, and contribute to this compartment expanding. Inhibition of B-cell receptor signaling leading to redistribution of CLL cells from sanctuary niches into the peripheral blood is a hallmark of ibrutinib treatment,45 the mechanism of which is, in part, disruption of microenvironmental interactions. Bone marrow specimens from ibrutinib-treated patients revealed disruption of (tumor-associated) macrophage-CLL cell contacts, with macrophage cellular protrusions contracting during therapy, likely reflecting a loss of NLC pro-survival signaling.33 Ibrutinib has been shown to block BTK and downstream transcription factors within macrophages, resulting in downregulated expression of the chemokines CXCL12 and CXCL13, thus suggesting a direct effect of ibrutinib on macrophages.46 The reduced levels of these chemokines further compromised adhesion and migration of malignant B cells in vitro.46 In accordance, ibrutinib-mediated inhibition of the migratory response of CLL cells towards these chemokines was demonstrated.33,47 Thus, direct effects of ibrutinib on macrophages seem to mediate inhibited CLL-cell chemotaxis and adhesion, thereby likely contributing significantly to the clinical reduction in lymph node and spleen size, and concomitant peripheral lymphocytosis.45 Contrariwise, unfavorable effects of ibrutinib, including impaired phagocytosis in macrophages and neutrophils, and inhibited NK-cell activation and suppressed antibody-dependent cellular cytotoxicity by NK cells have been demonstrated, likely related to direct inhibition of BTK and ITK by ibrutinib in these cells. This may have important clinical implications h

reg

haematologica | 2021; 106(9)

for combination treatment with CD20 antibodies.48 Reduction of MDSC and a concomitant increase in classical monocytes were recently demonstrated in patients with CLL after 12 months of ibrutinib treatment,36 and were likely due to both direct effects of BTK inhibition in MDSC,49 and indirect effects induced through reduced tumor burden. Given their suppressive effect on T-cell function,23 a reduction of MDSC may also further contribute indirectly to improved T-cell/immune functions. Moreover, ibrutinib abrogates the adherence of vascular cell adhesion molecule-1-positive CLL cells to fibronectin on stromal cells, thereby further reducing the ability of CLL cells to remain in the protective tissue niches.47,50 Although ibrutinib produces impressive clinical results, treatment resistance is emerging,6 and residual disease remains a challenge. In vitro studies have demonstrated a protective effect of NLC in the presence of ibrutinib, thereby implying a role for NLC in contributing to residual disease and the development of ibrutinib resistance.7 Furthermore, it was demonstrated that ibrutinib-resistant subclones harboring BTK mutations promote proliferation of BTK wild-type cells during ongoing ibrutinib treatment through paracrine stimulation, further implying a role of microenvironment crosstalk in the development of resistance.8 A number of studies seem to point toward improved clinical immune function due to the TME modulations mediated by ibrutinib.51,52 This issue, however, remains controversial, as there is still a lack of data demonstrating reduced risk of infections compared to prior ibrutinib treatment. However, studies do indicate that restoration of immune phenotypes and function establish after longterm treatment.33,34,36,51 This is in line with previous realworld data demonstrating that infectious adverse events in patients with CLL treated with ibrutinib are most frequent during the first 6 months, after which infection rates decline.53 Thus, the long-term indirect effects of ibrutinib due to reduced tumor burden and disrupted CLL-TME crosstalk may allow the various immune cell compartments to re-establish normal host immunity; however, further studies on this matter are warranted. Continued investigation of the impact of BTK inhibitors on the TME compartments is warranted in order to provide tailored treatment strategies to improve clinical outcome (residual and progressive disease) and immune function in patients with CLL, while evading emergence of drug resistance. The most important effects of BTK inhibitors on the TME are summarized in Table 1 and illustrated in Figure 2.

PI3K inhibitors In addition to BTK, PI3K constitutes another critical component of the B-cell receptor signaling pathway (Figure 2). Idelalisib is a selective inhibitor of PI3Kδ, the PI3K isoform generally restricted to hematopoietic cell types,54 and was the first PI3K inhibitor approved for CLL treatment. In preclinical studies, idelalisib induced caspase-dependent apoptosis of primary CLL cells and also reduced their chemokine secretion, independently of cytogenetics or IgHV mutational status.55,56 Although treatment of autologous T cells and NK cells with idelalisib does not induce apoptosis in these cells, it does decrease their production of inflammatory cytokines (IL-6, IL-10, tumor necrosis factor2315

R. Svanberg et al.

interfere with CXCL12-mediated chemotaxis, and abrogates adhesion of CLL cells to stromal cells, suggesting an indirect mechanism through disrupting the protection of CLL cells provided by the TME.9,56 This correlates with clinical findings of reduced lymphadenopathy and splenomegaly concomitant with lymphocytosis and significantly reduced levels of CLL-related chemokines.59 It has been demonstrated that idelalisib impairs neutrophil function ex vivo,60 which together with changes in cytotoxic T-cell subsets and strong suppression of T , likely contribute to the increased immune-related adverse events and increased risk of infections observed upon idelalisib treatment in clinical trials.61 The next-generation PI3K inhibitor, duvelisib, a dual inhibitor of PI3K isoforms δ and γ, was recently approved for the treatment of

α [TNF-α], interferon [IFN]-γ) and activation-induced molecules (CD40L).55 These changes could potentially have effects on both pro-tumor and antitumor immune functions. In addition, idelalisib antagonizes the CLL pro-survival functions of TNF-α and CD40L.55 The effect of idelalisib on the T subset has been a focus of previous studies, as inactivation of PI3Kδ in mice impaired T -mediated immune tolerance, enhancing CD8+ T-cell mediated cytotoxic responses towards tumor cells.57 Interestingly, PI3Kδ inhibition in a CLL mouse model resulted in reduced numbers and maturation of T ; however, this did not result in enhanced antitumor CD8+ T-cell function, likely due to concomitant direct inhibition of PI3Kδ downstream of Tcell receptor signaling.58 Similar to the effects of ibrutinib, idelalisib seems to reg

reg

regs

Table 1. Effect of novel therapeutic agents on the microenvironment in chronic lymphocytic leukemia.

Population

Agent

Functional changes

BTK inhibitors -ibrutinib (ibr) -acalabrutinib (aca) -zanubrutinib (zan)

Increased T-cell receptor diversity34,37 (ibr) Enhanced T-cell lytic immune synapse function52 (ibr) Skewing towards T 1 polarization23,53,54 (ibr) Reduced T-cell PD-1 expression/exhaustion phenotype49,51,67 (ibr, aca, zan) Reduced number of T 51,55 (ibr) Reduced secretion of inflammatory cytokines70 (id) Inhibition of T functions73,79 (id, duv)

T cells

reg

PI3K inhibitors -idelalisib (id) -duvelisib (duv) BCL-2 inhibitors -venetoclax (ven) IMiD/CELMoD -lenalidomide (len) -avadomide (ava)

reg

Reduced number of T cells70 (ven) Reduced number of T 70 (ven) Decreased T-cell PD-1 expression70 (ven) Immune activation, repaired T-cell dysfunction20,24,90,91 (len) Suppressed T-cell proliferation93 (len) Promotion of T 1 polarization92 (len) Induction of inflammatory IFN type I and II signaling in previously exhausted T-cells27 (ava) reg

Myeloid cells BTK inhibitors -ibrutinib (ibr)

PI3K inhibitors -idelalisib (id) IMiD -lenalidomide (len)

Abrogation of the protective contact49 (ibr) Inhibited chemokine signaling and mediation of CLL cell homing58,59 (ibr) Impaired phagocytosis by macrophages and neutrophils48 (ibr) Reduced number of MDSC and increased classical monocytes36 (ibr) Impaired neutrophil inflammatory responses76 (id) Impaired migration/chemotaxis, abrogated CLL cell protective capability, increased phagocytosis93 (len)

Stromal cells

NK cells

BTK inhibitors -ibrutinib (ibr) PI3K-inhibitors -idelalisib (id)

Revoked adherence to stromal cells in protective niches59,60 (ibr)

BTK inhibitors -ibrutinib (ibr) PI3K inhibitors -idelalisib (id) BCL-2 inhibitors -venetoclax (ven)

Inhibited NK-cell activation48 (ibr) Suppressed ADCC48 (ibr) Reduced secretion of inflammatory cytokines70 (id)

Reduced chemotaxis and impaired adhesion71,74 (id)

Decreased number of NK cells70 Improved NK-cell function70

CLL: chronic lymphocytic leukemia; BTK: Bruton tyrosine kinase; ITK; interleukin-2-inducible T-cell kinase; BCL-2: B-cell lymphoma 2; T : T helper; PD-1: programmed cell death protein 1; T : regulatory T-cell; PI3K: phosphoinositide 3’-kinase; IMiD: immunomodulatory drugs; CELMoD: cereblon E3 ligase modulators; IFN: interferon; MDSC: myeloid derived suppressor cells; ADCC: antibody-dependent cellular cytotoxicity. h

reg

2316

haematologica | 2021; 106(9)

Targeting the tumor microenvironment in CLL

relapsed/refractory CLL. Similar to idelalisib, treatment with duvelisib entails increased risk of immune-related toxicities and infections in patients with CLL,62 likely also due to strong direct inhibitory effects on T and cytotoxic T-cell effector function as demonstrated in a CLL mouse model.63 In contrast, another next-generation PI3K inhibitor, umbralisib, with dual PI3Kδ/casein kinase-1-e (CK1e) inhibitory activity, did not modulate T function, which was associated with lower toxicity in a murine model.63 Thus, the disadvantageous direct effects of idelalisib and duvelisib on T-cell subsets contributing to a risk of infections and toxicity, which have hampered their clinical usage, could potentially be mitigated with the use of umbralisib due to altered PI3K specificity. The most important effects of PI3K inhibition on the TME are summarized in Table 1 and illustrated in Figure 2. reg

reg

BCL-2 inhibitors The anti-apoptotic regulatory protein B-cell lymphoma 2 (BCL-2) is constitutively upregulated in several lymphomas including CLL, hence playing a dominant role in blocking apoptotic signaling and promoting survival in these malignancies.64 Venetoclax (ABT-199), a selective BCL-2 inhibitor, demonstrated the ability to induce rapid

apoptosis in primary CLL cells in vitro and in xenograft models65 (Figure 3). In clinical trials, venetoclax alone or combined with an anti-CD20 antibody has achieved deep and durable undetectable minimal residual disease in patients with CLL.64,66 However, while leukemic cells are highly dependent on BCL-2, the dependence of nonleukemic cells on this protein seems to vary substantially. The high prevalence of grade 3/4 neutropenia among patients treated with venetoclax likely reflects the relatively marked dependency of granulopoiesis on BCL-2.67 T-cell homeostasis also depends on BCL-2, however with variable impact on different T-cell subsets. While murine naïve T cells were found to be highly dependent on BCL2, the protein was dispensable for memory T cells.68 Coherently, a decrease in naïve T-cell subsets and increased memory T cells have also been reported in both mice and healthy human subjects receiving venetoclax69 (Figure 3). A study on CLL patients treated with venetoclax and the CD20 antibody obinutuzumab documented decreased numbers of normal B cells, NK cells, and T cells, including T , in the peripheral blood. In addition, a decrease in the exhausted/chronically activated PD1+ Tcell phenotype was observed, along with improved NKcell function, and reductions of the levels of elevated inflammatory cytokines70 (Figure 3). The authors interpreted these changes as being indirect effects due to eradicaregs

Figure 2. Effects of BTK inhibitors on the chronic lymphocytic leukemia tumor microenvironment. Inhibitory effects are represented by bars, stimulatory effects are represented by arrows. Upward arrows indicate increases, downward arrows indicate decreases. CLL: chronic lymphocytic leukemia; TME: tumor microenvironment; BTK: Bruton tyrosine kinase; BTKi: BTK inhibitor; ITK:interleukin-2-inducible T-cell kinase; PI3Kδ: phosphoinositide-3-kinase δ; TCR: T-cell receptor; CD: cluster of differentiation; CD40L: CD40 ligand; IL: interleukin; TNF: tumor necrosis factor; IFN: interferon; PD-1: programmed cell death protein 1; T : T helper; T : regulatory T cell; ADCP: antibody-dependent cellular phagocytosis; VCAM: vascular cell adhesion molecule; VLA, very late antigen; CXCL: CXC motif chemokine; MDSC: myeloid-derived suppressor cells. h

haematologica | 2021; 106(9)

reg

2317

R. Svanberg et al.

tion of the leukemic cells, and any direct effects on these cells by venetoclax were not investigated. Critically, the TME also appears to play a role in venetoclax resistance. In a previous study, in vitro CD40/CD40L co-stimulation strongly reduced sensitivity to venetoclax through upregulation of other anti-apoptotic proteins, such as myeloid cell leukemia 1 (MCL-1) and B-cell lymphoma extra large (BCL-XL), in CLL cells.71 The varying dependency on BCL2 among different microenvironmental cell types, as well as between patients, warrants further investigation, in order to optimize the advantages of targeting the apoptotic pathway in malignant cells, and utilize potential immunomodulatory effects in the immune TME while minimizing disadvantageous on-target-but-off-leukemic effects leading to adverse events.

Immunomodulatory drugs Lenalidomide is an immunomodulatory drug (IMiD) widely used to treat multiple myeloma. Despite having no direct cytotoxicity against CLL cells in vitro,72 clinical activity in patients with CLL has been demonstrated,73,74 supporting anti-CLL immunomodulatory effects in the TME as a principle mode of action. In vitro, lenalidomide induces downregulation of CLL immune checkpoint receptors on T cells, suggesting treatment-induced immune activation or reversal of exhaustion.15 Moreover, lenalidomide treatment of autologous T cells and CLL cells triggers repair of T-cell dysfunction. This results in improved synapse formation, granzyme B- and IL-21mediated cytotoxicity, enhanced CD8+ T-cell effector killing, and restored LFA-1-mediated T-cell motility.14,75–77 Supporting this, in vivo samples from treated patients revealed changes in the composition of the T-cell subpopulations and their cytokine production.78 Lenalidomide also affects CLL monocytes/NLC. The presence of lenalidomide impaired migration of CLL-supportive monocytes towards CCL2, CCL3, and CXCL12 in in vitro chemotaxis assays.79 The same study demonstrated downregulation of genes associated with pro-survival signals for CLL cells and impaired protective ability of NLC.79 Moreover, CLL-induced immunosuppression was reversed by lenalidomide, with improved phagocytotic activity, cytokine production, T-cell stimulatory and proliferative activity.79 Lenalidomide has produced clinical responses as monotherapy,74 in combination with rituximab or with chemotherapy,80 and as maintenance following chemotherapy.73 However, increased risk of toxicities and infections with treatment remains a concern,73 potentially reflecting potent activation of the immune TME with this class of drug. Thus, the place and dosing regimen for lenalidomide in clinical practice remain unclear. A novel option emerging for CLL therapy are next-generation cereblon E3 ligase modulators (CELMoD), with avadomide recently investigated in a preclinical study. Avadomide stimulated T-cell activation, the expression of immunostimulatory chemokines, and the formation of lytic synapses with CLL cells by triggering inflammatory IFN type I and II signaling in previously exhausted T cells from patients.81 The potential and optimal roles of IMiD and CELMoD in the context of the CLL-TME remain to be determined; however, the favorable immunomodulatory effects on the T-cell/NK-cell compartments imply a role for IMiD and CELMoD in 2318

developing novel combination treatment strategies. The most important effects of IMiD/CELMoD on the TME are summarized in Table 1 and illustrated in Figure 3.

Immune checkpoint blockade The PD-1:PD-L1 is an immune checkpoint pathway used by tumor cells to inhibit T cells and escape immune surveillance. Thus, this pathway constitutes an attractive therapeutic target (Figure 4).82 Blocking PD-L1 in CLLtransplanted mice resulted in repressed disease development and restored T-cell immune effector functions including improved cytotoxicity, cytokine production, and immune synapse formation.83 Despite this, the sparse clinical data on immune checkpoint blockade (ICB) in CLL are disappointing. In a phase II study of the PD-1 blocking antibody drug, pembrolizumab, four out of nine patients with Richter transformation showed clinical response to treatment, whereas none of the 16 CLL patients responded.84 The clinical efficacy of ICB-based therapy correlates with upregulated levels of tumor PDL1 expression that is associated with an “inflamed” microenvironment with the presence of activated cytotoxic tumor-infiltrating T cells attempting to engage tumor cells, which can be unleashed as checkpoint inhibitory signals are abrogated.85 PD-L1 expression on CLL cells is relatively low, likely reflecting low activity of cytolytic T cells.81,82 Furthermore, the immunosuppressive state of the TME in CLL, with profoundly exhausted effector T cells that exhibit multiple functional defects, likely contributes significantly to the lack of clinical response to checkpoint inhibitor monotherapy in CLL patients. Consistent with this, a recent study of patients’ lymph node biopsies has provided evidence for a noninflamed microenvironment in CLL, incorporating low numbers of CD8+ T cells, low PD-L1 expression and profound T-cell exhaustion.81 Thus, strategies that can subvert the strong immunosuppressive pressure of the CLLTME and overcome T-cell dysfunction may be necessary to sensitize CLL to ICB immunotherapy and develop therapeutic options for CLL patients. Further research to unravel the complex immunosuppression in the CLLTME is warranted in order to develop and optimize immuno-oncology treatments.

T-cell-based therapy Chimeric antigen receptor (CAR) T cells have emerged as a powerful therapeutic option designed to transfer high numbers of tumor-targeted effector T cells into the TME to overcome a paucity of endogenous cytolytic T cells. Briefly, autologous T cells are genetically modified to express CAR with specificity for specific tumor antigens, such as CD19 in B-cell malignancies, thus creating an adoptive T-cell-mediated cytotoxic response (Figure 4).86 CAR T cells combine the effects of T-cell and antibodymediated immune responses by triggering T-cell activation with granule exocytosis upon antigen binding.87 Despite the first successful CAR T-cell trial being reported in CLL, few clinical trials have subsequently reported efficacy of CAR T cells in CLL.88 CLL-induced T-cell dysfunction, as well as understudied lymphoid TME barriers, likely reduce the efficacy of this approach in CLL. A haematologica | 2021; 106(9)

Targeting the tumor microenvironment in CLL

recent study revealed that CAR T cells from CLL patients responding well to CAR T-cell therapy expressed upregulated genes associated with T-cell memory. Furthermore, enriched T-memory subsets prior to CAR T-cell generation correlated with sustained remissions. Meanwhile, CAR T cells from non-responders had upregulated genes associated with effector T-cell differentiation, apoptosis and exhaustion, thus emphasizing that T-cell fitness is crucial for the efficacy of CAR T cells.89 Due to the multitude of (successful) treatment options for CLL currently, CAR T-cell therapy may first become a relevant option in treating multi-relapsed disease, and preliminary reports from current clinical trials of CD19-targeted CAR T-cell therapy in patients with multi-relapsed CLL show somewhat encouraging results.90 However, paradoxically, Tcell exhaustion in CLL is demonstrated to worsen with progressive disease,13 thus pointing towards a need for options that improve T-cell function prior to the manufacture of CAR T cells or during treatment. Furthermore, it was recently elucidated that CLL cells can directly impair CAR T-cell function and induce an exhausted phenotype through the release of plasma extracellular vesicles.91 Thus, a meaningful role for CAR T-cell therapy in CLL may rely on the ability of current and/or future therapies to successfully target the TME and improve T-cell fitness in patients with CLL, prior to the CAR T-cell treatment, during preparation of the product, and after its administration. A novel therapeutic approach that could constitute an alternative to CAR T-cell therapy is off-the-shelf bispecific CD19/CD3 or CD20/CD3 antibody treatment.

Bispecific antibodies simultaneously engage CD3 on T cells and CD19 or CD20 on target B cells, and thereby redirect T cells to recognize CLL cells, facilitating synapse formation and, thus, T-cell-mediated antitumor responses (Figure 4). Preclinical studies using bispecific antibodies have demonstrated antileukemic activity against CLL cells in vitro and in xenograft models.92 Thus, bispecific antibodies may constitute a promising T-cell-based immunotherapeutic approach for CLL, alone, or in combination with TME-modulating agents that help improve T-cell function.

Developing combination strategies targeting the chronic lymphocytic leukemia – tumor microenvironment It is becoming evident that improving clinical responses (residual and progressive disease), overcoming toxicity, infection risk, as well as drug resistance, likely require strategies aimed at reshaping the immunosubversive, protumor TME state. Our improved understanding of the direct and indirect CLL-TME modulations by novel therapeutic agents in recent years provides a unique opportunity to optimize CLL treatment with strategic drug combinations that target multiple CLL-TME interactions to achieve therapeutic synergy while controlling toxicity. Monoclonal antibodies targeting the B-cell surface protein CD20 have been the backbone of standard chemoimmunotherapy regimes used to treat CLL for decades, although they are rarely used as a monotherapy in CLL.1

Figure 3. Effects of BCL-2 inhibitors, immunomodulatory drugs, and cereblon E3 ligase modulation on the tumor microenvironment. Inhibitory effects are represented by bars, stimulatory effects are represented by arrows. Upward arrows indicate increases, downward arrows indicate decreases. CLL: chronic lymphocytic leukemia; TME: tumor microenvironment; BCL-2: B-cell lymphoma 2; BCL-2i: BCL-2 inhibitor; IMiD: immunomodulatory drug; CELMoD: cereblon E3 ligase modulator; TCR: T-cell receptor; HLA-DR: human leukocyte antigen DR-isotype; IFN: interferon; PD-1: programmed cell death protein 1; PD-1L: programmed cell death protein-1 ligand; CCL: chemokine ligand.

haematologica | 2021; 106(9)

2319

R. Svanberg et al.

Figure 4. Effects of immune checkpoint blockade, chimeric antigen receptor T cells, and bispecific antibodies. Arrows point out the actions of immune checkpoint blockade, chimeric antigen receptor T cells and bispecific antibodies, leading to engagement and activation of the tumor microenvironment for anti-chronic lymphocytic leukemia activity. Upward arrows indicate increases. CLL: chronic lymphocytic leukemia; PD-1: programmed cell death protein 1; PD-1L: programmed cell death protein-1 ligand; Ab: antibody; CAR-T: chimeric antigen receptor T cells; bsAB: bispecific antibodies; TCR: T-cell receptor; CD: cluster of differentiation.

Major mechanisms of action of anti-CD20 antibodies are activation of antibody-dependent cellular cytotoxicity and antibody-dependent cellular phagocytosis, which rely on engaging the antitumor activity of NK cells and monocytes/macrophages within the immune TME.1,48,93 The direct inhibitory effects of ibrutinib on macrophage phagocytosis and NK-cell activation may therefore interfere with the therapeutic efficacy of anti-CD20 treatment.48 Compared to ibrutinib as a single agent, adding an anti-CD20 antibody to ibrutinib was associated with faster remissions and lower levels of residual disease in a clinical trial, although it was not demonstrated that the combination improved progression-free survival.94 Thus, whether this combination is beneficial remains debatable. In contrast, combining anti-CD20 antibodies with venetoclax seems to improve the phagocytosis of CLL cells by macrophages in vitro,93 and reverse venetoclax resistance induced by TME signaling.71 Interestingly, although venetoclax plus anti-CD20 treatment produces impressive clinical responses in clinical trials, a recent retrospective study including real-world data demonstrated comparable efficacy between venetoclax as a single agent and venetoclax plus anti-CD20 combination treatment in high risk relapsed/refractory CLL patients.95 Thus, further validating prospective studies are warranted to determine whether the addition of an anti-CD20 antibody to venetoclax is truly necessary. The addition of venetoclax to ibrutinib constitutes another approach aiming to provide improved duration and depth of remissions as well as to enable fixed-duration treatment, which has already, in part, demonstrated success in clinical trials.96 Similarly, the PI3K inhibitor, duvelisib, increases sensitivity of CLL cells to venetoclax, providing the rationale for duvelisib-venetoclax combination treatment, currently being investigated in clinical trials.97 However, the biggest challenges ahead involve finding 2320

strategic combinations that overcome T-cell dysfunction, improve the efficacy of T-cell-based therapies and ICB in CLL, and work towards curative therapy. Data from a human xenograft model support the ability of ibrutinib to enhance CAR T-cell function when administered concurrently.98 Similarly, another murine study indicated that PI3Kδ inhibition during CAR T-cell production may have a positive effect on engraftment and antitumor activity.99 Consistently with this, a clinical pilot study recently demonstrated high response rates in relapsed/refractory CLL patients receiving ibrutinib concomitant with CD19targeted CAR T-cell therapy, and lower toxicities compared to those in patients treated without concomitant ibrutinib.100 Furthermore, T cells from ibrutinib-treated CLL patients seem to exhibit improved in vitro anti-CLL activity combined with bispecific antibodies.101 The lack of clinical activity of anti-PD-1 monotherapy in CLL,84 has highlighted the need to incorporate ICB therapies into more powerful combinations to unleash the power of antitumor immune cells. Studies of PD1:PD-L1 blockade combined with ibrutinib have demonstrated enhanced CD8+ T-cell function along with improved disease control in a CLL mouse model.102 However, preliminary clinical results have indicated that coupling anti-PD-1 with ibrutinib may not increase response rates in patients.103 PI3K inhibition improved the anticancer effect of ICB through modulatory effects on MDSC in a solid cancer in vitro model,104 thus highlighting additional roles of PI3K inhibition in modulation of the TME which could be exploited. The relative expansion of memory T-cell subsets due to direct effects of venetoclax on other, more BCL-2-dependent T-cell subsets, support a role for venetoclax in combination with ICB. In a recent sold cancer murine study, venetoclax augmented the antitumor effect of anti-PD-1 and anti-PD-L1 inhibitors in vivo.69 haematologica | 2021; 106(9)

Targeting the tumor microenvironment in CLL

Through their potent activation of T cells, CELMoD and IMiD could represent strong complementary treatment partners for combination (immune)therapy.14,15,73,75,76,80,81 It has been demonstrated preclinically that the CELMoD avadomide could sensitize CLL to anti-PD-1 or anti-PD-L1 immunotherapy.81 By inducing inflammatory interferon type I and II signaling in previously exhausted T cells from patients, avadomide stimulated the proliferation and release of chemokines by T cells which recruited additional CD8+ T cells, upregulated PD-L1 in the immune TME, and enhanced lytic synapse formation.81 Even more powerful combinations could include pairing ICB with CAR T cells or bispecific antibodies to increase tumor infiltrating T cells, or dual ICB combinations to overcome additional inhibitory barriers. T-cell bispecific antibodies combined with an anti-PD-L1 antibody showed enhanced antitumor efficacy compared to either given alone in a solid cancer mouse model.105 Furthermore, a recent CLL murine study demonstrated that anti-PD1 ICB combined with inhibition of the immune checkpoint receptor lymphocyte-antigen gene 3 (LAG3) was able to decrease tumor load significantly, while either as monotherapy had little effect.106 Thus, developing combination immunotherapy could represent a powerful strategy for deepening targeted drug (e.g., BTK inhibitor- and/or venetoclax)-induced responses and working towards curative therapy in CLL.

Future perspectives: novel targetable tumor microenvironment interactions The CLL-TME constitutes a landscape of potential targetable pathways. Antibodies interrupting the CXCL12/CXCR4 interaction have demonstrated anti-CLL activity in vitro and in mouse models, and have been tested in phase I clinical trials for multiple myeloma, but have not yet been further explored in CLL.107 The BAFF/BAFF-R axis constitutes another attractive CLL-TME interaction to target. An anti-BAFF-R antibody blocked protective survival signaling in CLL cells and enhanced antibody-dependent cellular cytotoxicity in vitro, and also enhanced efficacy of ibrutinib in a CLL mouse model.108 Targeting of an IL10-producing CD38 regulatory B cell-like CLL subset is also currently under investigation.27 Although aimed at CLL cells, the anti-leukemic potential here would be mediated indirectly by abrogating IL-10-mediated immunosuppression. The “don’t eat me” signal regulatory protein (SIRP)1α/CD47 axis, co-expressed by macrophages and malignant cells, respectively, in various lymphoid malignancies including CLL, constitutes a mechanism of hi

References 1. Hallek M. Chronic lymphocytic leukemia: 2020 update on diagnosis, risk stratification and treatment. Am J Hematol. 2019;94(11):1266-1287. 2. Herishanu Y, Pérez-Galán P, Liu D, et al. The lymph node microenvironment promotes Bcell receptor signaling, NF-κB activation, and tumor proliferation in chronic lymphocytic leukemia. Blood. 2011;117(2):563-574. 3. Burger JA, Ghia P, Rosenwald A, CaligarisCappio F. The microenvironment in mature

haematologica | 2021; 106(9)

myeloid immune tolerance. It is currently being explored as a potential target in lymphoma, and may also constitute an important macrophage-CLL interaction with targetable potential.109 Furthermore, MDSC-derived indoleamine 2,3dioxygenase (IDO) has been explored as a target in other cancers, as an antitumor vaccination antigen.110 Mediating strong immunosuppressive effects in the CLL-TME, IDO may constitute another potentially attractive interaction to target in CLL. In conclusion, the heterogeneous course of CLL is driven by (i) the genetic complexity and (ii) diverse and complex tissue TME interactions, including (iii) antigenic drive. The functionally and phenotypically skewed cell types within the TME not only promote CLL itself, but also compromise the induction of adequate immune responses towards developing and progressing CLL clones, as well as infectious agents. Thus, further exploration of the impact of different therapies on critical microenvironmental interactions is warranted. This review was intended to outline our current understanding of how CLL-TME interactions are influenced by current CLL therapies, with the view to encourage continued mapping of the CLL-TME for more specific future targeting of CLL-TME crosstalk. The inclusion of translational correlative studies assessing immunological and TME changes within clinical trials should inform development of novel combination therapies beyond BTK inhibitors and BCL-2 inhibitors. These include, but are not limited to, checkpoint inhibitors, T-cell based therapies, and TME modulation overcoming the inherent immune exhaustion in CLL. Such strategies together with further understanding of the TME should, eventually, lead to improved and more personalized treatment options aiming for a clinical cure with reduced burden of adverse events for patients with CLL. Disclosures None of the authors received funds or any other direct contribution to support the submitted work. There are no patents or copyrights relevant to this work. RS received research support from AstraZeneca, and travel grants from Abbvie outside of this work. CN received research support and/or personal consultancy fees from Abbvie, AstraZeneca, Janssen, Roche, Danish Cancer Society and Novo Nordisk Foundation (grant NNF16OC0019302) outside of this work. SJ, AR and PP have no conflicts of interest to disclose. Contributions RS and SJ wrote the first draft of the review under supervision of CN; the final version of the review was written by all five authors. All authors approved the final version.

B-cell malignancies: a target for new treatment strategies. Blood. 2009;114(16):33673375. 4. Dubois N, Crompot E, Meuleman N, Bron D, Lagneaux L, Stamatopoulos B. Importance of crosstalk between chronic lymphocytic leukemia cells and the stromal microenvironment: direct contact, soluble factors, and extracellular vesicles. Front Oncol. 2020;10:1-19. 5. Caligaris-Cappio F, Bertilaccio MTS, Scielzo C. How the microenvironment wires the natural history of chronic lymphocytic

leukemia. Semin Cancer Biol. 2014;24:4348. 6. George B, Chowdhury SM, Hart A, et al. Ibrutinib resistance mechanisms and treatment strategies for B-cell lymphomas. Cancers (Basel). 2020;12(5):1328. 7. Boissard F, Fournié JJ, Quillet-Mary A, Ysebaert L, Poupot M. Nurse-like cells mediate ibrutinib resistance in chronic lymphocytic leukemia patients. Blood Cancer J. 2015;5(10):e355. 8. Chen JG, Liu X, Munshi M, et al. BTK Cys481Ser drives ibrutinib resistance via

2321

R. Svanberg et al. ERK1/2 and protects BTK wild-type MYD88-mutated cells by a paracrine mechanism. Blood. 2018;131(18):2047-2059. 9. Davids MS, Deng J, Wiestner A, et al. Decreased mitochondrial apoptotic priming underlies stroma-mediated treatment resistance in chronic lymphocytic leukemia. Blood. 2012;120(17):3501-3509. 10. Roessner PM, Seiffert M. T-cells in chronic lymphocytic leukemia: Guardians or drivers of disease? Leukemia. 2020;34(8):20122024. 11. Os A, Bürgler S, Ribes AP, et al. Chronic lymphocytic leukemia cells are activated and proliferate in response to specific T helper cells. Cell Rep. 2013;4(3):566-577. 12. van Attekum MHA, van Bruggen JAC, Slinger E, et al. CD40 signaling instructs chronic lymphocytic leukemia cells to attract monocytes via the CCR2 axis. Haematologica. 2017;102(12):2069-2076. 13. Palma M, Gentilcore G, Heimersson K, et al. T cells in chronic lymphocytic leukemia display dysregulated expression of immune checkpoints and activation markers. Haematologica. 2017;102(3):562-572. 14. Ramsay AG, Johnson AJ, Lee AM, et al. Chronic lymphocytic leukemia T cells show impaired immunological synapse formation that can be reversed with an immunomodulating drug. J Clin Invest. 2008;118(7):24272437. 15. Ramsay AG, Clear AJ, Fatah R, Gribben JG. Multiple inhibitory ligands induce impaired T-cell immunologic synapse function in chronic lymphocytic leukemia that can be blocked with lenalidomide: establishing a reversible immune evasion mechanism in human cancer. Blood. 2012;120(7):14121421. 16. Riches JC, Davies JK, McClanahan F, et al. T cells from CLLpatients exhibit features of Tcell exhaustion but retain capacity for cytokine production. Blood. 2013;121(9): 1612-1621. 17. Giannopoulos K, Schmitt M, Kowal M, Wlasiuk P, Bojarska-Junka A, Dmoszynska A. Characterization of regulatory T cells in patients with B-cell chronic lymphocytic leukemia. Oncol Rep. 2008;20(3):677-682. 18. Burger JA, Tsukada N, Burger M, Zvaifler NJ, Dell’Aquila M, Kipps TJ. Blood-derived nurse-like cells protect chronic lymphocytic leukemia B cells from spontaneous apoptosis through stromal cell-derived factor-1. Blood. 2000;96(8):2655-2663. 19. Burger JA, Burger M, Kipps TJ. Chronic lymphocytic leukemia B cells express functional CXCR4 chemokine receptors that mediate spontaneous migration beneath bone marrow stromal cells. Blood. 1999;94(11):36583667. 20. Borge M, Nannini PR, Morande PE, et al. CXCL12 is a costimulator for CD4+ T cell activation and proliferation in chronic lymphocytic leukemia patients. Cancer Immunol Immunother. 2013;62(1):113-124. 21. Filip AA, Ciseł B, Koczkodaj D, Wa̧sikSzczepanek E, Piersiak T, Dmoszyńska A. Circulating microenvironment of CLL: are nurse-like cells related to tumor-associated macrophages? Blood Cells Mol Dis. 2013;50(4):263-270. 22. Gustafson MP, Abraham RS, Lin Y, et al. Association of an increased frequency of CD14+ HLA-DR lo/neg monocytes with decreased time to progression in chronic lymphocytic leukaemia (CLL). Br J Haematol. 2012;156(5):674-676. 23. Jitschin R, Braun M, Büttner M, et al. CLLcells induce IDOhi CD14+HLA-DRlo

2322

myeloid-derived suppressor cells that inhibit T-cell responses and promote TRegs. Blood. 2014;124(5):750-760. 24. Lagneaux L, Delforge A, Bron D, De Bruyn C, Stryckmans P. Chronic lymphocytic leukemic B cells but not normal B cells are rescued from apoptosis by contact with normal bone marrow stromal cells. Blood. 1998;91(7):2387-2396. 25. Burger JA, Wiestner A. Targeting B cell receptor signalling in cancer: preclinical and clinical advances. Nat Rev Cancer. 2018;18(3):148-167. 26. Nishio M, Endo T, Tsukada N, et al. Nurselike cells express BAFF and APRIL, which can promote survival of chronic lymphocytic leukemia cells via a paracrine pathway distinct from that of SDF-1α. Blood. 2005;106(3):1012-1020. 27. Manna A, Kellett T, Aulakh S, et al. Targeting CD38 is lethal to Breg-like chronic lymphocytic leukemia cells and Tregs, but restores CD81 T-cell responses. Blood Adv. 2020;4(10):2143-2157. 28. Alhakeem SS, McKenna MK, Oben KZ, et al. Chronic lymphocytic leukemia–derived IL-10 suppresses antitumor immunity. J Immunol. 2018;200(12):4180-4189. 29. Smallwood DT, Apollonio B, Willimott S, et al. Extracellular vesicles released by CD40/IL-4-stimulated CLL cells confer altered functional properties to CD4+ T cells. Blood. 2016;128(4):542-552. 30. Ravandi F, O’Brien S. Immune defects in patients with chronic lymphocytic leukemia. Cancer Immunol Immunother. 2006;55(2):197-209. 31. Sharman JP, Egyed M, Jurczak W, et al. Acalabrutinib with or without obinutuzumab versus chlorambucil and obinutuzmab for treatment-naive chronic lymphocytic leukaemia (ELEVATE TN): a randomised, controlled, phase 3 trial. Lancet. 2020;395(10232):1278-1291. 32. Tam CS, Robak T, Ghia P, et al. Efficacy and safety of zanubrutinib in patients with treatment-naive chronic lymphocytic leukemia (CLL) or small lymphocytic lymphoma (SLL) with del(17p): initial results from arm C of the Sequoia (BGB-3111-304) trial. Blood. 2019;134(Suppl_1):499. 33. Niemann CU, Herman SEM, Maric I, et al. Disruption of in vivo chronic lymphocytic leukemia tumor-microenvironment interactions by ibrutinib - findings from an investigator-initiated phase II study. Clin Cancer Res. 2016;22(7):1572-1582. 34. Yin Q, Sivina M, Robins H, et al. Ibrutinib therapy increases T cell repertoire diversity in patients with chronic lymphocytic leukemia. J Immunol. 2017;198(4):17401747. 35. Long M, Beckwith K, Do P, et al. Ibrutinib treatment improves T cell number and function in CLL patients. J Clin Invest. 2017;127(8):3052-3064. 36. Solman IG, Blum LK, Hoh HY, et al. Ibrutinib restores immune cell numbers and function in first-line and relapsed/refractory chronic lymphocytic leukemia. Leuk Res. 2020;97:106432. 37. Baptista MJ, Basumallik N, Herman SEM, et al. Ibrutinib increases the clonality of TCR repertoire in patients with chronic lymphocytic leukemia. Blood. 2018;132(Suppl 1):238. 38. Dubovsky JA, Beckwith KA, Natarajan G, et al. Ibrutinib is an irreversible molecular inhibitor of ITK driving a Th1-selective pressure in T lymphocytes. Blood. 2013;122 (15):2539-2549.

39. Herman SEM, Montraveta A, Niemann CU, et al. The Bruton tyrosine kinase (BTK) inhibitor acalabrutinib demonstrates potent on-target effects and efficacy in two mouse models of chronic lymphocytic leukemia. Clin Cancer Res. 2017;23(11):2831-2841. 40. Zou YX, Zhu HY, Li XT, et al. The impacts of zanubrutinib on immune cells in patients with chronic lymphocytic leukemia/small lymphocytic lymphoma. Hematol Oncol. 2019;37(4):392-400. 41. Papazoglou D, Lesnick CE, Wang V, Kay NE, Shanafelt TD, Ramsay AG. Ibrutinib-based therapy improves anti-tumor T cell killing function allowing effective pairing with anti-PD-L1 immunotherapy compared to traditional FCR chemoimmunotherapy; implications for therapy and correlative immune functional data from the phase III. Blood. 2018;132(Suppl 1):236. 42. De Weerdt I, Hofland T, Lameris R, et al. Improving CLL Vγ9Vδ2-T-cell fitness for cellular therapy by ex vivo activation and ibrutinib. Blood 2018;132(21):2260-2272. 43. Podhorecka M, Goracy A, Szymczyk A, et al. Changes in T-cell subpopulations and cytokine network during early period of ibrutinib therapy in chronic lymphocytic leukemia patients: the significant decrease in T regulatory cells number. Oncotarget. 2017;8(21):34661-34669. 44. Jadidi-Niaragh F, Ghalamfarsa G, Memarian A, et al. Downregulation of IL-17-producing T cells is associated with regulatory T cell expansion and disease progression in chronic lymphocytic leukemia. Tumor Biol. 2013;34(2):929-940. 45. Herman SEM, Niemann CU, Farooqui M, et al. Ibrutinib-induced lymphocytosis in patients with chronic lymphocytic leukemia: correlative analyses from a phase II study. Leukemia. 2014;28(11):2188-2196. 46. Ping L, Ding N, Shi Y, et al. The Bruton’s tyrosine kinase inhibitor ibrutinib exerts immunomodulatory effects through regulation of tumorinfiltrating macrophages. Oncotarget. 2017;8(24):39218-39229. 47. De Rooij MFM, Kuil A, Geest CR, et al. The clinically active BTK inhibitor PCI-32765 targets B-cell receptor- and chemokine-controlled adhesion and migration in chronic lymphocytic leukemia. Blood. 2012;119(11): 2590-2594. 48. Da Roit F, Engelberts PJ, Taylor RP, et al. Ibrutinib interferes with the cell-mediated anti-tumor activities of therapeutic CD20 antibodies: implications for combination therapy. Haematologica. 2015;100(1):77-86. 49. Stiff A, Trikha P, Wesolowski R, et al. Myeloid-derived suppressor cells express Bruton’s tyrosine kinase and can be depleted in tumor-bearing hosts by ibrutinib treatment. Cancer Res. 2016;76(8):2125-2136. 50. Herman SEM, Mustafa RZ, Jones J, Wong DH, Farooqui M, Wiestner A. Treatment with ibrutinib inhibits BTK- and VLA-4dependent adhesion of chronic lymphocytic leukemia cells in vivo. Clin Cancer Res. 2015;21(20):4642-4651. 51. Sun C, Tian X, Lee YS, et al. Partial reconstitution of humoral immunity and fewer infections in patients with chronic lymphocytic leukemia treated with ibrutinib. Blood. 2015;126(19):2213-2219. 52. Xiao Y, Zou P, Wang J, Song H, Zou J, Liu L. Lower phosphorylation of p38 MAPK blocks the oxidative stress-induced senescence in myeloid leukemic CD34(+)CD38 () cells. J Huazhong Univ Sci Technolog Med Sci. 2012;32(3):328-333. 53. Aarup K, Rotbain EC, Enggaard L, et al. Real-

haematologica | 2021; 106(9)

Targeting the tumor microenvironment in CLL

world outcomes for 205 patients with chronic lymphocytic leukemia treated with ibrutinib. Eur J Haematol. 2020;105(5):646654. 54. Lannutti BJ, Meadows SA, Herman SEM, et al. CAL-101, a p110δ selective phosphatidylinositol-3-kinase inhibitor for the treatment of B-cell malignancies, inhibits PI3K signaling and cellular viability. Blood. 2011;117(2):591-594. 55. Herman SEM, Gordon AL, Wagner AJ, et al. Phosphatidylinositol 3-kinase-δ inhibitor CAL-101 shows promising preclinical activity in chronic lymphocytic leukemia by antagonizing intrinsic and extrinsic cellular survival signals. Blood. 2010;116(12):20782088. 56. Hoellenriegel J, Meadows SA, Sivina M, et al. The phosphoinositide 3′-kinase delta inhibitor, CAL-101, inhibits B-cell receptor signaling and chemokine networks in chronic lymphocytic leukemia. Blood. 2011;118(13):3603-3612. 57. Ali K, Soond DR, Piñeiro R, et al. Inactivation of PI(3)K p110δ breaks regulatory T-cell-mediated immune tolerance to cancer. Nature. 2014;510(7505):407-411. 58. Hanna BS, Roessner PM, Scheffold A, et al. PI3Kδ inhibition modulates regulatory and effector T-cell differentiation and function in chronic lymphocytic leukemia. Leukemia. 2019;33(6):1427-1438. 59. Brown JR, Byrd JC, Coutre SE, et al. Idelalisib, an inhibitor of phosphatidylinositol 3-kinase p110δ, for relapsed/refractory chronic lymphocytic leukemia. Blood. 2014;123(22):3390-3397. 60. Alflen A, Stadler N, Aranda Lopez P, et al. Idelalisib impairs TREM-1 mediated neutrophil inflammatory responses. Sci Rep. 2018;8(1):1-10. 61. O’Brien SM, Lamanna N, Kipps TJ, et al. A phase 2 study of idelalisib plus rituximab in treatment-naïve older patients with chronic lymphocytic leukemia. Blood. 2015;126(25): 2686-2694. 62. Flinn IW, Hillmen P, Montillo M, et al. The phase 3 DUO trial: duvelisib vs ofatumumab in relapsed and refractory CLL/SLL. Blood. 2018;132(23):2446-2455. 63. Maharaj K, Powers JJ, Achille A, et al. The dual PI3Kδ/CK1e inhibitor umbralisib exhibits unique immunomodulatory effects on CLL T cells. Blood Adv. 2020;4(13):30723084. 64. Valentin R, Grabow S, Davids MS. The rise of apoptosis: targeting apoptosis in hematologic malignancies. Blood. 2018;132(12): 1248-1264. 65. 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. 66. Roberts AW, Ma S, Kipps TJ, et al. Efficacy of venetoclax in relapsed chronic lymphocytic leukemia is influenced by disease and response variables. Blood. 2019;134(2):111122. 67. Leverson JD, Phillips DC, Mitten MJ, et al. Exploiting selective BCL-2 family inhibitors to dissect cell survival dependencies and define improved strategies for cancer therapy. Sci Transl Med. 2015;7(279):279ra40. 68. Wojciechowski S, Tripathi P, Bourdeau T, et al. Bim/Bcl-2 balance is critical for maintaining naive and memory T cell homeostasis. J Exp Med. 2007;204(7):1665-1675. 69. Kohlhapp FJ, Haribhai D, Mathew R, et al. Venetoclax increases intratumoral effector T cells and antitumor efficacy in combination

haematologica | 2021; 106(9)

with immune checkpoint blockade. Cancer Discov. 2020;11(2):68-79. 70. De Weerdt I, Hofland T, De Boer R, et al. Distinct immune composition in lymph node and peripheral blood of CLL patients is reshaped during venetoclax treatment. Blood Adv. 2019;3(17):2642-2652. 71. Thijssen R, Slinger E, Weller K, et al. Resistance to ABT-199 induced by microenvironmental signals in chronic lymphocytic leukemia can be counteracted by CD20 antibodies or kinase inhibitors. Haematologica. 2015;100(8):e302-305. 72. Chanan-Khan AA, Chitta K, Ersing N, et al. Biological effects and clinical significance of lenalidomide-induced tumour flare reaction in patients with chronic lymphocytic leukaemia: in vivo evidence of immune activation and antitumour response. Br J Haematol. 2011;155(4):457-467. 73. Fink AM, Bahlo J, Robrecht S, et al. Lenalidomide maintenance after first-line therapy for high-risk chronic lymphocytic leukaemia (CLLM1): final results from a randomised, double-blind, phase 3 study. Lancet Haematol. 2017;4(10):e475-e486. 74. Chen C, Paul H, Wang T, et al. Long-term follow-up of a phase 2 trial of single agent lenalidomide in previously untreated patients with chronic lymphocytic leukaemia. Br J Haematol. 2014;165(5):731733. 75. Ramsay AG, Evans R, Kiaii S, Svensson L, Hogg N, Gribben JG. Chronic lymphocytic leukemia cells induce defective LFA-1-directed T-cell motility by altering Rho GTPase signaling that is reversible with lenalidomide. Blood. 2013;121(14):2704-2714. 76. Ramsay AG, Gribben JG. Immune dysfunction in chronic lymphocytic leukemia T cells and lenalidomide as an immunomodulatory drug. Haematologica. 2009;94(9):1198-1202. 77. Browning RL, Byrd WH, Gupta N, et al. Lenalidomide induces interleukin-21 production by T cells and enhances IL21-mediated cytotoxicity in chronic lymphocytic leukemia B cells. Cancer Immunol Res. 2016;4(8):697-707. 78. Aue G, Sun C, Liu D, et al. Activation of Th1 immunity within the tumor microenvironment is associated with clinical response to lenalidomide in chronic lymphocytic leukemia. J Immunol. 2018;201(7):19671974. 79. Fiorcari S, Martinelli S, Bulgarelli J, et al. Lenalidomide interferes with tumor-promoting properties of nurse-like cells in chronic lymphocytic leukemia. Haematologica. 2015;100(2):253-262. 80. Thompson PA, Keating MJ, Hinojosa C, et al. Lenalidomide and rituximab in combination as initial treatment of chronic lymphocytic leukemia: initial results of a phase II study. Blood. 2014;124(21):1988. 81. Ioannou N, Hagner PR, Stokes M, et al. Triggering interferon signaling in T cells with avadomide sensitizes CLL to anti-PDL1/PD-1 immunotherapy. Blood. 2020;137 (2):216-231. 82. Brusa D, Serra S, Coscia M, et al. The PD1/PD-L1 axis contributes to T-cell dysfunction in chronic lymphocytic leukemia. Haematologica. 2013;98(6):953-963. 83. McClanahan F, Hanna B, Miller S, et al. PDL1 checkpoint blockade prevents immune dysfunction and leukemia development in a mouse model of chronic lymphocytic leukemia. Blood. 2015;126(2):203-211. 84. Ding W, LaPlant BR, Call TG, et al. Pembrolizumab in patients with CLL and Richter transformation or with relapsed

CLL. Blood. 2017;129(26):3419-3427. 85. Yan X, Zhang S, Deng Y, Wang P, Hou Q, Xu H. Prognostic factors for checkpoint inhibitor based immunotherapy: an update with new evidences. Front Pharmacol. 2018;9:1050. 86. Kalos M, Levine BL, Porter DL, et al. T cells with chimeric antigen receptors have potent antitumor effects and can establish memory in patients with advanced leukemia. Sci Transl Med. 2011;3(95):95ra73. 87. Laurin D, Marin V, Biagi E, et al. Exploration of the lysis mechanisms of leukaemic blasts by chimaeric T-cells. J Biomed Biotechnol. 2010;2010:234540. 88. Porter DL, Levine BL, Kalos M, Bagg A, June CH. Chimeric antigen receptor–modified T cells in chronic lymphoid leukemia. N Engl J Med. 2011;365(8):725-733. 89. Fraietta JA, Lacey SF, Orlando EJ, et al. Determinants of response and resistance to CD19 chimeric antigen receptor (CAR) T cell therapy of chronic lymphocytic leukemia. Nat Med. 2018;24(5):563-571. 90. Siddiqi T, Soumerai JD, Dorritie KA, et al. Rapid undetectable MRD (uMRD) responses in patients with relapsed/refractory (R/R) chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL) treated with lisocabtagene maraleucel (liso-cel), a CD19directed CAR T cell product: updated results from transcend CLL 004, a phase 1/2 study including patients with high-risk disease previously treated with ibrutinib. Blood. 2019;134(Suppl_1):503. 91. Cox MJ, Lucien F, Sakemura R, et al. Leukemic extracellular vesicles induce chimeric antigen receptor T cell dysfunction in chronic lymphocytic leukemia. Mol Ther. 2021;29(5):1918-1932. 92. Robinson HR, Qi J, Cook EM, et al. A CD19/CD3 bispecific antibody for effective immunotherapy of chronic lymphocytic leukemia in the ibrutinib era. Blood. 2018;132(5):521-532. 93. Elías EE, Almejún MB, Colado A, et al. Autologous T-cell activation fosters ABT199 resistance in chronic lymphocytic leukemia: rationale for a combined therapy with SYK inhibitors and anti-CD20 monoclonal antibodies. Haematologica. 2018;103 (10):e458-e461. 94. Burger JA, Sivina M, Jain N, et al. Randomized trial of ibrutinib vs ibrutinib plus rituximab in patients with chronic lymphocytic leukemia. Blood. 2019;133(10): 1011-1019. 95. Mato AR, Roeker LE, Eyre TA, et al. A retrospective comparison of venetoclax alone or in combination with an anti-CD20 monoclonal antibody in R/R CLL. Blood Adv. 2019;3(10):1568-1573. 96. Niemann CU, Levin M-D, Dubois J, et al. Venetoclax and ibrutinib for patients with relapsed/refractory chronic lymphocytic leukemia. Blood. 2020;137(8):1117-1120. 97. Patel VM, Balakrishnan K, Douglas M, et al. Duvelisib treatment is associated with altered expression of apoptotic regulators that helps in sensitization of chronic lymphocytic leukemia cells to venetoclax (ABT199). Leukemia. 2017;31(9):1872-1881. 98. Fraietta JA, Beckwith KA, Patel PR, et al. Ibrutinib enhances chimeric antigen receptor T-cell engraftment and efficacy in leukemia. Blood. 2016;127(9):1117-1127. 99. Stock S, Übelhart R, Schubert ML, et al. Idelalisib for optimized CD19-specific chimeric antigen receptor T cells in chronic lymphocytic leukemia patients. Int J Cancer. 2019;145(5):1312-1324.

2323

R. Svanberg et al. 100. Gauthier J, Hirayama AV, Purushe J, et al. Feasibility and efficacy of CD19-targeted CAR T cells with concurrent ibrutinib for CLL after ibrutinib failure. Blood. 2020;135(19):1650-1660. 101. Long M, Williams E, Berard C, et al. Ibrutinib treatment in CLL patients improves T cell function and blinatumomab redirected cytotoxicity. Blood. 2019;134 (Suppl_1):1049. 102. Hanna BS, Yazdanparast H, Demerdash Y, et al. Combining ibrutinib and checkpoint blockade improves CD8+ T-cell function and control of chronic lymphocytic leukemia in Eμ-TCL1 mice. Haematologica. 2021;106(4):968-977. 103. Younes A, Brody J, Carpio C, et al. Safety and activity of ibrutinib in combination with nivolumab in patients with relapsed nonHodgkin lymphoma or chronic lymphocytic

2324

leukaemia: a phase 1/2a study. Lancet Haematol. 2019;6(2):e67-e78. 104. Davis RJ, Moore EC, Clavijo PE, et al. AntiPD-L1 efficacy can be enhanced by inhibition of myeloid-derived suppressor cells with a selective inhibitor of PI3Kδ/γ. Cancer Res. 2017;77(10):2607-2619. 105. Sam J, Colombetti S, Fauti T, et al. Combination of T-cell bispecific antibodies with PD-L1 checkpoint inhibition elicits superior anti-tumor activity. Front Oncol. 2020;10:1-15. 106. Wierz M, Pierson S, Guyonnet L, et al. Dual PD1/LAG3 immune checkpoint blockade limits tumor development in a murine model of chronic lymphocytic leukemia. Blood. 2018;131(14):1617-1621. 107. Kashyap MK, Amaya-Chanaga CI, Kumar D, et al. Targeting the CXCR4 pathway using a novel anti-CXCR4 IgG1 antibody

(PF-06747143) in chronic lymphocytic leukemia. J Hematol Oncol. 2017;10(1):1-16. 108. McWilliams EM, Lucas CR, Chen T, et al. Anti–BAFF-R antibody VAY-736 demonstrates promising preclinical activity in CLL and enhances effectiveness of ibrutinib. Blood Adv. 2019;3(3):447-460. 109. Chao MP, Alizadeh AA, Tang C, et al. AntiCD47 antibody synergizes with rituximab to promote phagocytosis and eradicate nonHodgkin lymphoma. Cell. 2010;142(5):699713. 110. Kjeldsen JW, Iversen TZ, EngellNoerregaard L, Mellemgaard A, Andersen MH, Svane IM. Durable clinical responses and long-term follow-up of stage III-IV non-small-cell lung cancer (NSCLC) patients treated with IDO peptide vaccine in a phase I study - a brief research report. Front Immunol. 2018;9:2145.

haematologica | 2021; 106(9)

ARTICLE

Acute Myeloid Leukemia

Monitoring of clonal evolution of acute myeloid leukemia identifies the leukemia subtype, clinical outcome and potential new drug targets for post-remission strategies or relapse Esther Onecha,1,2,3 Inmaculada Rapado,1,2,3,4 María Luz Morales,1,2,3 Gonzalo Carreño-Tarragona,1,2 Pilar Martinez-Sanchez,1,2,3 Xabier Gutierrez,1 José María Sánchez Pina,1 María Linares,1,2,3 Miguel Gallardo,1,2,3 Joaquín Martinez-López,1,2,3,4,5 and Rosa Ayala1,2,3,4,5

Ferrata Storti Foundation

Haematologica 2021 Volume 106(9):2325-2333

Hematology Department, Hospital Universitario 12 de Octubre; 2Instituto de Investigacion Hospital 12 de Octubre, Imas12; 3Hematological Malignancies Clinical Research Unit, CNIO; 4CIBERONC, Instituto Carlos III and 5Complutense University of Madrid, Madrid, Spain

ABSTRACT

n cases of treatment failure in acute myeloid leukemia (AML), the utility of mutational profiling in primary refractoriness and relapse is not established. We undertook a perspective study using next-generation sequencing (NGS) of clinical follow-up samples (n=91) from 23 patients with AML with therapeutic failure to cytarabine plus idarubicin or fludarabine. Cases of primary refractoriness to treatment were associated with a lower number of DNA variants at diagnosis than cases of relapse (median 1.67 and 3.21, respectively, P=0.029). The most frequently affected pathways in patients with primary refractoriness were signaling, transcription and tumor suppression, whereas methylation and splicing pathways were mainly implicated in relapsed patients. New therapeutic targets, either by an approved drug or within clinical trials, were not identified in any of the cases of refractoriness (zero of ten); however, eight potential new targets were found in five relapsed patients (five of 13, P=0.027): one IDH2, three SF3B1, two KRAS, one KIT and one JAK2. Sixty-five percent of all variants detected at diagnosis were not detected at complete response. Specifically, 100% of variants in EZH2, RUNX1, VHL, FLT3, ETV6, U2AF1, PHF6 and SF3B1 disappeared at complete response, indicating their potential use as markers to evaluate minimal residual disease for follow-up of AML. Molecular follow-up using a custom NGS myeloid panel of 32 genes in the post-treatment evaluation of AML can help in the stratification of prognostic risk, the selection of minimal residual disease markers to monitor the response to treatment and guide post-remission strategies targeting AML, and the selection of new drugs for leukemia relapse.

Introduction Approximately 20–30% of all patients with acute myeloid leukemia (AML) show primary refractoriness to induction therapy without achieving complete remission (CR) and approximately 50% will relapse.1 Both primary refractoriness and relapse are therapeutic failures associated with adverse prognosis, with cure rates no higher than 10%.2-5 AML is often an oligoclonal disease at its origin, because tumor clones with diverse genetic identity are present within the same patient in greater or lesser representation. In the last decade, much progress has been made in our understanding of tumor purity and the representation of clonal and sub-clonal mutations, particularly in the role played by sub-clones in the clonal architecture in AML.6 In this context, the clonal architecture can be driven not only by a single predominant clone, but also in some

haematologica | 2021; 106(9)

Correspondence: ROSA AYALA rosam.ayala@salud.madrid.org/rayala@ucm.es Received: April 4, 2020. Accepted: July 20, 2020. Pre-published: July 30, 2020. https://doi.org/10.3324/haematol.2020.254623

2325

E. Onecha et al.

cases by several leukemic clones participating in the leukemic process,7 and even by genetically distinct clones segregating or combining, providing more tumor diversity.8,9 Consequently, the predominating clone at diagnosis may differ from the clone predominating in states of relapse or refractoriness.5 Previous reports on the origin and evolution of genomic mutations in AML suggest that the majority are random events that arose in hematopoietic stem/progenitor cells before they acquired the driver mutation. Patients with clonal hematopoiesis frequently present with mutations in the genes TET2, RUNX1 and EZH2, whereas patients without clonal hematopoiesis are associated with mutations in NPM1 and FLT3.10 Furthermore, one-third of all patients with myelodysplastic syndromes (MDS) evolve to AML through a process of clonal evolution involving mutations in several genes including NPM1, RUNX1, TP53 and NRAS.11 The reappearance of leukemic disease after relapse can be through several distinct mechanisms: i) the founding clone acquires new mutations, expands and emerges as the predominant clone at relapse; ii) a non-founding clone or subclone resists chemotherapy, acquires new mutations, expands and becomes the predominant clone in relapse; iii) an ancestral, pre-diagnostic clone evolves and emerges as the major clone at relapse; and iv) the treatment triggers the appearance of a new clone, not previously present, and generates a second pathology (not a relapse per se).12,13 In a very recent study examining a cohort of adult patients with AML with NPM1 mutated at diagnosis, common cancer pathways such as MAPK and WNT were found to be enriched in relapsed samples with loss of the NPM1 mutation, whereas MYC and SCF-KIT signaling pathways were enriched in relapsed samples with persistent NPM1 mutation.14 Similarly, in an examination of the genetic mechanisms of primary chemotherapy resistance in pediatric AML, mutations in FRMD8, DHX32, PIK3R1, SHANK3, MKLN1, WT1 and TP53 were maintained or even enriched in refractory disease with respect to diagnosis, and mutations in FLT3, PTPN11 and NRAS genes were eradicated.15 In the present study, we investigated the clinical impact of the molecular evolution of AML in patients after standard induction treatment. We performed genetic mutational studies along the follow-up in refractory and relapsed AML using targeted next-generation sequencing (NGS) with a 32-gene panel. This approach involved a complete analysis of paired samples at the time of diagnosis versus refractoriness and versus relapse from 23 patients who failed induction chemotherapy and/or who relapsed after CR.

Methods

arm, n=2) as induction treatment according to PETHEMA (Programa Español de Tratamientos en Hematología) protocols. Other clinical characteristics are summarized in Table 1. The study was conducted according to the Spanish law 14/2007 on biomedical research and was approved by the research ethics board of each participating institution. All patients provided informed consent.

Mutational profile workflow DNA was extracted using Maxwell® 16 MDx (Promega Biotech Iberica SL, Madrid, Spain) and quantified on a Qubit® 2.0 Fluorometer (Invitrogen, Thermo Fisher Scientific Inc., Waltham, MA). The sequencing workflow was done with a custom NGS myeloid panel of 32 genes frequently mutated in myeloid diseases (Online Supplementary Table S1). In addition, the detection and quantification of mutated NPM1 sequences was performed by allele specific quantitative polymerase chain reaction (qPCR), as previously described,17 using RNA as biological sample and ABL1 as the expression reference gene for normalization.18 Internal tandem duplications in FLT3 were detected using GENSCAN.19 Fastq files were processed and genomic variants were detected using RUbioSeq3.8.20 which filtering and prioritization are detailed in Online Supplementary Figure S2.

Statistical analysis All analyses were performed using the R environment (v3.4.4) for statistical computing. Fisher’s exact test was used to determine differences between two categorical variables. The median follow-up time was 18.5 (range, 2.8–127) months. Primary refractory AML was defined as the failure to achieve CR after the first cycle of induction treatment. Partial response AML was defined as having 5–19% blast cells in bone marrow with

Table 1. Clinical description of patients.

Patients (n=23) Sex Age at diagnosis Blasts at diagnosis WBC at diagnosis AML type Karyotype at diagnosis Cytogenetics Risk Group (ELN-2010)

Patients The NGS-based mutational dynamics study was performed in a cohort of 23 AML patients with therapeutic failure, refractory to induction treatment (n=8), relapsed after reaching CR (n=13), and first refractory then relapsed (n=2), diagnosed between 2007 and 2015 in the Hospital 12 de Octubre, Madrid. Patients were selected from a previous sequencing study at diagnosis (n=190).16,17 The study evaluated 91 samples in total from the 23 cases at different time points: diagnosis (n=23), CR (n=31), partial remission (n=3), primary refractoriness (n=13), second-line refractoriness (n=4) and relapse (n=17). The median age at diagnosis was 59 ( range, 24–77) years and patients were treated with cytarabine and idarubicin (3+7 scheme, n=21) or with or in FLUGAZA clinical trial (azacytidine

2326

HSCT

Induction therapy

Male 13 (57%) Years, median (range) %, median (range) 109/L, median (range) De novo t-AML Normal Altered Low Intermediate High Autologous Allogenic Not done 3+7 scheme* Azacytidine**

Female 10 (43%) 59 (24–78) 70 (8–06) 8.7 (1.2–145) 20 (87%) 3 (13%) 7 (30%) 16 (70 %) 3 (13%) 12 (52%) 8 (35%) 2 (9 %) 9 (39%) 12 (52%) 21 (91%) 2 (9%)

The table represents clinical data of patients included in the NGS study. WBC: white blood cells; t-AML: secondary AML to other chemotherapy; HSCT: hematopoietic stem cell transplantation. *3+7 regimen of chemotherapy: one or two induction cycles of cytarabine and idarubicin during seven and three days, respectively; and two or three consolidation cycles at high doses of cytarabine, twice a day for three alternate days followed by allogenic- or autologous-HSCT. ** Azacytidine scheme (azacytidine days 1 to 7).

haematologica | 2021; 106(9)

Monitoring of clonal evolution of AML

>50% decrease after one cycle of induction treatment; for statistical evaluation a partial remission sample was considered as a refractory sample.21 Relapse AML was defined as the recurrence of disease after CR, provided that it was detected in ≥5% blasts in the bone marrow or peripheral blood. The classification of clonal or subclonal mutation was derived from the variant allele frequency (VAF), which provides information about how many cells in a sample carry a particular variant. The VAF is defined as the ratio of sequence reads carrying the mutation to the total number of reads at a specific nucleotide position. In this study, VAF≥10% discriminates a clonal mutation and VAF<10% a subclonal mutation. Accordingly, the predominant clone is the one with a higher VAF. Also, additional molecular abnormalities (AMA) in patients with refractory or relapsed disease were defined as new mutations not present at diagnosis.

Results Patient cohorts and clinical-biological characteristics We observed a dynamic mutational profile along the

course of AML evaluation, both for samples from patients refractory to induction treatment (patients 1 to 10; Figure 1A and B) and from patients who relapsed after reaching CR (patients 9 to 23; Figure 1A to C). A detailed description of biological and clinical events, and specific treatments of the patients is reported in the Online Supplementary Appendix.

Mutational landscape at diagnosis and treatment failure Overall, 71 non-recurrent somatic variants were detected with a median coverage of 1,044 (range, 20–6,123) and a median VAF of 36% (range, 1–95%) (Online Supplementary Table S2). Fifty-one variants (71.8%) were single nucleotide variants (SNV) and 20 (28.2%) were small insertions (n=15) or deletions (n=5). Fifty missense variants were identified, in addition to one stop-gain, one in-frame deletion, four frameshift deletions, 14 frameshift insertions and one nonframeshift insertion. At diagnosis (n=23), 57 variants were detected with a median of twp (range, 0–5) variants/sample. We did not find any mutations in three patients, either at the beginning of

Figure 1A. Legend on page 2329.

haematologica | 2021; 106(9)

2327

E. Onecha et al.

Figure 1B. Legend on following page.

the disease or at any other time during evaluation; furthermore three of 23 cases presented with an altered karyotype at diagnosis. In the analysis of the samples of treatment refractoriness, we detected a median of one (range, 0–4) variant/sample in 20 samples evaluated, with three samples having no mutations. At CR, we detected a median of one (range, 0–4) variant/sample in 31 samples evaluated, with 11 samples having no mutations. At relapse, we detected a median of three (range, 0–11) variants/sample in 17 samples evaluated, one of which did not have a mutation (Online 2328

Supplementary Figure S1). Differences in genomic features between primary refractoriness and relapse at diagnosis Cases of primary refractoriness were associated with high-risk cytogenetics (ELN-2010 criteria)22 (six of nine), whereas relapsed cases were related to intermediate-risk cytogenetics ELN-201022 (nine of 14, P=0.085). Also, the cases of primary refractoriness were associated with a lower number of variants at diagnosis (median 1.67) than leukemia relapses cases, with a median of 3.21 variants (P=0.029). haematologica | 2021; 106(9)

Monitoring of clonal evolution of AML

Figure 1. Mutational features of patients. (A) The graphs show the mutational profile after induction treatment in primary refractory cases (patients 1 to 8), and patients who present with both primary refractory disease and relapse after complete remission (CR) (patient 9). (B) The graphs show the mutational profile after induction treatment in patients who present with both primary refractory disease and relapse after CR (patient 10), and patients who relapse after induction treatment (patients 11 to 18). (C) The graphs show the mutational profile after induction treatment in relapse cases (patients 19 to 23). The balls represent the presence of mutations, with their size representing the percentage variant allele frequency (%VAF) and the color representing the moment of evaluation (dx=blue, CR=green, PR and Rf=purple, R=red). Sex, age and type of acute myeloid leukemia are indicated in the individual table, as well as karyotype and treatment administrated. The percentage of blasts is indicated above the mutation, and the time-frame of treatment cycles is indicated in the same way. y: years; HD-Cyt: high-dose cytosine; Ams: amsacrine; Mit: mitoxantrone; Eto: etoposide, GO: gemtuzumab; Clo: clofarabine; Flu: fludarabine; Ida: idarubicine; Pxf: plerixafor; Dau: daunorubicin; Mid. Midostaurin; allo-HSCT: allogeneic hematopoietic stem cell transplant; auto-HSCT: autologuos hematopoietic stem cell transplant; NA: not available. Dx=diagnosis, CR=complete remission, PR=partial remission and Rf=refractoriness and R=relapse. In the case of several samples at time of evaluation, the samples were labeled sequentially (s1=sample 1, s2=sample 2, s3 = sample 3 and successively). Also, the mutated gene and variant (protein coding) and the exact % of VAF is indicated.

At diagnosis, we observed that in the group of patients who had shown primary refractoriness, the most frequently mutated genes were those associated with tyrosine kinases (KIT, NRAS, CBL, RUNX1) and TP53. By contrast, genes related to epigenetic regulation (DNMT3A, IDH1/2, KMT2A) and SF3B1 were more frequently mutated in the group of patients who subsequently relapsed (Online Supplementary Table S2).

Clonal evolution is involved with dynamics of variant allele frequency Almost 68% (67.7%) of the variants in the follow-up samples evaluated after induction cycles were the same as those detected at diagnosis (Table 2): TP53 (n=3/4), NRAS (n=3/3), KIT (n=3/3), CBL (n=3/3), RUNX1 (n=2/2), TET2 (n=2/2), ASXL1 (n=1/1), CALR (n=1/1), EZH2 (n=1/3), FLT3-SNV (n=1/1) and ETV6 (n=1/2). We also detected the following newly acquired mutations at treatment refractoriness: MPL (n=2), CBL (n=1), TP53 (n=1), and VHL (n=1). Notably, several variants were detected at diagnosis and disappeared at refractoriness (32.3%): PHF6 (n=3/3), U2AF1 (n=2/2), JAK2 (n=1/1), EZH2 (n=2/3), ETV6 (n=1/2) and TP53 (n=1/4). When we analyzed paired relapsed and diagnosis samples (Table 2), the variants that were maintained (80.7%) in relapsed samples were located in the following genes: DNMT3A (n=6/7), SF3B1 (n=3/3), KMT2A (n=4/5), TP53 (n=2/2), IDH2 (n=4/4), FLT3-SNV (n=4/5), TET2 (n=3/4), haematologica | 2021; 106(9)

ASXL1 (n=3/3), JAK2 (n=2/2), RUNX1 (n=2/2), EZH2 (n=2/2), IDH1 (n=2/2), CBL (n=1/1), NRAS (n=1/5), ETV6 (n=1/1), PHF6 (n=1/1), SRSF2 (n=1/1) and ZRSR2 (n=1/1). In addition, we detected 18 variants that were newly acquired during the progression: SF3B1 (n=3), EPOR (n=3), KRAS (n=2), IDH2 (n=1), KMD6A (n=1), KMT2A (n=1), KIT (n=1), PRPF40B (n=1), SF3A1 (n=1), U2AF1 (n=1), JAK2 (n=1), VHL (n=1) and TP53 (n=1). By contrast, the variants detected at diagnosis but that disappeared (19.3%) in relapsed samples were located in genes: NRAS (n=4/5), DNMT3A (n=1/8), KMT2A (n=1/5), FLT3-SNV (n=1/5), TET2 (n=1/4), VHL (n=2/2) and PTEN (n=1/1). These latter clones could be sensitive to treatment. We observed a decreased mutational load of 8.1% in samples from treatment refractory patients versus diagnosis samples (Figure 2A). While, an increased mutational load of 3.74% in the relapsed versus diagnosis samples (Figure 2B).

Molecular findings in complete remission of leukemia Regarding CR evaluation, we detected 81 variants in the study of paired CR and diagnosis samples (Table 2). The variants detected at diagnosis and maintained in CR (34.6%) were located in TET2 (n=4/8), DNMT3A (n=6/10), ASXL1 (n=3/3), NRAS (n=3/9), KMT2A (n=1/2), SF3B1 (n=3/6), IDH2 (n=2/7), SRSF2 (n=3/3) and CBL (n=2/2). In addition, the variants that disappeared in CR (65.4 %) were located in IDH2 (n=5/7), NRAS (n=6/9), DNMT3A (n=4/10), TET2 2329

E. Onecha et al.

Table 2. Variants detected in refractoriness, complete remission and relapse versus diagnosis.

Dx + Rf + ASXL1 CALR CBL DNMT3A EPOR ETV6 EZH2 FLT3-SNV IDH1 IDH2 JAK2 KMD6A KIT KMT2A KRAS MPL NRAS PHF6 PRPF40B PTEN RUNX1 SF3A1 SF3B1 SRSF2 TET2 TP53 U2AF1 VHL ZRSR2 N

Refractoriness Dx + Rf -

1 1 3

Dx Rf +

Dx + R +

Relapse Dx + R -

Dx R +

3 1

1 6

Dx + CR +

Complete Remission Dx + Dx CR CR +

3 2 6

3 1 1 1

1 2

1 2 4 2 4 2

2 3 6 6

3 4

1 1

1 1 1 1 1 2

6 2

2 3 3

1 1 2

2 3

1 2

3 1 2 2

1 21

1 42

5 1 3 1

1 1 1

4 3 4

3 4 3 5

The table lists the number of samples with allelic variants detected in the different genes included in the study. The table specifies the samples studied in primary refractoriness, relapse and complete remission. The last row indicates the number of samples that are present at diagnosis (Dx +), at refractoriness (Rf +), at relapse (R +) or at complete remission (CR +).Variants that are present in the diagnosis (Dx +) but absent in refractoriness (Rf -) or relapse (R -) or in complete remission (CR-). Thus, there are variants absent in the diagnosis (Dx -) but are present in refractoriness (Rf +) or relapse (R +) or complete remission (CR +).

(n=4/8), EZH2 (n=6/6), RUNX1 (n=5/5), VHL (n=5/5), FLT3 (n=6/6), SF3B1 (n=4/6), ETV6 (n=3/3), U2AF1 (n=3/3), PHF6 (n=2/2), y KMT2A (n=1/2). Also, five variants arose de novo in DNMT3A (n=2) and MPL (n=3). These results provide potential markers that could be used to detect minimal residual disease (MRD) in our series, including EZH2, RUNX1, VHL, FLT3, ETV6, U2AF1, PHF6 and SF3B1, as these variants disappear in CR.

Branching clonal evolution is predominant in acute myeloid leukemia Analysis of the molecular dynamics of the clones according to the VAF identified three patients who showed a change in the predominant clone from diagnosis to primary refractoriness: clones characterized by mutations in VHL (patient 1), ETV6 (patient 2) and TP53 (patient 6) became the predominant clones at refractoriness (Figure 1A). Likewise, four relapsed patients showed changes in the predominant clone from diagnosis, characterized by mutations in EPOR 2330

(patient 9), TP53 (patient 11), VHL (patient 16) and PHF6 (patient 22) (Figure 1A to C). In addition, clonal evolution was observed in 12 patients. A linear clonal evolution model was identified in four patients in primary refractoriness and two in relapse. By contrast, a branching clonal evolution model was identified in two patients in primary refractoriness and in six patients in relapse. Subclonal mutations (VAF <10%) were detected at diagnosis in signaling pathway genes (JAK2, FLT3, NRAS) and in splicing genes (U2AF1, SF3B1). Of these, JAK2, NRAS and U2AF1 variants disappeared in the treatment failure samples, whereas FLT3 and SF3B1 variants persisted. Also, we detected variants in TP53 and PHF6 (tumor suppressor genes), but only the PHF6 variant became the predominant clone in a relapsed sample (VAF=83%). Two TP53-subclonal mutations were detected at diagnosis in the same patient, but only one of them was maintained at a similar frequency in the refractory sample.

haematologica | 2021; 106(9)

Monitoring of clonal evolution of AML

Figure 2. Variation of the allelic frequency of the variants. Box plots representing the increase or decrease of allelic frequencies detected in genes where variants have been detected at refractoriness. (A) and at relapse (B). The genes are grouped according to metabolic pathways, with different colors representing: transcriptional regulator genes (ASXL1, EZH2 and PHF6) in green, CALR in black, epigenetic regulator genes (DNMT3A, TET2, IDH1, IDH2, KDM6A and KMT2A) in yellow, splicing genes (SF1, SF3A1, SF3B1, SRSF2, U2AF1, ZRSR2 and PRPF40B) in brown, cytokine signaling and JAK/STAT pathway genes (EPOR, FLT3, JAK2, KIT, SH2B3, MPL and CBL) in blue, GTPase activity genes (HRAS, KRAS and NRAS) in pink, transcription factors genes (ETV6 and RUNX1) in grey and tumor suppressor genes (VHL, TP53 and PTEN) are represented in red.

Clonal evolution is not associated with a worse outcome Conventional molecular alterations detected at diagnosis in ten patients were lost both at refractoriness (n=3) and leukemia relapse (n=7). AMA were identified in eight patients, who all achieved CR (eight of 17 cases with any CR vs. zero of six cases with no CR; P=0.037). Median overall survival was 77.4 (range, 21.5–133.3) months for the group with AMA features versus 11.8 (range, 1.2–22.4) months for the group without AMA features (P=0.083, Online Supplementary Figure 3SA). Clonal evolution detected with AMA identified patients with a trend for a better prognosis for disease-free survival (median disease-free survival was 22.1 vs. 10.8 months; P=0.065, Online Supplementary Figure 3SB); however, neither loss of molecular abnormalities nor combined additional and lost molecular abnormalities had an impact on prognosis. Accrual of AMA was mainly related to signaling pathway genes (five of eight cases cases with gain of mutations in the genes of the signaling pathway compared with only three of 15 cases without gain of mutations; P=0.042). Loss of AMA in relapsed samples was also mainly found for signaling pathway genes (seven of ten cases with loss of AMAs present in signaling pathways genes compared with one of 13 cases without loss of AMA; P=0.002). Other variables with a trend for an association with cases who achieved CR were normal karyotype (six of 17 cases with any CR vs. zero of six cases with no CR; P=0.091) and cytogenetic risk (only six of ten high-risk, but 11 of 13 intermediate- or low-risk achieved CR; P=0.183). haematologica | 2021; 106(9)

New therapeutic targets, either by an approved drug or within clinical trials, were not identified in cases of refractoriness; however, eight potential new targets were found in five relapsed cases (zer of ten refractoriness cases vs. five of 13 relapse cases, P=0.027): one IDH2, three SF3B1, two KRAS, one KIT and one JAK2.

Discussion Patients with AML who show primary resistance to induction treatment or leukemic relapse have a dismal prognosis. Our study identifies differences in the mutational landscape between primary refractory and relapsed AML. In this line, we report the usefulness of monitoring different leukemic clones, and particularly detecting the appearance of new clones, using an NGS-targeted panel in post-treatment AML, allowing us to: i) stratify patients into prognostic risk groups; ii) select MRD marker/s to monitor response to treatment; and ii) define targeted post-remission strategies including the selection of new drugs for leukemia relapse. The genetic follow-up of leukemic clones was performed using NGS technology with a high coverage of the variants (>1,000×), allowing the detection of sub-clones with a high sensitivity (<3%) in a cohort of 23 patients with AML, and with almost 100 samples evaluated. The NGS technology also allowed us to estimate the mutational load based on the VAF level, and to, therefore, infer the clonal architecture of the tumor and the model of clonal evolution.6 We confirmed the high clonal heterogeneity associated with this disease and the mutational profile associated with treatment refrac2331

E. Onecha et al.

toriness and relapse. The genes most frequently mutated at diagnosis in patients showing subsequent primary refractoriness were CBL, KIT, NRAS, RUNX1 and TP53, and those more frequently mutated in relapsed patients were DNMT3A, IDH1/2, KMT2A and SF3B1. Thus signaling, transcription and tumor suppression pathways were the more affected biological categories at diagnosis in the treatment refractory group, whereas methylation and splicing were the pathways most affected at diagnosis in the relapsed group. This perhaps indicates that methylation and splicing are rescue pathways used by leukemia cells to develop resistance to treatment. Our findings may lead to the development of a new focused therapeutic approach for patients belonging to the high-risk cytogenetics group at relapse but who have nevertheless achieved CR, because the oncologist could anticipate maintenance treatments based on drugs targeting methylation and splicing pathways, for example, hypomethylants or splicing inhibitors. Rescue pathways were not identified for cases of primary refractoriness, as clones remained stable and alterations were found in signaling and tumor suppressor genes, which are the most clinically relevant. We identified new potential therapeutic targets at the leukemia relapse stage affecting IDH2, SF3B1, KRAS, KIT and JAK2 genes, all of which are targets for approved drugs or available within clinical trials. However, some identified variants in SF3B1 and FLT3 are categorized as variant of uncertain (or unknown) significance and for clinical decision making only pathogenic or probably pathogenic variants can be used. Supporting our findings, a recent study described de novo mutations in transcription factors, signaling, cohesin and splicing pathways at the time of leukemia relapse in the t(8;21) AML patient subgroup;23 however, mutations detected at diagnosis in epigenetic regulators and genes involved in cell cycle control were stable or disappeared.23 We also detected an equal percentage of cases (32.3%) where the dominant clone changed within the refractory group versus the leukemia relapsed group with evolution of the following genes: ETV6, VHL, EPOR, JAK2, TP53, and PHF6. Consequently, therapeutic approaches must be targeted specifically to these clones if they are detected at diagnosis. The impact of subclonal mutations detected at diagnosis and their usefulness as MRD markers is not yet defined. In our AML series, 55% of subclones detected at diagnosis were lost in failure samples and the other 45% remained as subclones. Only the PHF6 mutation became the predominant clone in a failure sample, supporting the concept that treatment can result in subclonal eradication, but whether a resistance-mediating mutation determines the presence of a corresponding subclone from diagnosis could found a future relapse. We observed an increased mutational load as a strong molecular feature of relapse, as previously established in other hematological malignancies.24 In addition, the authors of the aforementioned study suggest that the knowledge of the tumor burden is important in the identification of subclones, with the aim of targeted and specific therapies to eradicate them and to follow their evolution. In an analysis of over 4,000 patients with newly diagnosed AML, several biological-clinical variables (age, performance status, white blood cell count, secondary disease, cytogenetic risk and NPM1/FLT3-ITD mutational status) 2332

were each strongly and independently associated with resistance (P<0.001); however, their ability to predict resistance was only fair.25 Our study includes other molecular markers (AMA) that improve the prediction of failure to response to leukemia relapse treatment (P=0.066), although it does not predict it for refractoriness (P=not signifcant). This is likely justified by the fact that the refractory group was enriched with high-risk cytogenetics features at diagnosis, whereas the relapsed group was associated with a greater number of pathogenetic or likely pathogenetic variants at diagnosis. In contrast to other reported findings,26 we did not detect differences in the age of the patients between the group with persistence of these mutations and the group without these mutations. We found persisting DNMT3A, TET2 and ASXL1 variants in CR, which have been previously reported,26 and are related to clonal hematopoiesis of indeterminate potential (with oncogenic potential).27 Other mutations, such as those involving KMT2A, CBL and NRAS, could be associated with AML transformation from a prior clonal disease, as described by Bejar.28 These differ from reported mutations that were involved in leukemic hematopoiesis in NPM1, PHF6, SRSF2, RUNX1 and TP53,27,29-36 although de novo cases predominated in our series. To the best of our knowledge, we provide the first clinical evidence that clonal evolution defined as AMA is a feature associated with cases that achieve CR and that have a better prognosis for disease-free survival. A previous study in AML provided a contrary prediction, which in an analogous manner defined the clonal evolution as additional cytogenetic abnormalities, and observed a worse prognosis.37 Our prediction agrees with reported findings in NPM1mutated AML,14 whereby AML patients with clonal evolution (NPM1-) at relapse have a significantly longer remission duration than patients without clonal evolution (NPM1+). The better prognosis associated with patients with new clones (AMA) might be due to the fact that the treatment has been effective in the basal clone. Previous studies have shown an association between persisting clonal cytogenetic markers in first remission and an increased risk of relapse.38,39 Somatic mutations that activate signaling pathways (FLT3, KRAS, or NRAS) were usually cleared on day 30, suggesting that subclones containing these mutations may be more sensitive to induction chemotherapy.2 Although new advances in induction treatment for AML have improved the rates of CR and overall survival, most patients ultimately relapse without effective post-remission therapy.40 The utility of clonal dynamics studies can be tested with new treatments such as FLT3, IDH1 and IDH2 inhibitors. The recommended study time to perform the monitoring of clonal evolution would be at diagnosis, at the end of the induction in CR, and at the refractory or relapsed stages, with the main utility in those patients with AML who achieve CR or blast clearance. Although some conclusions obtained need to be validated in another wide series, the results of the relevance of clonal kinetics and its implications are robust. Our results suggest that the monitoring of clonal evolution by genomic approaches can help to select post-remission strategies to target AML, and may improve prediction of clonal evolution and response of treatment. Disclosures No conflicts of interest to disclose. haematologica | 2021; 106(9)

Monitoring of clonal evolution of AML

Contributions EO collected samples, performed experiments, analyzed and interpreted data, and wrote the manuscript; IR analyzed and interpreted data; LM, GC-T, XG, MG, JS interpreted data; PM collected samples and clinical data; RA and JML designed and supervised research and experiments, analyzed and interpreted data, and wrote the manuscript; all authors prepared the report and approved the final version. Acknowledgments The study was conducted according to the Declaration of

References 1. Kuehn EW, Walz G, Benzing T. Von hippellindau: a tumor suppressor links microtubules to ciliogenesis and cancer development. Cancer Res. 2007;67(10):4537-4540. 2. Klco JM, Miller CA, Griffith M, et al. Association between mutation clearance after induction therapy and outcomes in acute myeloid leukemia. JAMA. 2015; 314(8):811-822. 3. Breems DA, Van Putten WL, Huijgens PC, et al. Prognostic index for adult patients with acute myeloid leukemia in first relapse. J Clin Oncol. 2005;23(9):1969-1978. 4. Pemmaraju N, Kantarjian H, Garcia-Manero G, et al. Improving outcomes for patients with acute myeloid leukemia in first relapse: a single center experience. Am J Hematol. 2015;90(1):27-30. 5. Bose P, Vachhani P, Cortes JE. Treatment of relapsed/refractory acute myeloid leukemia. Curr Treat Options Oncol. 2017; 18(3):17. 6. Vosberg S, Greif PA. Clonal evolution of acute myeloid leukemia from diagnosis to relapse. Genes Chromosomes Cancer. 2019;58(12):839-849. 7. De S, Ganesan S. Looking beyond drivers and passengers in cancer genome sequencing data. Ann Oncol. 2017;28(5):938-945. 8. Barber LJ, Davies MN, Gerlinger M. Dissecting cancer evolution at the macroheterogeneity and micro-heterogeneity scale. Curr Opin Genet Dev. 2015;30:1-6. 9. Gerlinger M, Rowan AJ, Horswell S, et al. Intratumor heterogeneity and branched evolution revealed by multiregion sequencing. N Engl J Med. 2012;366(10):883-892. 10. Welch JS, Ley TJ, Link DC, et al. The origin and evolution of mutations in acute myeloid leukemia. Cell. 2012;150(2):264-278. 11. Genovese G, Kahler AK, Handsaker RE, et al. Clonal hematopoiesis and blood-cancer risk inferred from blood DNA sequence. N Engl J Med. 2014;371(26):2477-2487. 12. Grimwade D, Ivey A, Huntly BJ. Molecular landscape of acute myeloid leukemia in younger adults and its clinical relevance. Blood. 2016;127(1):29-41. 13. Ramos NR, Mo CC, Karp JE, Hourigan CS. Current approaches in the treatment of relapsed and refractory acute myeloid leukemia. J Clin Med. 2015;4(4):665-695. 14. Cocciardi S, Dolnik A, Kapp-Schwoerer S, et al. Clonal evolution patterns in acute myeloid leukemia with NPM1 mutation. Nat Commun. 2019;10(1):2031. 15. McNeer NA, Philip J, Geiger H, et al. Genetic mechanisms of primary chemotherapy resistance in pediatric acute myeloid leukemia. Leukemia. 2019; 33(8): 1934-1943.

haematologica | 2021; 106(9)

Helsinki; the protocol was reviewed and approved by the Institutional Review Board/Independent Ethics Committe of the participating centers. Funding This study was supported by the Subdirección General de Investigación Sanitaria (Instituto de Salud Carlos III, Spain) grants PI13/02387 and PI16/01530, and the CRIS against Cancer foundation, grant 2014/0120. ML holds a postdoctoral fellowship of the Spanish Ministry of Economy and Competitiveness (FPDI-201316409).

16. Onecha E, Linares M, Rapado I, et al. A novel deep targeted sequencing method for minimal residual disease monitoring in acute myeloid leukemia. Haematologica. 2019;104(2):288-296. 17. Cedena MT, Rapado I, Santos-Lozano A, et al. Mutations in the DNA methylation pathway and number of driver mutations predict response to azacitidine in myelodysplastic syndromes. Oncotarget. 2017;8(63):106948106961. 18. Gorello P, Cazzaniga G, Alberti F, et al. Quantitative assessment of minimal residual disease in acute myeloid leukemia carrying nucleophosmin (NPM1) gene mutations. Leukemia. 2006;20(6):1103-1108. 19. Burge C, Karlin S. Prediction of complete gene structures in human genomic DNA. J Mol Biol. 1997;268(1):78-94. 20. Rubio-Camarillo M, Lopez-Fernandez H, Gomez-Lopez G, et al. RUbioSeq+: a multiplatform application that executes parallelized pipelines to analyse next-generation sequencing data. Comput Methods Programs Biomed. 2017;138:73-81. 21. Cheson BD, Bennett JM, Kopecky KJ, et al. Revised recommendations of the International Working Group for diagnosis, standardization of response criteria, treatment outcomes, and reporting standards for therapeutic trials in acute myeloid leukemia. J Clin Oncol. 2003;21(24):4642-4649. 22. Dohner 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. 23. Christen F, Hoyer K, Yoshida K, et al. Genomic landscape and clonal evolution of acute myeloid leukemia with t(8;21): an international study on 331 patients. Blood. 2019;133(10):1140-1151. 24. Jones JR, Weinhold N, Ashby C, et al. Clonal evolution in myeloma: the impact of maintenance lenalidomide and depth of response on the genetics and sub-clonal structure of relapsed disease in uniformly treated newly diagnosed patients. Haematologica. 2019; 104(7):1440-1450. 25. Walter RB, Othus M, Burnett AK, et al. Resistance prediction in AML: analysis of 4601 patients from MRC/NCRI, HOVON/SAKK, SWOG and MD Anderson Cancer Center. Leukemia. 2015; 29(2):312320. 26. Rothenberg-Thurley M, Amler S, Goerlich D, et al. Persistence of pre-leukemic clones during first remission and risk of relapse in acute myeloid leukemia. Leukemia. 2018;32(7):1598-1608.

27. Shlush LI. Age-related clonal hematopoiesis. Blood. 2018;131(5):496-504. 28. Bejar R. What biologic factors predict for transformation to AML? Best Pract Res Clin Haematol. 2018;31(4):341-345. 29. Metzeler KH, Herold T, Rothenberg-Thurley M, et al. Spectrum and prognostic relevance of driver gene mutations in acute myeloid leukemia. Blood. 2016;128(5):686-698. 30. Corces-Zimmerman MR, Hong WJ, Weissman IL, Medeiros BC, Majeti R. Preleukemic mutations in human acute myeloid leukemia affect epigenetic regulators and persist in remission. Proc Natl Acad Sci U S A. 2014;111(7):2548-2553. 31. Sottoriva A, Kang H, Ma Z, et al. A Big Bang model of human colorectal tumor growth. Nat Genet. 2015;47(3):209-216. 32. Jan M, Majeti R. Clonal evolution of acute leukemia genomes. Oncogene. 2013; 32(2):135-140. 33. Kronke J, Bullinger L, Teleanu V, et al. Clonal evolution in relapsed NPM1-mutated acute myeloid leukemia. Blood. 2013; 122(1):100108. 34. Jaiswal S, Fontanillas P, Flannick J, et al. Agerelated clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014;371(26):2488-2498. 35. Klco JM, Spencer DH, Miller CA, et al. Functional heterogeneity of genetically defined subclones in acute myeloid leukemia. Cancer Cell. 2014;25(3):379-392. 36. Quek L, Ferguson P, Metzner M, et al. Mutational analysis of disease relapse in patients allografted for acute myeloid leukemia. Blood Adv. 2016;1(3):193-204. 37. Shimizu H, Yokohama A, Ishizaki T, et al. Clonal evolution detected with conventional cytogenetic analysis is a potent prognostic factor in adult patients with relapsed AML. Hematol Oncol. 2018;36(1):252-257. 38. Marcucci G, Mrozek K, Ruppert AS, et al. Abnormal cytogenetics at date of morphologic complete remission predicts short overall and disease-free survival, and higher relapse rate in adult acute myeloid leukemia: results from cancer and leukemia group B study 8461. J Clin Oncol. 2004;22(12):24102418. 39. Chen Y, Cortes J, Estrov Z, et al. Persistence of cytogenetic abnormalities at complete remission after induction in patients with acute myeloid leukemia: prognostic significance and the potential role of allogeneic stem-cell transplantation. J Clin Oncol. 2011;29(18):2507-2513. 40. Derman BA, Larson RA. Post-remission therapy in acute myeloid leukemia: Are we ready for an individualized approach? Best Pract Res Clin Haematol. 2019;32(4): 101102.

2333

ARTICLE Ferrata Storti Foundation

Haematologica 2021 Volume 106(9):2334-2344

Chronic Lymphocytic Leukemia

Three-dimensional co-culture model of chronic lymphocytic leukemia bone marrow microenvironment predicts patient-specific response to mobilizing agents Federica Barbaglio,1,2* Daniela Belloni,2* Lydia Scarfò,2,3,4 Francesca Vittoria Sbrana,1 Maurilio Ponzoni,3,5 Lucia Bongiovanni,5 Luca Pavesi,1 Desiree Zambroni,6 Kostas Stamatopoulos,7,8 Valeria R. Caiolfa,6,9 Elisabetta Ferrero,2 Paolo Ghia2,3,4 and Cristina Scielzo1 Unit of Malignant B Cells Biology and 3D Modelling, Division of Experimental Oncology, IRCCS Ospedale San Raffaele, Milan, Italy; 2Unit of B Cell Neoplasia, Division of Experimental Oncology, IRCCS Ospedale San Raffaele, Milan, Italy; 3 Università Vita-Salute San Raffaele, Milan, Italy; 4Strategic Research Program on CLL, Division of Experimental Oncology, IRCCS Ospedale San Raffaele, Milan, Italy; 5 Pathology Unit, IRCCS Ospedale San Raffaele, Milan, Italy; 6Center for Experimental Imaging, IRCCS, Ospedale San Raffaele, Milan, Italy; 7Hematology Department and HCT Unit, G. Papanicolaou Hospital, Thessaloniki, Greece; 8Institute of Applied Biosciences, Centre for Research and Technology Hellas, Thessaloniki, Greece and 9Centro Nacional de investigaciones Cardiovasculares, (CNIC), Madrid, Spain 1

*FB and DB contributed equally as co-first authors.

ABSTRACT

Correspondence: CRISTINA SCIELZO scielzo.cristina@hsr.it Received: January 28, 2020. Accepted: July 17, 2020. Pre-published: July 30, 2020. https://doi.org/10.3324/haematol.2020.248112

2334

hronic lymphocytic leukemia (CLL) cells disseminate into supportive tissue microenvironments. To investigate the mechanisms involved in leukemic cell tissue retention we developed a threedimensional bone marrow (BM) microenvironment that recreates the interactions between CLL and BM stromal cells inside a scaffold within a bioreactor. Our system allows the parallel analysis of CLL cells retained inside the scaffold and those released in the presence/absence of pharmacological agents, mimicking tissue and circulating cell compartments, respectively. CLL cells can be retained within the scaffold only in the presence of microenvironmental elements, which through direct contact downregulate the expression of HS1 cytoskeletal protein in CLL cells. Consistent with this, the expression of HS1 was lower in CLL cells obtained from patients’ BM than in CLL cells circulating in the peripheral blood. Moreover, we demonstrate that CLL cells with inactive HS1, impaired cytoskeletal activity and a more aggressive phenotype are more likely to be retained within the scaffold despite the presence of ibrutinib, whose mobilizing effect is mainly exerted on those with active HS1, ensuing dynamic cytoskeletal activity. This differential effect would not otherwise be assessable in a traditional two-dimensional system and may underlie a distinctive resistance of single CLL clones. Notably, CLL cells mobilized in the peripheral blood of patients during ibrutinib therapy exhibited activated HS1, underscoring that our model reliably mirrors the in vivo situation. The three-dimensional model described herein is suitable for reproducing and identifying critical CLL-BM interactions, opening the way to pathophysiological studies and the evaluation of novel targeted therapies in an individualized manner.

Introduction Chronic lymphocytic leukemia (CLL) is characterized by a progressive expansion of clonal CD5+ B lymphocytes that accumulate and traffic between the peripheral blood (PB), bone marrow (BM) and secondary lymphoid organs.1,2 In those sites, CLL cells are extremely dependent on and reactive to the microenvironment (i.e., stromal, endothelial cells and immune cells) and proliferate in so-called “proliferation centers”,

haematologica | 2021; 106(9)

3D co-culture model of CLL cells within BM microenvironment

mainly found in the lymph nodes, and/or sheltered in vaguely characterized niches in the BM.3-5 CLL cells accumulating within tissues tend to spill over into the circulating blood where they acquire a more resting phenotype, indicating that the most clinically relevant events occur in tissues. This feature also underlines the importance of the host tissues in CLL which conceivably contribute to disease progression and ultimately to treatment resistance.3,5,6 Cytoskeleton regulation is clearly implicated in the dynamic behavior of CLL cells, contributing to the homing and trafficking in and out of tissues, also during treatment. In particular, we previously reported that the activated status of the cytoskeletal protein hematopoietic lineage cell-specific protein 1 (HS1) defines a distinct signaling pathway and cytoskeletal activity in CLL, while also having prognostic implications, with the active and inactive forms of HS1 correlating with a favorable or adverse prognosis, respectively.79 In parallel, we demonstrated that downregulation of HS1 expression interferes with secondary lymphoid organ (lymph nodes and spleen) infiltration by CLL cells and leads to increased BM homing associated with impaired cytoskeletal activity.9,10 More recently, HS1 has been found to associate with ROR1 in enhancing CLL cell migration,11 further underlining its potential clinical significance. New targeted therapies, namely kinase inhibitors, have multiple modes of action, including the mobilization of leukemic cells from tissues into the bloodstream, where CLL cells lose the protective effect exerted by the microenvironment, eventually becoming more susceptible to cell apoptosis.12-14 Effectively, the use of the BTK inhibitor ibrutinib for CLL treatment has been a game-changer in the management of patients with this disease,14 although it is not curative and patients may relapse after several years of response.15 Inhibition of VLA-4-dependent adhesion of CLL cells to stroma and stromal components has been proposed as an explanation for the lymphocytosis induced by ibrutinib treatment,16 while other studies suggest a role of ibrutinib in modulating migration of CLL cells to chemokine gradients, in particular through CXCR4.17 However, a major limitation of investigating tissue retention and egress (or mobilization) in CLL originates from the lack of suitable in vitro models for recreating the close interactions between leukemic cells and the microenvironment. Calissano et al. first showed a relationship between in vivo CLL cell kinetics and the expression of CD38, a protein involved in CLL cell retention and trafficking.4 More recently, Pasikowska et al. reported differences between lymph node-derived CLL cells versus PB-derived cells by taking advantage of an in vitro system that models trans-endothelial migration,18 while Chen et al. demonstrated the dynamic expression of CXCR4 following BTK inhibition in vivo in a CLL mouse model.17 Despite these advances, none of the existing models is suitable for deeply characterizing what is happening to human CLL cells in the tissues. In order to partially overcome this limitation, we have exploited, and adapted to CLL, a three-dimensional (3D) co-culture model, already thoroughly validated for multiple myeloma, which is able to reproduce malignant cellmicroenvironment interactions.19 This 3D model is based on the integrated use of cell-repopulated scaffolds and a rotating bioreactor. This combination enables reciprocal interactions to be established between tumoral and nontumoral compartments inside the scaffolds and to promote CLL cell survival. Moreover, CLL cells can be recovered from both inside and outside the scaffolds, counted and haematologica | 2021; 106(9)

characterized for expression of lineage markers and of molecules putatively involved in their mobilization, providing the possibility to elucidate this mechanism, also in response to mobilizing agents, particularly ibrutinib. As a proof-of-principle, we here provide evidence of HS1 modulation in the presence of the drug, ultimately regulating CLL cell tissue homing and egress. Moreover, we report that this innovative 3D model is able to reliably reproduce the events occurring in vivo during homing and migration, thus potentially contributing to better understanding the pathogenic mechanisms leading to the dissemination and homing of CLL cells, particularly in response to treatment.

Methods Study subjects and ethics statement Patients with CLL were diagnosed according to the updated National Cancer Institute Working Group guidelines.20 PB samples were obtained after informed consent from patients who were (i) either untreated or off treatment for at least 6 months; or (ii) under ibrutinib treatment. The study was approved by the “San Raffaele” Hospital ethics committee under the protocol VIVI-CLL entitled: “In vivo and in vitro characterization on CLL”; and the CERTH ethics committee in response to the application entitled “Molecular and functional studies of B cell malignancies”. The clinical and biological characteristics of the patients with CLL who provided samples for the experiments are reported in Online Supplementary Table S1.

Scaffold preparation Scaffolds were populated as described by Belloni et al.19 and adapted to CLL cells. Briefly: scaffold discs were cut from SpongostanTM sheets (Ethicon, Inc. USA) using a sterile 4 mm2 biopsy punch and then pre-seeded with BM-derived stromal cells HS5 (200,000/scaffold) in 96-well suspension culture plates (Greiner bio-one, Germany). Scaffolds were then transferred to 10 mL High Aspect Ratio Vessels (HARV) in 1 mL TCM (DMEM culture medium supplemented with 10% v/v fetal bovine serum) and cultured overnight in the RCCSTM bioreactor at the lower speed (rpm). Twenty-four hours later, CLL cells were added to the vessels, using the optimal ratio of CLL cells to stromal cells established in preliminary experiments (MEC1 cells=2x106, primary CLL cells =3x10^6). After 5 h, vessels were filled with growth medium (RPMI1640 culture medium supplemented with 10% v/v or 20% fetal bovine serum for MEC1 or primary CLL cells, respectively). At the end of the culture period, cells outside and inside the scaffold were recovered from the scaffolds by means of liberase (Roche) (25 μg/mL) treatment for further analysis (see Online Supplementary Methods). The cells outside and inside the scaffold were counted using the trypan blue exclusion test for viability, which showed that more than 90% of the cells were viable. Alternatively, scaffolds were formalin-fixed for IF or lysed with 100 μL RIPA buffer for western blotting analysis (see Online Supplementary Methods).

Bioreactor RCCSTM The 3-D dynamic culture was performed using the RCCSTM bioreactor RCCS-4DQ equipped with four rotating 10 mL-HARV culture vessels, which work as culture chambers (Synthecon Inc., USA).19 Vessels are provided with a gas exchange membrane made of silicon rubber, which allows optimal diffusion of O . The bioreactor was kept inside an incubator, in a humidified atmosphere, at 37°C with 95% air and 5% CO . During the experimental proce2

2335

F. Barbaglio et al.

Figure 1. Optimization of the 3D model. (A) Schematic representation of the experimental setup: HS5 cells were seeded into 3D Spongostan scaffolds and cultured under microgravity in a RCCSTM bioreactor for 24 h. MEC1 cells or primary CLL cells were added to the same scaffold and co-cultured for 72 h under microgravity. At the end of the incubation period, cells and scaffolds were collected and analyzed. (B) Representative confocal section taken from an X,Y,Z-stack of a representative scaffold, after 72 h of MEC1-GFP culture in the bioreactor. Nuclei were stained with DAPI and the scaffold was imaged by transmitted light (TL). The square-marked region of interest is also shown at higher magnification for each acquired channel. (C) The graph shows the total number of cells recovered from the medium outside the scaffold after 72 h of dynamic culture in the bioreactor. MEC1 cells were distinguished from HS5 cells by their smaller size. Each experiment was run in triplicate. MEC1 cells were significantly retained into the scaffold only in the presence of HS5 cells (*P=0.01). (D) The graph shows the total number of cells recovered from the medium outside the scaffold after 72 h of dynamic culture in the bioreactor of MEC1 control cells and MEC1-HS1KD cells: the latter were retained in the scaffold more than MEC1 control cells (**P=0.0013). (E) Evaluation of HS1 expression by quantitative real-time polymerase chain reaction analysis in MEC1 control cells inside and outside the scaffolds (experiment run in triplicate) showing that HS1 is down-regulated inside the scaffolds (*P=0.028).

2336

haematologica | 2021; 106(9)

3D co-culture model of CLL cells within BM microenvironment

dures, the operational conditions of the RCCS™ were set and constantly monitored in order to keep the samples in a “free fall” condition, which minimizes sedimentation of the scaffold while maximizing mass transfer and cell viability for the extended culture period.

Ibrutinib treatment in the bioreactor After 72 h of 3D dynamic culture in the bioreactor, supernatants were withdrawn from the vessels and centrifuged at 1,500 rpm for 5 min. Recovered cells were counted. Clarified supernatants were put into the vessels again, with or without 10 μM ibrutinib. We compared two different concentrations of ibrutinib (1 and 10 μM) to exclude a possible role of cell apotpotosis in the mobilization from the scaffold, due to the possible increased toxicity of ibrutinib at the higher concentration, and did not observe any significant differences (Online Supplementary Figure S2E). Cultures were stopped after 5 h of treatment and cells in the supernatants and in the scaffolds were recovered and submitted to the above mentioned analysis (see Online Supplementary Methods).

Microenvironmental elements are required to establish a 3D culture bone marrow model for chronic lymphocytic leukemia We customized a new 3D co-culture model, previously validated by our group for myeloma cell survival19 to recreate CLL and BM-stromal cell interactions inside a scaffold kept in culture in a rotating bioreactor (Figure 1A). We selected scaffolds made of Spongostan, which has an ultrastructure similar to the trabecular structure of BM, also because of their superior performance in supporting CLL cell retention compared to either gelatin or collagen-coated beads (data not shown). To set the optimal experimental conditions, we first defined the best ratio of cellular components and the most appropriate co-culture medium for supporting cell viability (see Methods). The scaffolds were sequentially populated with the human BM-derived stromal cell line HS5 and the CLL cell line MEC1. Scaffolds

Figure 2. HS1 expression is regulated by the stromal bone marrow microenvironment. (A) The graph shows the total number of cells recovered in the medium outside the scaffold after 72 h of dynamic culture in the bioreactor. For the experiments with primary chronic lymphocytic leukemia (CLL) cells, control samples were run using CLL cells alone and HS5 cells alone under the same culture conditions. CLL cells were retained inside the scaffolds only in the presence of HS5 (**P=0.0047). (B, C) The line plots show CXCR4 and HS1 expression, respectively, as determined by quantitative real-time polymerase chain reaction (RT-qPCR) in primary CLL cells collected from inside the scaffolds or from the outside medium. CLL cells retained inside the scaffold expressed significantly lower levels of CXCR4 (n=8; *P=0.015) and HS1 (**P=0.005). (D) Line plot illustrating the down-regulation of HS1 expression in CLL cells isolated from peripheral blood when they were in direct co-culture with HS5 cells, as determined by RT-qPCR. (****P<0.0001). (E) Line plot illustrating the HS1 expression in CLL cells isolated from peripheral blood when they were cultured in a 1 μm pore trans-well without direct contact with HS5 cells.

haematologica | 2021; 106(9)

Results

2337

F. Barbaglio et al.

retrieved from the vessels after 3 days of co-culture and submitted to confocal analysis showed that GFP-tagged MEC1 cells populated the entire scaffold efficiently and homogeneously (Figure 1B). The model allows parallel analysis of CLL cells inside and outside the scaffold, revealing that HS5 cells were needed in order to retain MEC1 cells

efficiently within the scaffold (Figure 1C). Next, we used MEC1 cells genetically modified to downregulate HS1 expression (MEC1-HS1KD), already known to display increased BM homing capacity in vivo in a CLL xenograft model,9 and tested whether this could be reproduced in our 3D ex-vivo model. We co-cultured in 3D either

C Figure 3. Analysis of HS1 expression in chronic lymphocytic leukemia cells isolated from peripheral blood and bone marrow. (A) HS1 mRNA expression levels, determined by quantitative real-time polymerase chain reaction in primary chronic lymphocytic leukemia (CLL) cells isolated from the peripheral blood (PB) and the bone marrow (BM) of patients (n=9). HS1 was significantly downregulated in the BM, compared to its levels in the PB of the same patient (*P=0.0498). (B) Image Stream analysis of intra-clonal expression of HS1 in PB versus BM from CLL patients (n=4). By gating on the CLL pool (CD5+CD19+), we found mainly within BM a population that was negative for HS1 expression (red rectangle and arrow indicate a representative image of a single CLL cell negative for HS1). We also found a population positive for HS1 (black rectangle and arrow indicate a representative image of a single CLL cell positive for HS1). The percentage of the PB and BM populations negative for HS1 are also shown (right panel) (*P=0.0187). Immunohistochemical (IHC) analysis of HS1 expression in BM sections from CLL patients (n=4)

2338

haematologica | 2021; 106(9)

3D co-culture model of CLL cells within BM microenvironment

Figure 4. Ibrutinib treatment in 3D co-culture. (A) Schematic representation of the experimental setup: HS5 cells were seeded into 3D Spongostan scaffolds and cultured under microgravity in a RCCSTM bioreactor for 24 h. MEC1 cells or primary chronic lymphocytic leukemia (CLL) cells were added to the same scaffold and cocultured for 72 h under microgravity. At the end of the incubation period, the supernatant was collected and depleted of cells. The same supernatant was added again to the culture with or without ibrutinib for 5 h. At the end of the incubation period the scaffolds were collected and analyzed. (B) The histogram plot shows the total number of cells (MEC1-GFP+ HS5) that migrated outside the scaffold after 72 h of dynamic culture in the bioreactor in the presence of 10 μM ibrutinib (for 5 h) or RPMI medium only (untreated). MEC1 cells were significantly mobilized (*P=0.02) from the scaffolds. (C) On the left, representative confocal images of the scaffolds analyzed in panel (B), and on the right the histogram showing the mean number of MEC1 GFP+ cells quantified in the scaffold by counting the GFP+ cells in four different stacks for both treated and untreated conditions (P=0.002). (D) Line plot illustrating the effect of 10 μM ibrutinib treatment on the mobility of primary peripheral blood-derived CLL cells that were recovered outside the scaffold (***P=0.0005). (E) Representative confocal images of examples of the scaffolds analyzed in panel (D).

haematologica | 2021; 106(9)

2339

F. Barbaglio et al.

unmodified control cells (MEC1-CNTR) or MEC1-HS1KD cells with HS5 stromal cells as described above (Figure 1A) and observed that, outside the scaffolds, there were significantly fewer MEC1-HS1KD cells than MEC1-CNTR cells (P=0.0013) (Figure 1D). We then used quantitative real-time polymerase chain reaction (RT-qPCR) analysis to quantify the expression of HS1 in MEC1-CNTR cells cultured in 3D and found that HS1 was downregulated in MEC1 cells retained inside the scaffold compared to its level in the cells outside the scaffold (n=3 replicates; P=0.028) (Figure 1E). Collectively, these findings indicate that our 3D system can reliably reproduce the BM-CLL interactions occurring in vivo and further underscore the relevance of HS1 downregulation as a putative mechanism associated with CLL cell retention in the BM microenvironment, as previously suggested in mouse models.9

The bone marrow microenvironment regulates HS1 expression in primary chronic lymphocytic leukemia cells in 3D co-culture We then co-cultured primary leukemic CLL cells isolated from the PB of six patients with CLL in the Spongostan scaffolds in the presence and absence of HS5 cells, as represented in Figure 1A. We also confirmed for primary cells that stromal HS5 cells are needed to efficiently retain primary CLL cells in the scaffolds (Figure 2A). To further validate our model for CLL, we confirmed, by flow cytometry, the ability of CLL cells to retain the surface expression of their lineage markers CD19/CD5 throughout the whole culture period both inside and outside the scaffolds (Online Supplementary Figure S1A). We then quantified the expression of CXCR4 by RTqPCR and flow cytometry in CLL cells recovered from inside and outside the scaffolds after 3 days of co-culture with HS5 cells and found that CXCR4 was downregulated in the cells retained inside the scaffold (n=8, P=0.015 Figure 2B; n= 3, P=0.03, Online Supplementary Figure 1C), mimicking the in vivo finding of CXCR4 downregulation in response to the binding of its cognate ligand CXCL12 (SDF1α).21 Next, we focused again on the HS1 gene and, similar to the findings in MEC1 cells (Figure 1E), we observed a sig-

nificant downregulation of HS1 in the CLL fraction inside the scaffolds as compared to the outside fraction (n=8 replicates, P=0.005) (Figure 2C). These findings raised the question of whether the expression of HS1 reflected a direct influence of the BM microenvironment or, conversely, a fraction of CLL cells constitutively expressing low levels of HS1 might preferentially home to the BM. In order to answer this question, we co-cultured primary CLL cells isolated from the PB of ten patients with the stromal cell line HS5 in two-deimensional (2D) monolayers, and evaluated the expression of HS1 by RT-qPCR. HS1 expression was significantly downregulated after 24 h of co-culture (n=15, P<0.0001) (Figure 2D). We then co-cultured CLL primary cells from the PB of eight patients with HS5 cells in the presence and absence of a trans-well (1 μm pore) to avoid direct tumor-stroma contact. HS1 expression was not downregulated in the presence of the trans-well (Figure 2E), suggesting that its regulation requires direct contact between leukemic cells and the BM stromal microenvironment.

HS1 is heterogeneously expressed in chronic lymphocytic leukemia tissues The results described above suggest that HS1 might be differentially expressed in CLL depending on the tissue in which the leukemic cells are located. To assess whether our 3D model recapitulates what occurs in vivo, we compared HS1 expression in primary human CLL cells from PB and BM. Using RT-qPCR, we observed that HS1 expression was significantly downregulated in CLL cells isolated from the BM as compared with its expression in paired samples isolated from the PB (n=9, P=0.0498) (Figure 3A). In the same cohort of patients, we also confirmed that CXCR4 expression was downregulated in the BM as compared to the PB (n=10, P=0.003) (Online Supplementary Figure S1B). When we quantified HS1 expression at a single-cell level in CLL cells isolated from paired PB and BM samples using the Image Stream instrument,22 we observed that the majority of CLL cells in the PB strongly expressed HS1, while in the BM a fraction of CLL cells were HS1 negative (n=4, P=0.0187) (Figure 3B). Accordingly, immunohistochemistry performed on BM (n=4 patients analyzed) revealed a het-

Figure 5. HS1 activation following ibrutinib treatment. (A) Dot plot showing the number of chronic lymphocytic leukemia (CLL) cells mobilized outside the scaffold after ibrutinib treatment and fractionated according to HS1 activation. CLL cells with active HS1 were mobilized more efficiently from the scaffolds compared to those with inactive HS1 (*P=0.04). (B) The graph shows the densitometric analysis of the active HS1-Y378 phosphorylated form in CLL cells retained inside the scaffold or recovered from outside the scaffolds with or without ibrutinib treatment. The CLL cells mobilized and recovered outside the scaffold show significantly higher HS1 activation than that of the cells retained inside the scaffold (**P=0.0035).

2340

haematologica | 2021; 106(9)

3D co-culture model of CLL cells within BM microenvironment

erogeneous pattern of HS1 expression among the patients (Figure 3C). This finding confirms that our 3D model can reliably reproduce the native BM microenvironment and provide significant insights into CLL cells in the tissues.

Chronic lymphocytic leukemia cells are mobilized from the scaffolds following exposure to ibrutinib On the basis of the evidence presented here, HS1 appears to be involved in CLL cell compartmentalization, prompting the question of whether it could also be involved in the process of CLL cell mobilization from the tissues. In order to address this point, we exploited our 3D model, as described in Figure 4A, and evaluated whether the cytoskeletal activity of HS1 also plays a role in the CLL cell mobilization promoted by the BTK inhibitor ibrutinib.16 MEC1 cells co-cultured with HS5 cells within scaffolds

were efficiently mobilized upon 5 h of treatment with ibrutinib, as shown by both the number of MEC1 cells recovered in the medium (Figure 4B) and by the confocal images of the scaffolds, which exhibited significantly fewer GFPtagged MEC1 cells in the untreated condition (P=0.002) (Figure 4C, Online Supplementary Movies 1-2). Of note, HS5 cells were not mobilized by the drug (Online Supplementary Figure S2A). We next studied the response of primary CLL samples (n=21) to ibrutinib and found that the number of cells outside the scaffolds was significantly higher upon drug treatment than the number in untreated samples (n=21, P=0.0005) (Figure 4D). Consistent with this, confocal microscopy analysis performed on the scaffolds showed that they were depopulated of CLL cells after incubation with ibrutinib (Figure 4E).

Figure 6. HS1 activation and expression in patients during ibrutinib treatment. Top left panel: schematic representation of the experimental protocol. Peripheral blood (PB) was collected at 1, 2, 3, 4, 8 and 12 weeks after drug treatment, PB lymphocyte (LY) count was determined at each time-point with a hemocytometer and chronic lymphocytic leukemia cells were isolated and stored frozen at -80°C. For each patient, we plotted the western blot densitometry quantification of HS1-Y378 and the number of lymphocytes in the PB. The values are normalized to the basal level (n=8 patients).

haematologica | 2021; 106(9)

2341

F. Barbaglio et al.

As expected, the expression levels of HS1 of cells moving out of the scaffolds decreased while they remained lower in CLL cells inside the scaffold in both untreated and treated (ibritunib) settings (Online Supplementary Figure S2B). Conversely, CXCR4 expression on CLL cells inside the scaffold decreased following exposure to ibrutinib, confirming previous in vivo results from Chen et al.17 (P=0.03) (Online Supplementary Figure S2C). In parallel, we observed that CXCL12 levels in the medium did not change significantly during the drug treatment (Online Supplementary Figure S2D).

Cells with inactive HS1 are less efficiently mobilized following ibrutinib treatment To elucidate the mechanism underlying CLL mobilization, we evaluated whether HS1-mediated cytoskeletal rearrangements might be involved. We have previously shown that HS1 activation differs among patients with CLL and is associated with the clinical course (active HS1 is associated with a favorable prognosis while inactive HS1 is associated with an adverse prognosis).10 CLL cells with active HS1 (carrying HS1 phosphorylated in Y378, as determined by western blot analysis; data not shown) show efficient cytoskeletal functionality, while CLL cells with inactive HS1 (not phosphorylated in HS1-Y378) display reduced cytoskeletal activity associated with a higher propensity to accumulate within the BM microenvironment.10 We detected that both CLL cells with active HS1 and those with inactive HS1 were capable of homing to the scaffolds without significant differences (data not shown). Interestingly, we observed that the CLL cells released from the scaffolds upon exposure to ibrutinib were enriched for those with active HS1 (10 cases) as compared to those with inactive HS1 (10 cases) (P=0.04) (Figure 5A). Similarly, when we studied, in the same co-culture model, HS1 activation, i.e. the levels of HS1-Y378 determined by western blot, in primary CLL cells (n=7) exposed or not to ibrutinib, we observed that the levels of HS1-Y378 were higher in the cells released into the supernatant than in those retained inside the scaffolds (n=7, P=0.0035) (Figure 5B). These results demonstrate that more aggressive CLL cells (i.e., those with inactive HS1) are less efficiently mobilized from the BM surrogate scaffold and confirm that the segregation of CLL cells between the two compartments is not random but rather affected by the drug that, in turn, mediates changes in HS1 activation. Arguably, therefore, our 3D model may discriminate between cases that will respond more robustly to the CLL cell-mobilizing effect of ibrutinib.

found that HS1 expression increased during the first weeks of treatment in all patients analyzed either at the protein or at the gene level; however, we could not find a correlation with circulating lymphocyte count (Online Supplementary Figure S3A). Collectively, these results indicate that our 3D model can mirror the events occurring in vivo during ibrutinib therapy.

Discussion The clinical scenario of CLL is rapidly changing, in particular thanks to new targeted therapies,23 although the disease is still incurable. CLL is strongly influenced by the tissue microenvironment, as evidenced by the fact that circulating CLL cells are more sensitive to drug-induced apoptosis, suggesting that a supportive microenvironment is necessary for the survival of leukemic cells. This has encouraged the development of mobilizing agents24 and points to the key role of the cytoskeleton in recirculation and accumulation of CLL cells in different tissues. A key prerequisite for investigating the mechanisms underlying human CLL cell homing and mobilization is the capability to reliably and accurately reproduce a native CLL tissue microenvironment in vitro, which is so far unavailable. Our previous studies pointed to the importance of the BM microenvironment, showing specific homing of aggressive CLL (with inactive HS1) to this site in vivo in mouse models.9 For this reason, following our previously published experience from the analysis of multiple myeloma cell survival and response to borte-

Chronic lymphocytic leukemia cells mobilized into the peripheral blood by ibrutinib show active HS1 during the first weeks of treatment Finally, we analyzed HS1 activation in CLL cells of patients (n=8) under ibrutinib treatment for different periods (from week 1 to week 12) and correlated it with the lymphocyte count in the PB at the different time-points. In accordance with the results obtained in our 3D model (Figure 5B), we observed that PB CLL cells from six of the eight patients displayed HS1 activation during the first weeks (weeks 1, 2, and 3) of treatment as compared to the basal level (Figure 6A), in parallel with an increase in lymphocyte count in the PB. In contrast, peripheral lymphocytosis was not seen in the two patients in whom HS1 did not undergo activation. In parallel, we analyzed HS1 expression in CLL cells by both western blotting and RT-qPCR and 2342

Figure 7. Schematic summary of the presented model. CLL: chronic lymphocytic leukemia; BM: bone marrow.

haematologica | 2021; 106(9)

3D co-culture model of CLL cells within BM microenvironment

zomib,19 we customized, for CLL, a 3D BM model based on scaffolds within a bioreactor. The model allows us to analyze in parallel CLL cells retained inside the reconstructed BM microenvironment and those recovered from outside. We were able to define BM stromal cells as the minimal requirement to support CLL cell retention and homing inside the scaffolds, paving the way for future improvements aimed at testing the individual contribution of additional types of cells in this process. The observation that BM stromal cells support the retention of a consistent albeit variable fraction of primary CLL cells among different patients indicates that CLL compartmentalization in a bioreactor is not a random phenomenon. The demonstration of differential HS1 expression and activation status of CLL cells inside and outside the scaffolds, along with a similar modulation of CXCR4 expression, indicates a potential molecular basis for this process that does indeed mirror the events taking place in vivo in the BM and PB. Kinase inhibitors, such as the BTK inhibitor ibrutinib, influence the kinetics of leukemic cell recirculation and, interestingly, mobilize CLL cells more efficiently from the lymph nodes than from the BM, suggesting a specific tissue-dependent effect.25 Taking advantage of our 3D BM model, we here provide evidence that CLL cells mobilized from the scaffolds upon exposure to ibrutinib are mainly those with active HS1 and suggest that ibrutinib may exert its mobilizing effect through HS1 activation. It remains to be elucidated whether ibrutinib affects HS1 activation on CLL cells directly or indirectly in the scaffold. We have evidence that ibrutinib does not affect HS1 activation in the absence of BM-derived stromal cells (data not shown), suggesting a specific role exerted by the microenvironment, possibly toward HS1 downregulation, following direct contact with BM-derived stromal cells. Accordingly, we observed that HS1 undergoes activation during the first weeks of ibrutinib treatment in patients. We may then infer that CLL cells with low/inactive HS1 preferentially home to the BM niche where they encounter a protective microenvironment against the mobilization effect promoted by ibrutinib. This further indicates that our 3D model may reproduce the events occurring in vivo under ibrutinib treatment and may help to understand the slower clearance of the BM in patients.25 In conclusion, we here present and validate a reproducible

References 1. Caligaris-Cappio F, Bertilaccio MT, Scielzo C. How the microenvironment wires the natural history of chronic lymphocytic leukemia. Semin Cancer Biol. 2014;24:4348. 2. Burger JA, Gribben JG. The microenvironment in chronic lymphocytic leukemia (CLL) and other B cell malignancies: insight into disease biology and new targeted therapies. Semin Cancer Biol. 2014;24:71-81. 3. Herishanu Y, Perez-Galan P, Liu D, et al. The lymph node microenvironment promotes Bcell receptor signaling, NF-kappaB activation, and tumor proliferation in chronic lymphocytic leukemia. Blood. 2011;117(2):563574. 4. Calissano C, Damle RN, Hayes G, et al. In vivo intraclonal and interclonal kinetic heterogeneity in B-cell chronic lymphocytic

haematologica | 2021; 106(9)

3D BM model to aid understanding of the mechanisms underlying CLL tissue retention and mobilization, but also capable of predicting patient-specific efficacy of CLL mobilizing agents,23 as schematically summarized in Figure 7. This may serve in the future as a precision medicine tool to test these and other drugs acting through similar mechanisms in a more suitable system than the traditional 2D models from which the dynamic effects of treatments cannot be inferred. Moreover, it represents the first step towards the development of new and more complex 3D in-vitro models mimicking different microenvironments such as lymph nodes. Disclosures PG and KS have received honoraria and research funding from AbbVie and Janssen not related to this project. The other authors have no conflicts of interest to disclose. Contributions CS wrote the manuscript. CS, FB, DB, FVS and LP performed the experiments and analyzed the data. LS, PG and KS provided patients' and clinical information. MP and LB performed immunohistochemistry. DZ performed image stream analysis. VRC helped in the microscopy. EF, VRC, PG, MP and KS revised the manuscript. Acknowledgments We thank Alembic for helpful suggestions and technical support. Funding This project was supported in part by: the Associazione Italiana per la Ricerca sul Cancro AIRC (Special Program on Metastatic Disease – 5 per mille #21198); My first grant AIRC (#17006) (principal investigator CS). The research leading to these results received funding from AIRC under IG 2018 - ID. 21332 project (principal investigator CS). Roche per la ricerca 2016. Leukemia Research Foundation grant 2018; GCH-CLL project funded by ERA NET TRANSCAN-2 Joint Transnational Call for Proposals 2014 (JTC 2014) and project #179 NOVEL funded by ERA-NET TRANSCAN-2 JTC 2016; by the European Commission/DG Research and Innovation. CNIC is supported by the Ministerio de Ciencia, Innovación y Universidades and the Pro CNIC Foundation, and it is a Severo Ochoa Center of Excellence (SEV2015-0505). VRC acknowledges the support of FEDER "Una manera de hacer Europa".

leukemia. Blood. 2009;114(23):4832-4842. 5. Ponzoni M, Doglioni C, Caligaris-Cappio F. Chronic lymphocytic leukemia: the pathologist's view of lymph node microenvironment. Semin Diagn Pathol. 2011;28(2):161166. 6. Burger JA. Chemokines and chemokine receptors in chronic lymphocytic leukemia (CLL): from understanding the basics towards therapeutic targeting. Semin Cancer Biol. 2010;20(6):424-430. 7. Yamanashi Y, Okada M, Semba T, et al. Identification of HS1 protein as a major substrate of protein-tyrosine kinase(s) upon Bcell antigen receptor-mediated signaling. Proc Natl Acad Sci U S A. 1993;90(8):36313635. 8. Yamanashi Y, Fukuda T, Nishizumi H, et al. Role of tyrosine phosphorylation of HS1 in B cell antigen receptor-mediated apoptosis. J Exp Med. 1997;185(7):1387-1392.

9. Scielzo C, Bertilaccio MT, Simonetti G, et al. HS1 has a central role in the trafficking and homing of leukemic B cells. Blood. 2010;116(18):3537-3546. 10. ten Hacken E, Scielzo C, Bertilaccio MT, et al. Targeting the LYN/HS1 signaling axis in chronic lymphocytic leukemia. Blood. 2013;121(12):2264-2273. 11. Hasan MK, Yu J, Chen L, et al. Wnt5a induces ROR1 to complex with HS1 to enhance migration of chronic lymphocytic leukemia cells. Leukemia. 2017;31(12):26152622. 12. Uy GL, Rettig MP, Motabi IH, et al. A phase 1/2 study of chemosensitization with the CXCR4 antagonist plerixafor in relapsed or refractory acute myeloid leukemia. Blood. 2012;119(17):3917-3924. 13. Burger JA. Targeting the microenvironment in chronic lymphocytic leukemia is changing the therapeutic landscape. Curr Opin Oncol.

2343

F. Barbaglio et al. 2012;24(6):643-649. 14. Thompson PA, Burger JA. Bruton's tyrosine kinase inhibitors: first and second generation agents for patients with chronic lymphocytic leukemia (CLL). Expert Opin Investig Drugs. 2018;27(1):31-42. 15. Komarova NL, Burger JA, Wodarz D. Evolution of ibrutinib resistance in chronic lymphocytic leukemia (CLL). Proc Natl Acad Sci U S A. 2014;111(38):13906-13911. 16. Herman SE, Mustafa RZ, Jones J, Wong DH, Farooqui M, Wiestner A. Treatment with ibrutinib inhibits BTK- and VLA-4-dependent adhesion of chronic lymphocytic leukemia cells in vivo. Clin Cancer Res. 2015;21(20):4642-4651. 17. Chen SS, Chang BY, Chang S, et al. BTK inhibition results in impaired CXCR4 chemokine receptor surface expression, signaling and function in chronic lymphocytic

2344

leukemia. Leukemia. 2016;30(4):833-843. 18. Pasikowska M, Walsby E, Apollonio B, et al. Phenotype and immune function of lymph node and peripheral blood CLL cells are linked to transendothelial migration. Blood. 2016;128(4):563-573. 19. Belloni D, Heltai S, Ponzoni M, et al. Modeling multiple myeloma-bone marrow interactions and response to drugs in a 3D surrogate microenvironment. Haematologica. 2018;103(4):707-716. 20. Hallek M, Cheson BD, Catovsky D, et al. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-Working Group 1996 guidelines. Blood. 2008;111(12): 5446-5456. 21. Burger JA, Burger M, Kipps TJ. Chronic lym-

phocytic leukemia B cells express functional CXCR4 chemokine receptors that mediate spontaneous migration beneath bone marrow stromal cells. Blood. 1999;94(11):36583667. 22. Jayappa KD, Portell CA, Gordon VL, et al. Microenvironmental agonists generate de novo phenotypic resistance to combined ibrutinib plus venetoclax in CLL and MCL. Blood Adv. 2017;1(14):933-946. 23. Schiattone L, Ghia P, Scarfo L. The evolving treatment landscape of chronic lymphocytic leukemia. Curr Opin Oncol. 2019;31(6):568573. 24. Burger JA. The CLL cell microenvironment. Adv Exp Med Biol. 2013;792: 25-45. 25. Badar T, Burger JA, Wierda WG, O'Brien S. Ibrutinib: a paradigm shift in management of CLL. Expert Rev Hematol. 2014;7(6):705717.

haematologica | 2021; 106(9)

ARTICLE

Chronic Lymphocytic Leukemia

Impaired nodal shrinkage and apoptosis define the independent adverse outcome of NOTCH1 mutated patients under ibrutinib therapy in chronic lymphocytic leukemia Giovanni Del Poeta,1 Annalisa Biagi,1 Luca Laurenti,2 Annalisa Chiarenza,3 Federico Pozzo,4 Idanna Innocenti,2 Massimiliano Postorino,1 Francesca Maria Rossi,4 Maria Ilaria Del Principe,1 Riccardo Bomben,4 Paolo de Fabritiis,1 Antonio Bruno,1 Maria Cantonetti,1 Francesco Di Raimondo,3 Antonella Zucchetto4 and Valter Gattei4 Division of Hematology, Department of Biomedicine and Prevention, University Tor Vergata, Roma; 2Division of Hematology, Università Cattolica del Sacro Cuore, Roma; 3 Division of Hematology, Ferrarotto Hospital, Catania and 4Clinical and Experimental Hematology Unit, Centro di Riferimento Oncologico di Aviano (CRO), IRCCS, Aviano (PN), Italy. 1

Ferrata Storti Foundation

Haematologica 2021 Volume 106(9):2345-2353

ABSTRACT

he introduction of agents inhibiting the B-cell receptor-associated kinases such as ibrutinib has dramatically changed treatments algorithms of chronic lymphocytic leukemia (CLL) as well as the role of different adverse prognosticators. We evaluated the efficacy of ibrutinib as a single agent, in a real-life context, in 180 patients with CLL mostly pretreated, recruited from three independent cohorts from Italy. Patients received 420 mg oral ibrutinib once daily until progression or occurrence of unacceptable side effects. Seventy-three patients discontinued ibrutinib for progression or for adverse events. NOTCH1 mutations (NOTCH1 M) were correlated with a reduced redistribution lymphocytosis, calculated at 3 months on ibrutinib (P=0.022). Moreover, NOTCH1 M patients showed inferior nodal response at 6 months on ibrutinib compared to NOTCH1 wild-type patients (P<0.0001). Significant shorter progression free survival (PFS) and overall survival (OS) were observed in NOTCH1 M patients (P=0.00002 and P=0.001). Interestingly, NOTCH1 M plus a lower BAX/BCL-2 ratio identified a CLL subset showing the worst PFS and OS (P=0.0002 and P=0.005). In multivariate analysis of PFS and OS, NOTCH1 M were confirmed an independent prognosticator (P=0.00006 and P=0.0039). In conclusion, NOTCH1 M are strongly associated with a lower BAX/BCL-2 ratio, consistent with defective apoptosis, lower redistribution lymphocytosis and lower nodal shrinkage under ibrutinib treatment, this last paramter being responsible for partial responses, subsequent relapses, as well as shorter PFS and OS. Either new small molecule combination approaches or antibodies targeting NOTCH1 could be future therapeutic options for NOTCH1 M patients.

Introduction Chronic lymphocytic leukemia (CLL) is the most frequent adult leukemia in Western countries and it is characterized by an extremely heterogeneous clinical course. New molecular aberrations with negative prognostic value in CLL, such as NOTCH1, MYD88, TP53 and SF3B1 gene mutations, were identified in the last decade mainly thanks to the advent of next-generation sequencing (NGS).1,2 In particular NOTCH1 mutations (M) are found in 10-14% of patients at diagnosis with frequency increasing with disease progression and during transformation to Richter syndrome.3 Furthermore NOTCH1 M are associated with the presence of trisomy 12 and with high CD49d expression which are negative prognostic factors in CLL. NOTCH1 M are also associated with an increased activation of the NF-kB pathway, promoting tumour cell proliferation and survival.4 NOTCH1 M were shown to affect

haematologica | 2021; 106(9)

Correspondence: GIOVANNI DEL POETA g.delpoeta@tin.it Received: February 29, 2020. Accepted: July 20, 2020. Pre-published: July 30, 2020. https://doi.org/10.3324/haematol.2020.251488

2345

G. Del Poeta et al.

the response to chemo-immunotherapy in CLL. In the CLL8 study Stilgenbauer et al.5 demonstrated that patients carrying NOTCH1 M did not benefit of the addition of rituximab to standard fludarabine and cyclophosphamide chemotherapy. Moreover, it has emerged that NOTCH1 M are associated with decreased duration of response in a large series of relapsed/refractory (R/R) patients treated with venetoclax.6 In a recent study, Tissino et al.7 have demonstrated that patients with CLL whose cells were characterized by high CD49d expression, underwent reduced lymphocytosis and inferior nodal response after treatment with ibrutinib. Several reports confirmed that in CLL the balance between the pro- and anti-apoptotic members of the BCL-2 family determines chemotherapy sensitivity and cell survival.8,9 Noteworthy, we demonstrated that a low BAX/BCL-2 ratio had an additive negative prognostic impact in both TP53 M and NOTCH1 M patients with CLL treated with chemoimmunotherapy.10 The recent introduction of novel B-cell receptor inhibitors such as ibrutinib and idelalisib and of novel potent oral BH3 peptidomimetics such as venetoclax in clinical practice, prompted us to evaluate the clinical impact of both NOTCH1 M and BAX/BCL-2 ratio in patients treated with targeted oral therapies and in particular in those treated with ibrutinib. The aims of this study were: i) to analyse the correlations between NOTCH1 M and other biological parameters including CD49d expression and the BAX/BCL-2 ratio; ii) to address the impact of NOTCH1 M both on redistribution lymphocytosis and on nodal responses after treatment with ibrutinib; iii) to evaluate the impact of NOTCH1 M and BAX/BCL-2 ratio on the overall response rate (ORR) to ibrutinib, progression free survival (PFS) and overall survival (OS); iiii) to assess whether NOTCH1 M could be considered an independent prognostic factor.

Methods Study design and patients In this study we retrospectively analysed 180 patients with CLL exposed to treatment with ibrutinib. Patients were recruited from three independent cohorts from Italy (Rome Tor Vergata University, Rome Cattolica Sacro Cuore University and Catania Ferrarotto Hospital), between 2014 and 2019. Informed consent was obtained in accordance with the Declaration of Helsinki. The study was performed under the Institutional Review Board of the Centro di Riferimento Oncologico (IRCSS) of Aviano (approval numbers: IRB-05-2010 and IRB-05-2015). Patients were 122 males and 58 females with a median age of 69 years (range, 36-90). According the modified Rai staging system,11 134 patients had an intermediate risk and 46 a high risk stage. All these parameters were considered at the time ibrutinib was initiated. All patients received 420 mg oral ibrutinib (Imbruvica; Janssen, Beerse, Belgium) once daily until progression or occurrence of unacceptable side effects. Median number of previous chemotherapy regimens were two (range, 0-4). Patients receiving first-line ibrutinib were 26 (14.4%), of whom 24 of 26 (92%) were TP53 mutated. Median follow-up was 25 months (range, 10-61). Seventy-three patients (40.6%) discontinued ibrutinib either for progression (n=42) or for adverse events (n=31) (Table 1): 32 patients were subsequently treated with venetoclax (11 for toxicity [grade 3 or 4 World Health Organization] and 21 for progression of disease), five patients

2346

were treated with idelalisib, and the remaining 36 patients received other lines of chemotherapy (n= 12) or no therapy (n=24). The clinical characteristics of patients are reported in Table 1. The clinical assessment of patients with CLL to establish diagnosis and response to therapy were based both on the International Workshop on Chronic Lymphocytic Leukemia (iwCLL) criteria.12 The clinical impact of NOTCH1 M and BAX/BCL-2 ratio on ibrutinib treatment was evaluated by measuring the kinetics of absolute lymphocyte counts (ALC), the reduction of lymphadenopathy, and the clinical outcome, as defined by ORR, PFS and OS.

Chronic lymphocytic leukemia characterization Flow cytometry was employed for immunophenotypical CLL characterization and was performed with FACSCalibur or FACSCanto I flow cytometer. BCL-2 and BAX oncoproteins were analysed by flow cytometry in samples taken before starting ibrutinib. BAX/BCL-2 ratio was calculated dividing mean fluorescence intensity (MFI) of BAX by MFI of BCL-2 on CLL cells, as previously described.10 The threshold of positivity was set at ≥1.5. Immunoglobulin heavy-chain variable region gene (IGHV) mutational status was performed by NGS, as previously described.13,14 TP53 exons 2 to 11 mutational status and NOTCH1 exon 34 and 3’ untranslated (UTR) region mutational status were analysed by NGS, as previously described.4,7 CLL samples were considered

Table 1. Patient characteristics (n = 180)

No. of patients/Total cases (%) Observation time Median age, y (range) Males Modified Rai stage Intermediate High Number of previous regimens 0 1 2 3 4 NOTCH1 mutation BAX/BCL-2 ratio <1.50 Trisomy 12 11q deletion TP53 mutations/17p deletion UM IGHV CD38 ≥ 30% CD49d ≥ 30% Median follow up (months) Response to ibrutinib therapy Complete response Partial response Partial response with lymphocytosis Stable disease/No response Discontinuation Progression Toxicity Richter Syndrome Progression-free Survival at 2 years Overall Survival at 2 years Overall Survival at 4 years

2014-2019 69 (36-90) 122/180 (68) 134 (74) 46 (26) 26/180 (14.5) 56/180 (31.1) 74/180 (41.1) 22/180 (12.2) 2/180 (1.1) 65/180 (36.1) 74/113 (65.5) 23/179 (13) 35/179 (20) 66/178 (37.1) 123/175 (70.3) 54/113 (47.8) 108/179 (60.3) 25 (10-61) 33/180 (18.3) 51/180 (28.3) 81/180 (45.1) 15/180 (8.3) 73/180 (40.6) 42/180 (35) 31/180 (65) 13/180 (7.2) 80% 84% 71%

y: years; IGHV: immunoglobulin heavy-chain variable region gene. No.: number.

haematologica | 2021; 106(9)

NOTCH1 mutations in CLL

mutated for NOTCH1 i.e., NOTCH1 M, if exceeding a variant allele frequency (VAF) of 1%.15,16

Redistribution lymphocytosis and nodal response The redistribution lymphocytosis was calculated as percent variation of ALC over the baseline values. Nodal response was calculated as percent reduction in sum of the product of diameter (SPD) values on the major lymph node regions over the baseline measurement, as reported previously.17 Additional details on the employed procedures and methods are reported in the Online Supplementary Materials and Methods.

Results NOTCH1 mutations and BAX/BCL-2: correlations with other biological parameters Sixty-five patients were NOTCH1 M (65 of 180, 36.11%), with VAF levels >1 (Online Supplementary Table 2S). With regard to the distribution of VAF levels, 21 patients had VAF between 1% and 10%, seven patients between 10.5% and 20% and 37 patients above 20%. Fifty-six NOTCH1 M cases bore a single mutation, eight cases two mutations and one case three mutations. NOTCH1 M cases were classified as follows: 45 delCT, six frameshift other than delCT (FS), 8 3’-UTR and six considering both missense (one) and nonsense (five) mutations (Online Supplementary Table S2). Seventy-four patients showed a BAX/BCL-2 ratio lower than 1.5 (74 of 113, 65.5%). NOTCH1 M were significantly associated with SPD ratio<1.5: in fact, 34 of 38 NOTCH1 M patients showed BAX/BCL-2 ratio less than 1.5 (P=0.0001). Moreover, NOTCH1 M were strongly correlated with CD49d expression: 51 patients were both NOTCH1 M and CD49d ≥30% (P=0.0001). Furthermore, a significant corre-

lation was found between a lower BAX/BCL-2 ratio and CD38>30% (41 of 54; P=0.030) as well as between CD38>30% and NOTCH1 M (27 of 38 patients; P=0.0004) (Table 2; Online Supplementary Table S3). Trisomy 12 was confirmed to be strongly correlated with NOTCH1 M (18 of 23; P=0.0002). There was only a trend towards significant association between NOTCH1 M and IGHV UM status (48 of 62; P=0.08). On the other hand, IGHV UM status was correlated with lower BAX/BCL-2 ratio (58 of 81; P=0.030). TP53 M and/or del17p were found in 66 of 178 patients (37.1%). Noteworthy, 23 of 178 patients (13%) were simultaneously NOTCH1 and TP53 mutated. The distribution of clinical and biological prognostic factors according to NOTCH1 M is shown in Table 2. The distribution of prognostic factors according to the BAX/BCL-2 ratio and CD38 was obtained in 113 patients from Rome and shown in Table 2 and the Online Supplementary Table S3.

Relevance of NOTCH1 mutations as biological prognostic parameter The mean peripheral lymphocyte percentage change from baseline, calculated at 3 months on ibrutinib, was lower in NOTCH1 M patients than in NOTCH1 wild-type (WT) patients (14% vs. 54%; P=0.022, Mann-Whitney test), thus confirming a reduced redistribution lymphocytosis (Figure 1A). Moreover, the mean percent SPD change, calculated at 6 months on ibrutinib, was lower in NOTCH1 M patients than in NOTCH1 WT patients (53% vs. 80%; P<0.0001, MannWhitney test), confirming a significant poor nodal response (Figure 1B). Moreover, we compared NOTCH1 M plus lower BAX/BCL-2 ratio versus NOTCH1 M plus higher BAX/BCL-2 ratio with respect to redistribution lymphocy-

Figure 1. Box plots by NOTCH1 wild-type and mutated groups showed significant lower redistribution lymphocytosis after 3 months on ibrutinib treatment in NOTCH1 M patients (A) and equally lower sum of the product of diameter (SPD) values after 6 months in NOTCH1 M patients (B). pts: points; WT: wild-type; M: mutated; B LYMPHS: B lymphocytes.

haematologica | 2021; 106(9)

2347

G. Del Poeta et al.

Table 2. Distribution of prognostic factors in chronic lymphocytic leukemia according to NOTCH1 mutations

Parameter

Age <60 years >60 years Sex Male Female Mod-Rai Intermediate High Lines of therapy ≤2 >2 CD49d <30% >30% CD38 <30% >30% FISH Normal/del13q +12, 11q-,17p- del11q, del17p) IGHV Mutated Unmutated TP53 Mutated Wild-type BAX/BCL-2 ratio <1.5 >1.5

NOTCH1 Mutated Wild-type

P¶

n§

4-year OS,%

4-year PFS, %

31 34

54 61

0.52

180

85 95

0.29

81 89

0.72

45 20

77 38

0.44

180

122 58

0.71

115 55

0.55

10 55

30 85

0.32

180

140 40

0.81

40 130

0.33

55 10

101 14

0.54

180

156 24

0.004

148 22

0.002

14 51

57 57

0.0001

179

71 108

0.23

68 101

0.045

11 27

48 27

0.0004

113

59 54

0.52

56 50

0.36

24 41

46 68

0.0002

179

71 107

0.49

70 98

0.46

14 48

38 75

0.080

175

52 123

0.76

48 117

0.036

23 40

43 72

0.52

178

66 112

0.028

59 109

0.022

34 4

40 35

0.0001

113

74 39

0.013

67 39

0.0019

¶ Fisher exact tests were performed to evaluate the association between NOTCH1 mutations or wild-type and other prognostic factors. § Values refer to the number of cases analysed for a given feature. *P-values were calculated by the log-rank test in univariate analysis. PFS: progression-free survival; OS: overall survival; FISH: fluorescence in situ hybridisation; IGHV: immunoglobulin heavy-chain variable region gene.

tosis and lymph node shrinkage. No significant differences were found between these two subsets (Online Supplementary Figure S3).

Table 3. Multivariate Cox regression analysis

NOTCH1 mutations, BAX/BCL-2 ratio and their impact on clinical outcome

NOTCH1 M >2 lines of therapy TP53 M

According to clinical endpoints, ORR was 91% [complete response (CR): 18%, partial response (PR): 28%, PR with lymphocytosis (PR-L): 45%] (Table 1). The estimated 2-year and 4-year OS were 84% and 71%, respectively (Table 1; Online Supplementary Figure S4). Noteworthy, OS was longer in patients previously treated with one line of chemoimmunotherapy before ibrutinib (P=0.02, Online Supplementary Figure S5). PR and PR-L were significantly correlated with NOTCH1 M (30 of 65 and 22 of 65, respectively; P=0.00001, Online Supplementary Table S4). Of note, PR, PRL and chemoresistance were also associated with lower BAX7BCL-2 ratio (23 of 29, 33 of 52 and nine of nine, respectively; P=0.002, Online Supplementary Table S5). Interestingly, discontinuation due to disease progression was more frequent in NOTCH1 M patients than in NOTCH1 WT patients (P=0.034, Online Supplementary Table S4). Significant shorter PFS and OS were observed in NOTCH1 M patients (34% vs. 76% and 56% vs. 83% at 3 years, respectively; P=0.00002 and P=0.001; Figure 2A and B). There were no significant differences among VAF range 2348

Parameter

PFS 168 patients HR P

3.89 2.88 2.05

0.00006 0.0040 0.028

OS 178 patients HR P

2.64 2.43 1.94

0.0039 0.015 0.047

PFS: progression-free survival, OS: overall survival: M: mutant; HR: hazard ratio.

1-10%, 10.5-20% and above 20% with respect to PFS and OS, as shown in the Online Supplementary Figures S6 and S7. Moreover, we restricted the analysis of NOTCH1 to the relapse setting only (154 of 180 patients) obtaining similar significant results regarding PFS and OS (Online Supplementary Figures S8 and S9).

Additive prognostic properties of NOTCH1 mutations and BAX/BCL-2 ratio In order to obtain a better refinement in the prognostic assessment of PFS and OS, we combined the values of the BAX/BCL-2 ratio with those of NOTCH1. Within the subset of 113 patients from Rome, shorter PFS and OS were detected both in patients with NOTCH1 M (46% vs. 83% and 68% vs. 86% at 3 years, respectively; P=0.0019 and haematologica | 2021; 106(9)

NOTCH1 mutations in CLL

P=0.031, Online Supplementary Figure S10A and B) and with lower BAX/BCL-2 ratio (60% vs. 97% and 72% vs. 94% at 3 years, respectively; P=0.019 and P=0.013, Figure 3A and B). Therefore, higher or lower BAX/BCL-2 ratio combined with NOTCH1 WT or NOTCH1 M identified two subsets of patients, the former with the best prognosis and the latter with the worst prognosis with respect to both PFS (97% vs. 42%; P=0.0002, Figure 4A) and OS (94% vs. 63%;P=0.005, Figure 4B), confirming the true additive prognostic properties of these two prognosticators.

together with >2 previous lines of therapy (p=0.015) and TP53 M (p=0.047) (Table 3). NOTCH1 M and >2 previous lines of therapy were confirmed as independent prognosticators for PFS (P=0.035 and P=0.015, respectively) also in a model that included the BAX7BCL-2 ratio, available in a smaller subset of cases (n=113, Online Supplementary Table S6). Conversely, in the same subset of patients, no factor emerged as independent prognosticator for OS (Online Supplementary Table S6).

Multivariate analysis

Discussion

The clinical impact of NOTCH1 as independent prognosticator was checked by multivariate Cox proportional hazards analysis applied to models including two other prognosticators proven to be significant in univariate analysis (Table 2). With respect to PFS, NOTCH1 M (P=0.0002) were confirmed as an adverse independent prognostic factor (P=0.00006) together with >2 previous lines of therapy (P=0.004) and TP53 M (P=0.028) (Table 3). Similarly, in a multivariate analysis of OS, NOTCH1 M retained an independent prognostic value (P=0.0039)

In the present study we evaluated the efficacy of ibrutinib treatment in the high-risk NOTCH1 M CLL group and correlated NOTCH1 M to BAX/BCL-2 ratio, a value reflecting the susceptibility of cells to apoptosis. Efficacy of ibrutinib remained high at 4-year follow-up in almost all pre-treated patients with CLL, with 71% of patients alive and progression free, similarly to other studies.17 Moreover, ibrutinib was more effective in patients previously treated with only one line therapy, compared to patients previously treated

Figure 2. Progression-free survival and overall survival curves based on NOTCH1. Kaplan-Meier plot comparing progression-free survival (PFS) (A) and overall survival (OS) (B) based on NOTCH1. NOTCH1 mutated (NOTCH1 M) patients experienced both a shorter PFS and OS.

haematologica | 2021; 106(9)

2349

G. Del Poeta et al.

with >2 lines of therapy (Online Supplementary Figure S5). On the other hand, the clinical outcome was similar for patients receiving first-line ibrutinib and patients with one previous therapy, probably due to the high incidence of TP53 mutated patients (24 of 26) in first-line setting (Online Supplementary Figure S5). In CLL, the frequency of NOTCH1 M cases between 612% if evaluated at presentation, increases to about 1520% in the context of fludarabine refractory patients.18,19 The higher frequency of NOTCH1 M characterizing our cohort of patients (36%) could be attributed both to the previous lines of chemotherapy and to the very low cut-off (>1%) chosen for NOTCH1 M. The adverse clinical outcome of patients with NOTCH1 M CLL was confirmed in univariate analysis in several independent cohorts of patients treated with chemo-immunotherapy.20-24 Since clonal CLL cells accumulate because of prolonged survival due to impairment of apoptosis, the analysis of the BAX/BCL-2 ratio could be a valid tool to provide information on the chemo-sensitivity of CLL cells.9,10 We addressed the clinical impact of both NOTCH1 M,

evaluated by NGS, and BAX/BCL-2 ratio, determined by flow cytometry, in patients with CLL homogeneously treated with ibrutinib, mainly in a (R/R) setting. Determination of both parameters was done prior to starting ibrutinib therapy. The NGS approach used for NOTCH1 M analysis allowed detection of allele frequency as low as 1%, highlighting the presence of subclonal mutations in 32% of total NOTCH1 M cases.1,16,26 Of note, subclonal NOTCH1 M (i.e., VAF<10%) had similar prognostic impact as clonal mutations (Online Supplementary Figures S6 and S7); consistently a receiver operating characteristic curve analysis confirmed the use of 1% as optimal cut-off (Online Supplementary Figures S2). In this context, detection of NOTCH1 M by NGS could be viewed as a useful tool for clinical follow-up of patients as well as for minimal residual disease studies, although the latter use remains speculative at the moment. From a biological point of view, we found a significant relationship between NOTCH1 and some other prognosticators. In particular, a significant correlation between NOTCH1 M and higher CD49d or CD38 expressions was

Figure 3. Progression-free survival and overall survival curves based on BAX/BCL-2 ratio within the subset of 113 patients from Rome. Kaplan-Meier plot comparing progression-free survival (PFS) (A) and overall survival (OS) (B) based on the BAX/BCL-2 ratio. Patients with a BAX/BCL-2 ratio <1.5 experienced both shorter PFS and OS.

2350

haematologica | 2021; 106(9)

NOTCH1 mutations in CLL

observed, as well a trend towards an association between NOTCH1 M and IGHV UM status, in keeping with previous observations by us and others.27, 21,4,28 Further, co-occurrence of NOTCH1 M and TP53 M characterized 13% of our patients (23 of 178), a rather high percentage if compared to previous reports where concomitant NOTCH1 M and TP53 M, preferentially affecting the same leukemic cells,29 accounted for 1.2-2.6% of CLL patients.20,23 This may be due to the high number of pre-treated patients and to the low cut-off chosen by us for NOTCH1 M detection. We confirmed that NOTCH1 M were strongly correlated with trisomy 12, in line with previous reports describing a high NOTCH1 M rate in CLL cases with isolated trisomy 12 and a lower frequency in cases characterized by additional chromosomal abnormalities.30-32 In particular, a mutation frequency of 41.9% was reported in aggressive trisomy 12 cases, suggesting a pivotal role of NOTCH1 activation in this group.33 Moreover, we observed here a lower BAX/BCL-2 ratio in NOTCH1 M patients, in keeping with our previous studies showing NOTCH1-dependent activation of the NF-kB pathway that may result in the upregulation of target genes, including BCL-2.4

The strong correlation between lower BAX/BCL-2 ratio and NOTCH1 M suggests that the poor prognosis of NOTCH1 M patients may be related to the lack of apoptosis, although these observations need further confirmation. The variability in the degree and kinetics of ibrutinibinduced recirculation lymphocytosis has been highlighted by several studies,34,35 and was also confirmed in the present study. Here we show that at 3 months on ibrutinib, the typical ibrutinib-induced peak of lymphocytosis is observed in NOTCH1 WT patients, but not in NOTCH1 M cases. Moreover, even though the analysis of nodal response confirmed an overall significant reduction in organomegaly and lymph node size in most cases at 6 months on ibrutinib, NOTCH1 M cases experienced a significant lower nodal response compared to NOTCH1 WT cases. These results may be explained by the strong correlation between CD49d overexpression and NOTCH1 M (51 of 65 cases), in line with the reported involvement of the NOTCH1 pathway in the regulation of CD49d expression.4 Consistently, CD49d associates with nodal presentation and subsequent development of lymphadenopathy in patients with CLL.36

Figure 4. Progression-free survival and overall survival curves in relation to combined BAX/BCL-2 ratio and NOTCH1. Progression-free survival (PFS) and overall survival (OS) were shorter within the NOTCH1 mutated (NOTCH1 M) plus BAX/BCL-2 <1.5 subgroup (A-B), showing additive prognostic properties.

haematologica | 2021; 106(9)

2351

G. Del Poeta et al.

Moreover, CD49d expression identifies cases with reduced lymphocytosis and inferior nodal response upon ibrutinib treatment, suggesting the retention of CD49d-expressing cells in tissue sites via activated VLA-4 .7 Consistently with the high frequency of pre-treated patients in our cohort (154 of 180), the OS values at 2 and 4 years (84% and 71% respectively), were similar to those reported for the phase III RESONATE study in patients with previously treated CLL/SLL.37 We have recently reported that NOTCH1 M identify a subgroup of patients with CLL with worse prognosis in the setting of a rituximab-based induction and consolidation treatment.38 Here, we described a negative prognostic impact of NOTCH1 M also in the ibrutinib setting. Our findings differ from those resulting from the extended follow up from the RESONATE study of relapsed/refractory CLL, where the presence of NOTCH1 M did not negatively affect the efficacy of ibrutinib on disease progression outcomes.37 This difference can be explained by the very low cut-off (>1%) chosen for NOTCH1 M in our study, although for the validation of these findings additional independent cohorts are needed. The here reported capacity of BAX/BCL-2 index to identify patients with a different response to ibrutinib could be of interest in the light of the treatments protocols associating B-cell receptor inhibitors and BH3 mimetics such as venetoclax.39 Moreover, the additive negative prognostic value of NOTCH1 M and low BAX/BCL-2 ratio described by us, further support the rationale to improve the efficacy of ibrutinib by using the BCL-2 inhibitor venetoclax in patients with NOTCH1 mutated CLL.10 Interestingly, an additive prognostic impact of the combination of BAX/BCL-2 and NOTCH1 M in the setting of chemo-immunotherapy was also reported by us.10 Several independent cohorts of patients confirmed the adverse clinical outcome of NOTCH1 M with CLL in univariate analysis,20-23,40 although conflicting results are reported about its independent prognostic effect. In particular, NOTCH1 M did not retain independent significance as a predictor of time-to-first treatment in one of the largest series of patients with CLL,41 while in another study it emerged as an independent predictor of shorter survival, along with TP53 abnormalities.42 Here NOTCH1 M were confirmed to be an independent prognostic factor together with previous lines of therapy and TP53 both with respect to PFS and OS. The apparent higher prognostic impact of NOTCH1 M compared to TP53 mutation, as emerged in our multivariable analysis, may be explained by the greater number of TP53 mutated cases treated first line with ibrutinib, hence with a better prognosis than NOTCH1 M cases that were more frequently treated with ibrutinib in second or further lines of therapy.

References 1. 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. 2. Puente XS, Beà S, Valdés-Mas R, et al. Noncoding recurrent mutations in chronic lymphocytic leukaemia. Nature. 2015;526 (7574):519-524.

2352

The current use of B-cell receptor and BCL-2 inhibitors led to high-rate improvement of outcome in CLL. However, several issues remain, resulting in resistance/progression thus limiting the eradication of the tumour. The growing evidence for a critical role of the NOTCH1 pathway in CLL makes this cancer gene a target to design tailored treatments for this peculiar subset through specific NOTCH1targeted therapies. In this context, γ-secretase inhibitors are the most extensively explored anti-NOTCH1 molecules and their combination with fludarabine demonstrated antitumour effects in primary CLL with NOTCH1 M.43 Noteworthy, a humanized antibody targeting NOTCH1 (clinicaltrials gov. Identifier: OMP-52M51) entered phase I trial in relapsed/refractory lymphoid malignancies.44 However, to date, the future treatment of CLL with NOTCH1 M relies on the association of small molecule inhibitors targeting both the BCR pathway and the antiapoptotic BCL-2 protein. Disclosures No conflicts of interest to disclose. Contributions GDP and VG designed the study, interpreted data, performed statistical analysis, wrote the manuscript and gave final approval of the manuscript; AB and AZ contributed to study design and data interpretation and to write the manuscript; LL, AC and MIDP contributed to interpret the data and to write the manuscript; AZ, FMR and GDP obtained flow cytometric data; FMR performed FISH cytogenetic analysis; VG, FP and RB investigated IGHV, NOTCH1 and TP53 mutations; FB, SA, GG, AV contributed to study design and data interpretation; II, MP, PdF, MC recruited the patients and collected clinical data. Acknowledgments This study was supported in part by Ministero dell’Università e della Ricerca Scientifica e Tecnologica (MURST), Programmi di Ricerca di Interesse Nazionale; Ministero della Salute (Ricerca Finalizzata Istituto di Ricovero e Cura a Carattere Scientifico [IRCCS], Rome, Italy; Associazione Italiana Ricerca Cancro (AIRC), Investigator Grant IG-21687 (to V.G.); Progetto Ricerca Finalizzata PE 2016-02362756, Ministero della Salute, Rome, Italy (to V.G.); Progetto Ricerca Finalizzata RF-2018-12365790 (to A.Z.); Fondazione Cariplo (grant 2012-0689); Associazione Italiana contro le Leucemie, Linfomi e Mielomi (AIL), Venezia Section, Pramaggiore Group, Italy; Fondazione per la Vita di Pordenone, Italy; Ricerca Scientifica Applicata, Regione Friuli Venezia Giulia (“Linfonet” Project), Trieste, Italy; “5x1000 Intramural Program”, Centro di Riferimento Oncologico, Aviano, Italy. The authors thank the members of their Departments of Hematology clinical staff for their invaluable support to this CLL clinical research program.

3. Rossi D, Spina V, Bomben R, et al. Association between molecular lesions and specific B-cell receptor subsets in chronic lymphocytic leukemia. Blood. 2013;121(24):4902-4905. 4. Benedetti D, Tissino E, Pozzo F, et al. NOTCH1 mutations are associated with high CD49d expression in chronic lymphocytic leukemia: link between the NOTCH1 and the NF-κB pathways. Leukemia. 2018;32(3):654-662.

5. Stilgenbauer S, Schnaiter A, Paschka P, et al. Gene mutations and treatment outcome in chronic lymphocytic leukemia: results from the CLL8 trial. Blood. 2014;123(21):32473254. 6. Roberts AW, Ma S, Kipps TJ, et al. Efficacy of venetoclax in relapsed chronic lymphocytic leukemia is influenced by disease and response variables. Blood. 2019;134(2):111122. 7. Tissino E, Benedetti D, Herman SEM, et al.

haematologica | 2021; 106(9)

NOTCH1 mutations in CLL

Functional and clinical relevance of VLA-4 (CD49d/CD29) in ibrutinib-treated chronic lymphocytic leukemia. J Exp Med. 2018;215 (2):681-697. 8. Pepper C, Hoy T, Bentley P. Elevated Bcl2/Bax are a consistent feature of apoptosis resistance in B-cell chronic lymphocytic leukaemia and are correlated with in vivo chemoresistance. Leuk Lymphoma. 1998;28 (3-4):355-361. 9. Williamson KE, Kelly JD, Hamilton PW, et al. Bcl-2/Bax ratios in chronic lymphocytic leukaemia and their correlation with in vitro apoptosis and clinical resistance. Br J Cancer. 1998;78(4):553-554. 10. Del Principe MI, Dal Bo M, Bittolo T, et al. Clinical significance of bax/bcl-2 ratio in chronic lymphocytic leukemia. Haematologica. 2016;101(1):77-85. 11. Rai KR, Han T. Prognostic factors and clinical staging in chronic lymphocytic leukemia. Hematol Oncol Clin North Am. 1990;4(2):447-456. 12. Hallek M, Cheson BD, Catovsky D, et al. iwCLL guidelines for diagnosis, indications for treatment, response assessment, and supportive management of CLL. Blood. 2018;131(25):2745-2760. 13. Degan M, Bomben R, Dal Bo M, et al. Analysis of IgV gene mutations in B cell chronic lymphocytic leukaemia according to antigen-driven selection identifies subgroups with different prognosis and usage of the canonical somatic hypermutation machinery. Br J Haematol. 2004;126(1):2942. 14. Bomben R, Dal Bo M, Zucchetto A, et al. Mutational status of IgV(H) genes in B-cell chronic lymphocytic leukemia and prognosis: percent mutations or antigen-driven selection? Leukemia. 2005;19(8):1490-1492. 15. Thorvaldsdóttir H, Robinson JT, Mesirov JP. Integrative Genomics Viewer (IGV): highperformance genomics data visualization and exploration. Brief Bioinform. 2013;14 (2):178-192. 16. D'Agaro T, Bittolo T, Bravin V, et al. NOTCH1 mutational status in chronic lymphocytic leukaemia: clinical relevance of subclonal mutations and mutation types. Br J Haematol. 2018;182(4):597-602. 17. Byrd JC, O'Brien S, James DF. Ibrutinib in relapsed chronic lymphocytic leukemia. N Engl J Med. 2013;369(13):1278-1279. 18. Lionetti M, Fabris S, Cutrona G, et al. Highthroughput sequencing for the identification of NOTCH1 mutations in early stage chronic lymphocytic leukaemia: biological and clinical implications. Br J Haematol. 2014;165(5):629-639. 19. Fabbri G, Rasi S, Rossi D, et al. Analysis of the chronic lymphocytic leukemia coding genome: role of NOTCH1 mutational activation. J Exp Med. 2011;208(7):1389-1401. 20. Weissmann S, Roller A, Jeromin S, et al.

haematologica | 2021; 106(9)

Prognostic impact and landscape of NOTCH1 mutations in chronic lymphocytic leukemia (CLL): a study on 852 patients. Leukemia. 2013;27(12):2393-2396. 21. Villamor N, Conde L, Martínez-Trillos A, et al. NOTCH1 mutations identify a genetic subgroup of chronic lymphocytic leukemia patients with high risk of transformation and poor outcome. Leukemia. 2013;27(5): 1100-1106. 22. Sportoletti P, Baldoni S, Del Papa B, et al. A revised NOTCH1 mutation frequency still impacts survival while the allele burden predicts early progression in chronic lymphocytic leukemia. Leukemia. 2014;28(2):436439. 23. Del Poeta G, Dal Bo M, Del Principe MI, et al. Clinical significance of c.7544-7545 delCT NOTCH1 mutation in chronic lymphocytic leukaemia. Br J Haematol. 2013;160(3):415-418. 24. Oscier DG, Rose-Zerilli MJ, Winkelmann N, et al. The clinical significance of NOTCH1 and SF3B1 mutations in the UK LRF CLL4 trial. Blood. 2013;121(3):468-475. 25. Chiorazzi N, Rai KR, Ferrarini M. Chroniclymphocyticleukemia. N Engl J Med. 2005;352(8):804-815. 26. Pozzo F, Bittolo T, Arruga F, et al. NOTCH1 mutations associate with low CD20 level in chronic lymphocytic leukemia: evidence for a NOTCH1 mutation-driven epigenetic dysregulation. Leukemia. 2016;30(1):182-189. 27. Chiaretti S, Marinelli M, Del Giudice I, et al. NOTCH1, SF3B1, BIRC3 and TP53 mutations in patients with chronic lymphocytic leukemia undergoing first-line treatment: correlation with biological parameters and response to treatment. Leuk Lymphoma. 2014;55(12):2785-2792. 28. Larrayoz M, Rose-Zerilli MJ, Kadalayil L, et al. Non-coding NOTCH1 mutations in chronic lymphocytic leukemia; their clinical impact in the UK CLL4 trial. Leukemia. 2017;31(2):510-514. 29. Kantorova B, Malcikova J, Brazdilova K, et al. Single cell analysis revealed a coexistence of NOTCH1 and TP53 mutations within the same cancer cells in chronic lymphocytic leukaemia patients. Br J Haematol. 2017;178(6):979-982. 30. Del Giudice I, Rossi D, Chiaretti S, et al. NOTCH1 mutations in +12 chronic lymphocytic leukemia (CLL) confer an unfavorable prognosis, induce a distinctive transcriptional profiling and refine the intermediate prognosis of +12 CLL. Haematologica. 2012;97(3):437-441. 31. Bulian P, Bomben R, Dal Bo M, et al. Mutational status of IGHV is the most reliable prognostic marker in trisomy 12 chronic lymphocytic leukemia. Haematologica. 2017;102(11):e443-e446. 32. López C, Delgado J, Costa D, et al. Different distribution of NOTCH1 mutations in

chronic lymphocytic leukemia with isolated trisomy 12 or associated with other chromosomal alterations. Genes Chromosomes Cancer. 2012;51(9):881-889. 33. Balatti V, Lerner S, Rizzotto L, et al.Trisomy 12 CLLs progress through NOTCH1 mutations. Leukemia. 2013;27(3):740-743. 34. Herman SE, Niemann CU, Farooqui M, et al. Ibrutinib-induced lymphocytosis in patients with chronic lymphocytic leukemia: correlative analyses from a phase II study. Leukemia. 2014;28(11):2188-2196. 35. Farooqui MZ, Valdez J, Martyr S, et al. Ibrutinib for previously untreated and relapsed or refractory chronic lymphocytic leukaemia with TP53 aberrations: a phase 2, single-arm trial. Lancet Oncol. 2015;16(2): 169-176. 36. Strati P, Parikh SA, Chaffee KG, et al. CD49d associates with nodal presentation and subsequent development of lymphadenopathy in patients with chronic lymphocytic leukaemia. Br J Haematol. 2017;178(1):99105. 37. Brown JR, Hillmen P, O'Brien S, et al. Extended follow-up and impact of high-risk prognostic factors from the phase 3 RESONATE study in patients with previously treated CLL/SLL. Leukemia. 2018;32(1):8391. 38. Dal Bo M, Del Principe MI, Pozzo F, et al. NOTCH1 mutations identify a chronic lymphocytic leukemia patient subset with worse prognosis in the setting of a rituximab-based induction and consolidation treatment. Ann Hematol. 2014;93(10):17651774. 39. Del Poeta G, Del Principe M, Postorino M, et al. Apoptosis resistance and NOTCH1 mutations impair clinical outcome in chronic lymphocytic leukemia (CLL) patients treated with ibrutinib. Blood. 2017;130 (Suppl 1):261. 40. Guièze R, Robbe P, Clifford R, et al. Presence of multiple recurrent mutations confers poor trial outcome of relapsed/refractory CLL. Blood. 2015;126(18):2110-2117. 41. Baliakas P, Hadzidimitriou A, Sutton LA, et al. Recurrent mutations refine prognosis in chronic lymphocytic leukemia. Leukemia. 2015;29(2):329-336. 42. Rossi D, Rasi S, Fabbri G, et al. Mutations of NOTCH1 are an independent predictor of survival in chronic lymphocytic leukemia. Blood. 2012;119(2):521-529. 43. López-Guerra M, Xargay-Torrent S, Rosich L, et al. The γ-secretase inhibitor PF03084014 combined with fludarabine antagonizes migration, invasion and angiogenesis in NOTCH1-mutated CLL cells. Leukemia. 2015;29(1):96-106. 44. Wu Y, Cain-Hom C, Choy L, et al. Therapeutic antibody targeting of individual Notch receptors. Nature. 2010;464(7291): 1052-1057.

2353

ARTICLE Ferrata Storti Foundation

Haematologica 2021 Volume 106(9):2354-2363

Chronic Lymphocytic Leukemia

Zanubrutinib monotherapy for patients with treatment-naïve chronic lymphocytic leukemia and 17p deletion Constantine S. Tam,1,2,3,4 Tadeusz Robak,5 Paolo Ghia,6 Brad S. Kahl,7 Patricia Walker,8 Wojciech Janowski,9 David Simpson,10,11 Mazyar Shadman,12,13 Peter S. Ganly,14 Luca Laurenti,15 Stephen Opat,16,17 Monica Tani,18 Hanna Ciepluch,19 Emma Verner,20,21 Martin Šimkovič,22,23 Anders Österborg,24,25 Marek Trněný,26 Alessandra Tedeschi,27 Jason C. Paik,11 Sowmya B. Kuwahara,11 Shibao Feng,11 Vanitha Ramakrishnan,11 Aileen Cohen,11 Jane Huang,11 Peter Hillmen28 and Jennifer R. Brown29 1 Peter MacCallum Cancer Center, Melbourne, Victoria, Australia; 2University of Melbourne, Parkville, Victoria, Australia; 3Royal Melbourne Hospital, Parkville, Victoria, Australia; 4St Vincent’s Hospital Melbourne, Fitzroy, Victoria, Australia; 5Medical University of Lodz, Lodz, Poland; 6Università Vita-Salute San Raffaele and IRCCS Ospedale San Raffaele, Milano, Italy; 7Washington University School of Medicine, St Louis, MO, USA; 8Peninsula Private Hospital, Frankston, Victoria, Australia; 9Calvary Mater Newcastle, Waratah, New South Wales, Australia; 10North Shore Hospital, Auckland, New Zealand; 11BeiGene USA, Inc., San Mateo, CA, USA; 12Fred Hutchinson Cancer Research Center, Seattle, WA, USA; 13Department of Medicine, University of Washington, Seattle, WA, USA; 14Department of Hematology, Christchurch Hospital, Christchurch, New Zealand; 15Fondazione Policlinico Universitario A. Gemelli IRCCS, Rome, Italy; 16Monash Health, Clayton, Victoria, Australia; 17Monash University, Clayton, Victoria, Australia; 18Hematology Unit, Santa Maria delle Croci Hospital, Ravenna, Italy; 19 Copernicus Wojewódzkie Centrum Onkologii, Gdánsk, Poland; 20Concord Repatriation General Hospital, Concord, New South Wales, Australia; 21University of Sydney, Concord, New South Wales, Australia; 22Fourth Department of Internal Medicine - Hematology, University Hospital, Hradec Kralove, Czech Republic; 23Faculty of Medicine, Charles University, Prague, Czech Republic; 24Department of Oncology-Pathology, Karolinska Institutet, Stockholm, Sweden; 25Department of Hematology, Karolinska University Hospital, Stockholm, Sweden; 26First Department of Medicine, First Faculty of Medicine, Charles University, General Hospital, Prague, Czech Republic; 27ASST Grande Ospedale Metropolitano Niguarda, Milan, Italy; 28St James’s University Hospital, Leeds, UK and 29 Dana-Farber Cancer Institute, Boston, MA, USA

ABSTRACT

Correspondence: CONSTANTINE S. TAM constantine.tam@petermac.org Received: May 29, 2020. Accepted: September 17, 2020. Pre-published: October 13, 2020. https://doi.org/10.3324/haematol.2020.259432

2354

atients with chronic lymphocytic leukemia or small lymphocytic lymphoma whose tumors carry deletion of chromosome 17p13.1 [del(17p)] have an unfavorable prognosis and respond poorly to standard chemoimmunotherapy. Zanubrutinib is a selective next-generation Bruton tyrosine kinase inhibitor. We evaluated the safety and efficacy of zanubrutinib 160 mg twice daily in treatment-naïve patients with del(17p) disease enrolled in a dedicated, nonrandomized cohort (Arm C) of the phase III SEQUOIA trial. A total of 109 patients (median age, 70 years; range, 42–86) with centrally confirmed del(17p) were enrolled and treated. After a median of 18.2 months (range, 5.0–26.3), seven patients had discontinued study treatment due to progressive disease, four due to an adverse event, and one due to withdrawal of consent. The overall response rate was 94.5% with 3.7% of patients achieving complete response with or without incomplete hematologic recovery. The estimated 18-month progression-free survival rate was 88.6% (95% CI: 79.0–94.0) and the estimated 18-month overall survival rate was 95.1% (95% CI: 88.4–98.0). Most common all-grade adverse events included contusion (20.2%), upper respiratory tract infection (19.3%), neutropenia/neutrophil count decreased (17.4%), and diarrhea (16.5%). Grade ≥ 3 adverse events were reported in 53 patients (48.6%), most commonly neutropenia (12.9%) and pneumonia (3.7%). An adverse event of atrial fibrillation was reported in three patients (2.8%). Zanubrutinib was active and well tolerated in this large, prospectively enrolled treatment cohort of previously untreated patients with del(17p) chronic lymphocytic leukemia/small lymphocytic lymphoma. This trial was registered as clinicaltrials.gov Identifier: NCT03336333.

haematologica | 2021; 106(9)

Zanubrutinib in treatment naïve del(17p) CLL/SLL

Introduction Patients with chronic lymphocytic leukemia (CLL) or small lymphocytic lymphoma (SLL) have historically been treated with combination chemotherapy and immunotherapy with success; however, many patients do not have sustained responses in part due to genomic aberrations that impair responsiveness to therapy. One aberration found in patients with CLL/SLL is the deletion of chromosome 17p13.1 [del(17p)]; these patients have an unfavorable prognosis and respond poorly to standard chemoimmunotherapy, with reduced rates of overall survival (OS) and worse clinical outcomes.1,2 del(17p) results in the mono or biallelic loss of the TP53 gene, which encodes the tumor suppressor p53, a multifunctional transcription factor important for cellular response to DNA damage, including cell cycle arrest and apoptosis.3 Most patients with del(17p) also have a mutation of the other TP53 allele and therefore lack wild-type TP53, leading to genomic instability and reduced responsiveness to cytotoxic chemotherapy.2 The incidence of del(17p) is approximately 5-8% of patients with CLL at diagnosis and increases with each relapse.4,5 Novel targeted therapies are therefore the preferred treatment modality for previously untreated patients whose disease bears the del(17p) mutation.6,7 Several new agents are approved by the US Food and Drug Administration and European Medicines Agency (EMA) for adult patients with CLL/SLL regardless of del(17p) status, including the BTK inhibitors ibrutinib and acalabrutinib. In the RESONATE study of ibrutinib versus ofatumumab in patients with relapsed or refractory (R/R) CLL/SLL,8 89% of ibrutinib-treated patients with del(17p) achieved an objective response with long term follow-up.9 Similar results were observed in the single-arm RESONATE-17 study in R/R CLL10 in which 83% of patients achieved an objective response; the 24-month progression-free survival (PFS) was 63%. Acalabrutinib, a secondgeneration BTK inhibitor, has been approved recently for treatment-naïve (TN) patients with CLL/SLL regardless of del(17p) status based on results from the ELEVATE TN study.11,12 Notably, of the patients with del(17p) in that study, only 16 were assigned to the single-agent acalabrutinib arm. Other small molecule inhibitors approved in patients with CLL/SLL and del(17p) include the BCL-2 inhibitor venetoclax13 and the phosphatidylinositide-3-kinase (PI3K)-δ inhibitor idelalisib.14 Stilgenbauer and colleagues reported the results of venetoclax treatment in 158 mostly R/R del(17p) patients,15 including a 77% overall response rate (ORR) and an estimated 24-month PFS of 50%. Notably, only five patients in this study were TN. In the CLL14 trial which compared venetoclax and obinutuzumab to chlorambucil and obinutuzumab in TN CLL, 17 patients assigned to the venetoclax arm had del(17p).16 Similarly, studies supporting the approval of idelalisib together with rituximab in the frontline setting by the EMA included few TN CLL/SLL patients with del(17p).17,18 Collectively, only limited data are available for novel targeted therapies for previously untreated patients with del(17p) CLL/SLL, and no large multi-center studies have systematically examined this specific population. Zanubrutinib (BGB-3111) is a next-generation BTK inhibitor with favorable oral bioavailability and high specificity for BTK, exhibiting comparatively lower offtarget activity than ibrutinib for structurally related kinashaematologica | 2021; 106(9)

es such as epidermal growth factor receptor (EGFR), interleukin-2 inducible kinase (ITK), and Src family kinases.19 The safety, pharmacokinetics, pharmacodynamics, and preliminary activity of zanubrutinib were investigated in a phase I/II study of patients with multiple B-cell malignancies in which high level, sustained BTK occupancy was noted in both peripheral blood and lymph nodes at the recommended phase II dose of 160 mg twice daily (bid).20 Encouraging activity was observed in a cohort of 78 patients with both TN and R/R CLL/SLL, including in a subset of 16 patients with del(17p) or TP53 mutation who achieved a 100% ORR.20 Activity of zanubrutinib was also observed in a separate phase II trial of 91 R/R CLL/SLL patients in China, including in a subset of 17 patients with del(17p) who achieved a 88.2% ORR.21 Zanubrutinib has recently received accelerated approval in the United States for adult patients with mantle cell lymphoma who have received at least one prior therapy22 and is currently undergoing further clinical testing in several prospective, multicenter, randomized phase III trials in CLL/SLL. The SEQUOIA trial (clinicaltrialsgov. Identifier: NCT03336333) is an open-label, multi center, randomized phase III study of TN patients with CLL/SLL. Patients without centrally confirmed del(17p) were randomized to receive either zanubrutinib 160 mg bid until unacceptable toxicity or disease progression (PD), or six cycles of rituximab and bendamustine. Considering the poor outcomes associated with any standard chemoimmunotherapy regimen in patients with del(17p), those with centrally confirmed del(17p) during screening for SEQUOIA were not randomized but assigned to single-agent zanubrutinib in a separate cohort (Arm C). This is the first report of the safety and efficacy results in this high-risk del(17p) patient cohort.

Methods Study design and population Eligible patients had confirmed CLL/SLL requiring treatment per International Workshop on Chronic Lymphocytic Leukemia (iwCLL) definition.23 TN adults were eligible if either aged ≥65 years or unsuitable for treatment with fludarabine, cyclophosphamide, and rituximab (FCR) and an Eastern Cooperative Oncology Group (ECOG) performance status ≤2.24 Centrally confirmed del(17p) by fluorescence in situ hybridization with >7% aberrant nuclei present was required. Patients had adequate endorgan function, including absolute neutrophil count (ANC) ≥1,000/mm3 and platelet count ≥75,000/mm3. For patients with bone marrow involvement, ANC ≥750/mm3 and platelet count ≥50,000/mm3 were allowed. Patients with history of atrial fibrillation and/or long-term anticoagulation, or those requiring moderate or strong CYP3A inhibitors, could enroll.25 All Arm C patients were assigned to receive zanubrutinib (160 mg bid) until intolerance or PD. This study was conducted according to principles of the Declaration of Helsinki and the International Conference on Harmonization guidelines for Good Clinical Practice and approved by the Institutional Review Board/Ethics Committee at each participating site. All patients provided written informed consent. This study adhered to CONSORT-10 guidelines for reporting.26

Objectives and assessments Key objectives for Arm C included assessments of ORR, PFS,

2355

C.S. Tam et al.

duration of response (DOR), and safety (frequency and severity of all treatment-emergent adverse events [AE]). ORR was assessed per modified iwCLL criteria for CLL23,27 and per Lugano criteria for SLL.28 Measurable disease, defined as ≥1 lymph node >1.5 cm in the longest diameter and measurable in two perpendicular diameters, was assessed by computed tomography/magnetic resonance imaging. Response assessments were performed every 12 weeks after the first dose day for 96 weeks, then every 24 weeks until PD or initiation of new, nonprotocol therapy, whichever came first. Patients underwent bone marrow examination at baseline and for confirmation of complete response (CR), or CR with incomplete hematologic recovery (CRi), or if PD was suspected due to cytopenia. All treatment-emergent AE, including AE of interest (AEI) based on the known toxicity profile for BTK inhibitors occurring on or after day 1 until 30 days after treatment discontinuation were summarized. AEI were categorized in accordance with predefined MedDRA search criteria (Online Supplementary Table S1). AE severity was assessed using the National Cancer Institute Common Terminology Criteria for Adverse Events v4.03 and the Grading Scale for Hematologic Toxicity in CLL Studies.23 Biomarkers were assessed at baseline and optionally at progression.

Statistical analyses Primary efficacy and safety analyses included all patients with centrally confirmed del(17p) CLL/SLL receiving ≥1 dose of zanubrutinib. ORR was summarized as percentage of responders (CR, CRi, nodular partial response [nPR], partial response [PR], or PR with lymphocytosis [PR-L]) with corresponding 95% Confidence Interval (CI). An evaluation of ORR in subgroups defined by key demographic and baseline disease characteristics was conducted and summarized in a forest plot. DOR was defined as time from first response until PD or death due to any cause. PFS was measured from time of first dose to PD or death due to any cause. Median DOR, PFS, and event-free survival rates were estimated using Kaplan-Meier methodology with corresponding 95% CI.

Results Patient characteristics and disposition Between February 3, 2018 and February 20, 2019, 109 patients with centrally confirmed del(17p) CLL/SLL were enrolled from 59 sites in 13 countries (Online Supplementary Table S2) and received ≥1 dose of zanubrutinib. Two additional patients without del(17p) were assigned in error to this study arm and are not included in the analysis. At the data cutoff date of April 15, 2020, the median duration of study follow-up was 18.2 months (range, 5.0–26.3). Median age at study entry was 70.0 years (range, 42–86); the majority of patients presented with CLL (90.8%). Reported reasons for treatment per iwCLL criteria in order of frequency included progressive marrow failure (41.3%); massive, progressive, or symptomatic lymphadenopathy (41.3%); significant fatigue (33.0%); night sweats (32.1%); progressive lymphocytosis with increase of > 50% over 2 months or doubling time of <6 months (28.4%); massive, progressive, or symptomatic splenomegaly (23.9%); and unintentional weight loss (14.7%). More than one reason for treatment may have been given, with a median of two reasons given for each patient. Many patients had other high-risk disease characteristics, including 40 patients with CLL who presented as Binet stage C (40.4%), 42 patients with bulky disease ≥5 2356

cm (38.5%), 78 patients with elevated β2-microglobulin (78.8% of 99 patients with available data), and 67 patients with an unmutated immunoglobulin heavy chain variable (IGHV) locus (65.0% of 103 patients with sufficient RNA for testing). Furthermore, 32 (37.2%) of 86 patients with sufficient metaphases available for analysis had at least three distinct karyotypic abnormalities defined as complex karyotype (Table 1).

Efficacy The ORR was 94.5%, which included three patients (2.8%) with CR, one patient (0.9%) with CRi, 95 (87.2%) with PR, and four (3.7%) with PR-L (Table 2). Five (4.6%) patients had a best response of stable disease (SD). One patient (0.9%) had PD at the first response assessment. Five (4.6%) patients met clinical CR criteria but did not undergo bone marrow biopsy. Ninety-seven patients (89.0%) remained on treatment at the time of analysis. Nine patients with an initial response (8.3%) progressed on study, four of whom had histologically confirmed Richter transformation. Median time to transformation was 13.7 months (time to transformation for each patient: 3.9, 13.6, 13.8, and 15.7 months). Two other patients had new lesions seen on CT with positron emission tomography (PET) avidity for which biopsy could not definitively confirm transformation, while one other patient had accelerated CLL. Seven patients who have progressed have discontinued treatment; two other patients with progression remained on treatment at time of data cutoff. Four patients who progressed have died; two due to progression, one due to an adverse event after progression (acute kidney injury), and one after progression due to septic shock. Four patients (3.7%) discontinued treatment due to AEs, of whom 2 have died (see Safety below), while one patient discontinued treatment after withdrawal of consent and was lost to follow up. Median PFS and OS were not reached. The estimated 18-month PFS rate was 88.6% (95% CI: 79.0–94.0) (Figure 1A), while the estimated 18-month OS rate was 95.1% (95% CI: 88.4–98.0) (Figure 1B). The median time to response was 2.8 months (range, 1.9–16.5) The median DOR was not reached; 92.8% of patients had a DOR ≥ 12 months (Figure 1C). ORR was consistent across all prespecified demographic and baseline disease characteristics (Figure 2). Transient treatment-related lymphocytosis was observed (Online Supplementary Figure S1A), but there was a significant reduction in target lesion size by the first scheduled response assessment, consistent with the short median time to response (Online Supplementary Figure S1A). The peak median change in absolute lymphocyte count (ALC) and time to resolution of lymphocytosis both appeared to be decreased from previous experience with zanubrutinib.19 In an exploratory analysis, changes in ALC were compared between patients with mutated and unmutated IGHV locus. Patients with unmutated IGHV showed a slight trend towards less treatment-related lymphocytosis (Online Supplementary Figure S1B), similar to previous reports.9,29 Sixty-one patients (56.0%) began the study with at least one cytopenia (Table 1), including 43 patients with anemia (39.4%), eight patients with neutropenia (7.3%), and 28 patients with thrombocytopenia (25.7%) (Online Supplementary Table S3). Eleven patients (10.1%) received at least one dose of a granulocyte colony-stimulating fachaematologica | 2021; 106(9)

Zanubrutinib in treatment naïve del(17p) CLL/SLL

tor, while one patient (0.9%) received at least one dose of an erythrocyte-stimulating growth factor. Of those patients with baseline anemia, 86.0% of patients demonstrated sustained improvement in hemoglobin; 75.0% of patients with baseline neutropenia also demonstrated sustained improvement in ANC, and 85.7% of patients

with baseline thrombocytopenia demonstrated sustained improvement in platelet count (Online Supplementary Table S3). In an exploratory post hoc analysis, baseline characteristics and response rate were compared between patients with a percentage of del(17p)-positive nuclei of ≥20%

Figure 1. Survival and response analyses using the Kaplan-Meier method. (A) Progression-free survival as determined by investigator assessment. Shaded area indicates 95% Confidence Interval (CI). (B) Overall survival. Shaded area indicates 95% CI. (C) Duration of response as determined by investigator assessment. Shaded area indicates 95% CI.

haematologica | 2021; 106(9)

2357

C.S. Tam et al.

Table 1. Key patient and disease characteristics.

Table 2. Summary of investigator-assessed efficacy.

TN del(17p) CLL/SLL (n=109) Follow-up, median (range), mo Demographics Age, median (range), years Male sex, n (%) ECOG PS of 2, n (%) Months since diagnosis, median (Q1 - Q3) Disease characteristics SLL, n (%) Binet stage C for patients with CLL, n (%) ALC (x 109/L), median Hemoglobin (g/L), median Platelet count (x 109/L), median β2-microglobulina > 3.5 g/dL, n (%) IGHV mutational status,b n (%) Mutated Unmutated Bulky disease,c n (%) Any TL LDi ≥ 5 cm Any TL LDi ≥ 10 cm Cytopenia present,d n (%) Proportion of cells with del(17p), n (%) 7.5% – 10% 10.5% – 20.0% 20.5% – 50.0% 50.5% – 100% Mean % (SD) Karyotype status,e n (%) Non-Complex (0 to 2 abnormalities) Complex 3 or more abnormalities 5 or more abnormalities

18.2 (5.0-26.3) 70.0 (42-86) 78 (71.6) 14 (12.8) 21.6 (7.7-54.8) 10 (9.2) 40/99 (40.4) 65.1 120.0 154.0 78/99 (78.8) 36/103 (35.0) 67/103 (65.0) 42 (38.5) 11 (10.1) 61 (56.0) 16 (14.7) 44 (40.4) 13 (11.9) 36 (33.0) 36.0 (31.6) 54/86 (62.8) 32/86 (37.2) 23/86 (26.7)

AE: adverse event; ALC: absolute lymphocyte count; CLL: chronic lymphocytic leukemia; ECOG PS: Eastern Cooperative Oncology Group performance status; IGHV: immunoglobulin heavy chain variable; LDi: longest diameter; SD: standard deviation; SLL: small lymphocytic lymphoma; TL: target lesion; TN: treatment-naïve; mo: months. a Ten patients had missing data. bSix patients had RNA quantity/quality not sufficient for polymerase chain reaction (PCR) amplification of heavy-chain variable (VH) region for sequencing. cPatients with any target lesion with longest diameter presented. dPatients having anemia (≤110 g/L), thrombocytopenia (≤ 100x109/L), or neutropenia (≤ 1.5 x 109/L). e23 patients had insufficient metaphases available for analysis.

[del(17p) high] versus patients with a percentage of >7% to <20% [del(17p) low] (Online Supplementary Table S4). Patients in the del(17p) high category were observed to have a higher rate of unmutated IGHV (75% vs. 56.4% of patients with a resulted test; P=0.0478) and complex karyotype status (56.8% vs. 22.4% of patients with sufficient metaphases for analysis; P=0.0011); no other differences in baseline characteristics were observed. Best ORR and estimated 18-month PFS were 98% and 89%, respectively, in the del(17p) high category and 92% and 88%, respectively, in the del(17p) low category.

Safety AE of any grade reported in ≥10% of patients included contusion (20.2%), upper respiratory tract infection (19.3%), neutropenia/neutrophil count decreased (17.4%), diarrhea (16.5%), nausea (14.7%), rash (13.8%), constipation (13.8%), back pain (12.8%), cough (11.9%), arthralgia (11.0%), and fatigue (10.1%) (Table 3). Grade ≥3 events were reported in 53 patients (48.6%), with neutropenia/decreased neutrophil count (12.9%) and pneumonia (3.7%) being the most common. Serious AE 2358

Efficacy Variable

TN del(17p) CLL/SLL (n = 109)

Best response, n (%) ORR (CR, PR, or PR-L), n (%) [95% CI]a CR CRi PR PR-L SD PD Time to response, mo PR-L or higher, median (range) PR or higher, median (range) DOR Response ≥ 12 mo, % [95% CI]a Estimated PFS rate, % 12 months [95% CI]a 18 months [95% CI]a

103 (94.5) [88-98] 3 (2.8) 1 (0.9) 95 (87.2) 4 (3.7) 5 (4.6) 1 (0.9) 2.79 (1.9-16.5) 2.89 (1.9-16.5) 92.8 [85.4-96.5] 94.5 [88.2-97.5] 88.6 [79.0-94.0]

CI: Confidence Interval; CLL: chronic lymphocytic leukemia; CR: complete response; DOR: duration of response; ORR: overall response rate; PD: progressive disease; PFS: progression-free survival; PR: partial response; PR-L: PR with lymphocytosis; SD: stable disease; SLL: small lymphocytic lymphoma; TN: treatment-naïve. aTwo-sided ClopperPearson 95% CI.

(SAE) were reported in 36.7% of patients, with pneumonia (3.7%) being the most common. AE led to dose reduction in 6 (5.5%) patients and included thrombocytopenia, diarrhea, gastritis, blood bilirubin increased, arthralgia, myalgia, and headache. All six patients remain on treatment. Four patients discontinued treatment due to AE; a 77year-old male patient had grade 4 pseudomonal sepsis associated with grade 4 neutropenia and atrial fibrillation who recovered, while another 72-year-old male patient had grade 3 melanoma requiring surgery and adjuvant therapy. One grade 5 event was reported in a 84-year-old female patient with health-care associated pneumonia diagnosed 8 days after the last dose of zanubrutinib which was held for an unrelated procedure. This case was complicated by the development of sepsis which was assessed as related to zanubrutinib. Finally, one grade 5 event was reported in a 72-year-old male patient who developed disease progression at the week 36 response assessment including massive enlargement of intra-abdominal lymph nodes associated with hypercalcemia. Prior to discontinuation of study drug, the patient shortly thereafter had renal failure requiring dialysis and subsequently died due to pulmonary edema. No sudden or unknown deaths were reported. AEI known to be associated with BTK inhibitors were characterized in greater detail, including grouping of similar AE by category (Online Supplementary Table S1). AEI reported in ≥10% of treated patients included infections (64.2%; 13.8% grade ≥3), minor bleeding (26.6%), bruising (24.8%; 0% grade ≥3), neutropenia (18.3%; 13.8% grade ≥3), diarrhea (15.6%; 0.9% grade ≥3), nausea (13.8%; 0% grade ≥3), arthralgia (11.0%; 0% grade ≥3), fatigue (10.1%; 0.9% grade ≥3) (Online Supplementary Table S5). The most common infections reported in ≥ 5% of patients included upper respiratory tract infection (19.3%), pneumonia (8.3%), nasopharyngitis (7.3%), and urinary tract infection (6.4%); most of these were grade 1 or 2 events. Prophylaxis against opportunistic infections was allowed per local standard of care but not required; haematologica | 2021; 106(9)

Zanubrutinib in treatment naïve del(17p) CLL/SLL

Figure 2. Subgroup analysis of overall response rate. Overall response rate presented as of November 1, 2019. Two-sided Clopper-Pearson 95% Confidence Interval (CI) are used. CLL: chronic lymphocytic lymphoma; ECOG PS: Eastern Cooperative Oncology Group performance status; LDH: lactate dehydrogenase; LDi: longest diameter; SLL: small lymphocytic leukemia. aPatients with any target lesion with longest diameter presented. bSix patients had RNA quantity/quality not sufficient for polymerase chain reaction amplification of immunoglobulin heavy chain variable (VH) region for sequencing. cPatients having anemia (≤110 g/L), thrombocytopenia (≤100x109/L), or neutropenia (≤1.5x109/L). d10 patients had missing data. e23 patients had insufficient metaphases available for analysis.

no opportunistic infections were reported. Of other malignancies reported on study, most were dermatological malignancies reported in ten patients (9.2%). Other than two patients with melanoma, all were basal and squamous cell carcinomas of the skin (grade 1/2) reported from patients in Australia and New Zealand, where skin cancers are frequent, especially in patients with CLL/SLL.30,31 One patient developed grade 3 melanoma requiring surgery and adjuvant therapy with pembrolizumab leading to discontinuation of zanubrutinib. Five patients (4.6%) reported a non-dermatologic other malignancy. One patient developed localized breast cancer for which axillary lymph node biopsy indicated disease transformation to DLBCL, while one other patient developed lung cancer for which interlobar lymph node biopsy also indicated disease transformation to DLBCL. Three patients reported a transitional cell carcinoma of the bladder or ureter, all of whom underwent resection withhaematologica | 2021; 106(9)

out known residual disease and remain on study drug treatment. The usage of therapeutic anticoagulation was not restricted on this study; 20 patients (18.3%) were reported to have taken a therapeutic anticoagulant including warfarin, direct-acting oral anticoagulants, and heparins during the study, while 27 patients (24.8%) were reported to have taken an oral platelet aggregation inhibitor, including aspirin, during the study. Bleeding of any type was reported in 47.7% of patients (4.6% grade ≥3). Major bleeding events were defined as any bleeding event with grade ≥3, any SAE, or any bleed affecting the central nervous system (CNS); these occurred in six patients (5.4%). A description of each event, including any confounding factors, is presented in Online Supplementary Table S6; in two patients, bleeding occurred in the setting of a surgical procedure without dose hold as advised per protocol. All patients continued study treatment after dose interruption. No 2359

C.S. Tam et al. Table 3. Most common adverse events regardless of causality. Adverse events of any grade occurring in ≥ 5% of patients and all grade ≥3 adverse events occurring in ≥2% of patients are shown.a

Term

Any Grade

Grade 1/2

Grade 3 n (%)

Grade 4

Grade 5

Patients with at least one AE Hematologic AE Neutropenia Neutrophil count decreased Nonhematologic AE Contusion Upper respiratory tract infection Diarrhea Nausea Constipation Rash Back pain Cough Arthralgia Fatigue Dyspepsia Headache Fall Pain in extremity Pneumonia Abdominal pain Dyspnea Epistaxis Hematuria Nasopharyngitis Pruritus Pyrexia Hypertension Muscle spasms Urinary tract infection Vomiting Hematoma Musculoskeletal pain Skin laceration

106 (97.2)

53 (48.6)

44 (40.4)

7 (6.4)

2 (1.8)

13 (11.9) 6 (5.5)

3 (2.8) 2 (1.8)

7 (6.4) 1 (0.9)

3 (2.8) 3 (2.8)

0 (0) 0 (0)

22 (20.2) 21 (19.3) 18 (16.5) 16 (14.7) 15 (13.8) 15 (13.8) 14 (12.8) 13 (11.9) 12 (11.0) 11 (10.1) 10 (9.2) 9 (8.3) 9 (8.3) 9 (8.3) 9 (8.3) 8 (7.3) 8 (7.3) 8 (7.3) 8 (7.3) 8 (7.3) 8 (7.3) 8 (7.3) 7 (6.4) 7 (6.4) 7 (6.4) 7 (6.4) 6 (5.5) 6 (5.5) 6 (5.5)

22 (20.1) 21 (19.3) 17 (15.6) 16 (14.7) 15 (13.8) 15 (13.8) 13 (11.9) 13 (11.9) 12 (11.0) 10 (9.2) 10 (9.2) 8 (7.3) 7 (6.4) 8 (7.3) 5 (4.6) 8 (7.3) 8 (7.3) 7 (6.4) 6 (5.5) 8 (7.3) 8 (7.3) 7 (6.4) 5 (4.6) 7 (6.4) 5 (4.6) 7 (6.4) 6 (5.5) 5 (4.6) 6 (5.5)

0 (0) 0 (0) 1 (0.9) 0 (0) 0 (0) 0 (0) 1 (0.9) 0 (0) 0 (0) 1 (0.9) 0 (0) 1 (0.9) 2 (1.8) 1 (0.9) 3 (2.8) 0 (0) 0 (0) 1 (0.9) 2 (1.8) 0 (0) 0 (0) 1 (0.9) 2 (1.8) 0 (0) 2 (1.8) 0 (0) 0 (0) 1 (0.9) 0 (0)

0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)

0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 1 (0.9) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0) 0 (0)

AE: adverse event. aData are for treatment-emergent adverse events in the 109 patients in this arm of the study.

bleeding events affecting the CNS were reported. A history of atrial fibrillation or flutter was reported in seven patients (6.4%); four patients (3.7%) entered the study with controlled and hemodynamically stable atrial fibrillation or flutter. An AE of atrial fibrillation or flutter was reported in three patients (2.8%). One grade 2 event was reported in a 71-year-old male patient with hypertension and prior history of atrial fibrillation, while a grade 3 event was reported in a 78-year-old male patient who was septic from cholecystitis. Both of these events resolved and did not require discontinuation of study treatment. Finally, one grade 4 event was reported in a 77-year-old male patient with grade 3 hypertension at baseline, who discontinued treatment due to sepsis secondary to Pseudomonas; atrial fibrillation resolved following recovery from sepsis.

Discussion In this report, we have shown the activity and safety of zanubrutinib in a large non-randomized cohort of treatment-naïve CLL/SLL patients with centrally confirmed del(17p), enrolled as part of the global SEQUOIA trial. As expected, these results compare favorably with those 2360

from previous studies of TN patients treated with chemoimmunotherapy, including the CLL8 trial of FCR.1,2,32 At the present time, prospective clinical trial data from BTK inhibitor-treated patients with TN del(17p) CLL/SLL are limited. Patients with del(17p) were not eligible for enrollment in RESONATE-233 and the ECOGE191234 trials evaluating ibrutinib in the TN setting. The Alliance A041702 study, comparing chemotherapy, ibrutinib, or ibrutinib and rituximab in older patients with TN CLL/SLL, did allow patients with del(17p) to enroll; of these, 6% had del(17p), and only nine patients were assigned to receive ibrutinib alone.35 Two studies examining combinations of novel targeted therapies with antiCD20 antibodies also enrolled a small number of TN patients with del(17p), including the iLLUMINATE study, where 14 patients with del(17p) received combination ibrutinib and obinutuzumab.36 In a single-center, phase 2 study, single-agent ibrutinib was evaluated in 35 patients with TN CLL; notably, this population was selected based on cytogenetics or TP53 sequencing and allowed younger and/or fit patients, leading to enrollment of a population with a median age of 62.37,38 At a median follow-up time of 15 months, ORR was reported as 97%, including 12% CR, 70% PR, and 15% PR-L. The ORR with zanubrutinib observed in the present haematologica | 2021; 106(9)

Zanubrutinib in treatment naïve del(17p) CLL/SLL

study appears at least comparable with the reported ibrutinib experience. Median time to response essentially was defined by the first scheduled response assessment (Figure 1C), consistent with reduction in target lesion size (Online Supplementary Figure S1A) and resolution of cytopenias (Online Supplementary Table S3). Consistent with previous studies of other BTK inhibitors,12,32,37 CR were uncommon with short follow-up; longer follow-up will be needed to more precisely define the CR rate. Response rates appear to be similarly high regardless of coincident risk factors, including IGHV mutational status and complex karyotype status, though the low number of non-responders may limit the ability to detect meaningful differences between subgroups. In the UK LRF CLL4 trial, del(17p) in >20% of nuclei was found to be independently associated with shorter PFS in patients treated with chemoimmunotherapy.39,40 As expected, patients with a higher burden of del(17p) were associated with a higher rate of poor prognostic factors such as unmutated IGHV status and complex karyotype. When examining ORR and progression events in the current study, ORR appeared to be similar in patients without and with del(17p) in ≥20% of nuclei (Online Supplementary Table S4), suggesting that zanubrutinib has preserved activity in high-risk patients with enrichment of malignant cells for del(17p). This is comparable to the activity seen in the RESONATE-17 trial of R/R patients with del(17p) treated with ibrutinib.10 Ten patients progressed on study. In addition to presence of del(17p), seven patients who progressed had an unmutated IGHV locus. Karyotype analysis was available for eight patients who had progressed, of whom two had complex karyotype (number of abnormalities: 5 and 6). Four patients had histologically confirmed transformation to aggressive lymphoma, while two patients had suspected transformation. The present results compare favorably to those reported with chemoimmunotherapy, where 23% of patients with del(17p) treated in the first line experienced disease transformation, with a median time to transformation of 12 months.41 In the RESONATE-17 trial of R/R CLL patients, 44% of progression events were due to transformation, most within the first 6 months of treatment.10 Similarly, both early progression events in the TN del(17p) or TP53 populations treated with ibrutinib, as reported by Farooqi and colleagues, occurred due to Richter transformation.37 These data are in line with the known association of del(17p) and transformation to aggressive lymphomas.42 Long term outcomes for patients with del(17p) treated with ibrutinib were reported by Ahn and colleagues,38 showing an approximate 5-year PFS of 75%. Further follow-up will be required to demonstrate the durability of responses to zanubrutinib. Additional analyses, including correlation of response and progression with concurrent genomic abnormalities and other genetic mutations (e.g., TP53, NOTCH1, BTK, and PLCG2 mutations), both at baseline and at the time of progression, are currently in progress. The clinically meaningful activity noted in this patient series appears to be associated with a favorable toxicity profile and is consistent with that reported in other studies of zanubrutinib to date.20,43 Despite enrolling a more elderly and comorbid population and allowing for therapeutic anticoagulation on study, the incidence of grade ≥3 AE or SAE leading to major bleeding was 5.6% with no CNS events reported, and all patients able to continue study haematologica | 2021; 106(9)

drug after dose interruption. Consistent with its greater selectivity for BTK and less inhibition of kinases such as EGFR, Src, and others, the incidence of grade 3 AE such as diarrhea, arthralgia, and myalgia were all ≤1%. Importantly, only three patients on this study reported treatment-emergent atrial fibrillation, six patients required ongoing dose reduction, and four patients discontinued zanubrutinib due to an AE. Two phase III randomized studies in patients with R/R CLL/SLL44 and Waldenström macroglobulinemia45 are ongoing to directly compare the efficacy and safety profiles of ibrutinib and zanubrutinib. Limitations of this study include the relatively short duration of follow-up and its single-arm design. A retrospective analysis for baseline TP53 mutations is currently being performed for this study arm and the larger randomized arms. Analogous to other BTK inhibitors, the singleagent activity of zanubrutinib is not expected to induce a deep response with eradication of minimal residual disease (MRD). Several studies have recently reported achievement of undetectable MRD in patients with TN CLL/SLL, including those patients with del(17p), by combining a BTK inhibitor with obinutuzumab or venetoclax.36,46,47 The SEQUOIA trial is currently enrolling patients in Arm D, which will evaluate the safety and activity of a combination of zanubrutinib with venetoclax in TN CLL/SLL patients with del(17p). In summary, these results indicate that single-agent zanubrutinib is active and generally well tolerated in this very high-risk population. Disclosures CST received research funding from Janssen, AbbVie, BeiGene, Pharmacyclics, TG Therapeutics and AbbVie and served as a consultant for BeiGene, Janssen, Roche, AbbVie and Loxo; TR served as a consultant for and received honoraria and research funding from Gilead; PG served as a consultant for Adaptive, AbbVie, ArQule, BeiGene, Celgene/Juno, Dynamo, Gilead, Janssen, Sunesis and received research funding from AbbVie, Gilead, Janssen, Novartis, and Sunesis; LL received honoraria from Roche and AbbVie, and served as a consultant for BeiGene, Roche, AbbVie, AstraZeneca, Janssen, and Gilead; BSK served as a consultant for AbbVie, Acerta, AstraZeneca, BeiGene, Janssen and Pharmacyclics; PW is an employee of Alfred Health (public hospital) and Peninsula Health (public hospital) and received travel funding from Roche; WJ served as a consultant for AstraZeneca and served on advisory boards for Janssen, Celgene, and Amgen; DS is an employee of and has equity ownership in BeiGene, received honoraria from AbbVie, Janssen, and Roche, received research funding from AbbVie, Amgen, Celgene, Roche, MSD, Acerta, Pharmacyclics, Sanofi, and GSK; MS served as a consultant for AbbVie, Genentech, AstraZeneca, Pharmacyclics, Verastem, ADC Therapeutics, Atara Biotherapeutics, Cellectar, and Bristol Myers Squibb, and received research funding from Mustang Bio, Celgene, Pharmacyclics, Gilead and Genentech, TG Therapeutics, BeiGene, AstraZeneca, Sunesis, Acerta Pharma, Atara Biotherapeutics, and Bristol Myers Squibb; SO received honoraria from and served as a consultant for AbbVie, Roche, AstraZeneca, Merck, Gilead, Janssen, and Novartis, received research funding from BeiGene, Roche, AstraZeneca, Janssen, Merck, Amgen, and Epizyme, and received travel funding from Roche; MT has nothing to disclose; HC is an employee of Copernicus Wojewódzkie Centrum Onkologii Gdańsk; EV received research funding from Janssen; MŠ served as a consultant for and received travel expenses from AbbVie, Gilead, Janssen-Cilag, and Acerta, and served on an advisory board for 2361

C.S. Tam et al.

AbbVie; AÖ received research funding from BeiGene; MT is an employee of the Charles University General Hospital, served as a consultant for Takeda, Bristol Myers Squibb, Incyte, AbbVie, Amgen, Roche, Gilead Sciences, Janssen, Celgene, MorphoSys, received honoraria from Janssen, Gilead Sciences, Takeda, Bristol Myers Squibb, Amgen, AbbVie, Roche, MorphoSys, and Incyte; AT served as a consultant for and received honoraria from Janssen-Cilag SpA, AstraZeneca, and AbbVie; JCP, SBK, and SF are employees of and have equity ownership in BeiGene USA; VR is an employee of, has a leadership role and equity ownership in, and received travel funding from BeiGene USA; AC is an employee of, has equity ownership in, and received travel funding from BeiGene USA; JH is an employee of, has a leadership role with, and has equity ownership in BeiGene USA; PH received research funding from AbbVie, Pharmacyclics, Janssen, Gilead, and Roche, received honoraria from AbbVie and Janssen, and served on advisory boards for Acerta, Janssen, and AbbVie; JRB served as a consultant for AbbVie, Acerta, AstraZeneca, BeiGene, Catapult Therapeutics, Dynamo Therapeutics, Genentech/Roche, Gilead, Juno/Celgene, Kite, Loxo, Novartis, Pfizer, Pharmacyclics, Sun, Sunesis, TG Therapeutics, Nextcea and Verastem, received research funding from Gilead, Loxo, Sun and Verastem, received honoraria from Janssen and Teva, and served on the data safety monitoring board for Morphosys and Invectys; PSG, LL and MT have nothing to disclose.

References 1. Fischer K, Cramer P, Busch R, et al. Bendamustine in combination with rituximab for previously untreated patients with chronic lymphocytic leukemia: a multicenter phase II trial of the German Chronic Lymphocytic Leukemia Study Group. J Clin Oncol. 2012;30(26):3209-3216. 2. Hallek M, Fischer K, Fingerle-Rowson G, et al. Addition of rituximab to fludarabine and cyclophosphamide in patients with chronic lymphocytic leukaemia: a randomised, open-label, phase 3 trial. Lancet. 2010;3 76(9747):1164-1174. 3. Kastenhuber ER, Lowe SW. Putting p53 in context. Cell. 2017;170(6):1062-1078. 4. Dohner H, Stilgenbauer S, Benner A, et al. Genomic aberrations and survival in chronic lymphocytic leukemia. N Engl J Med. 2000;343(26):1910-1916. 5. Yu L, Kim HT, Kasar S, et al. Survival of del17p CLL depends on genomic complexity and somatic mutation. Clin Cancer Res. 2017;23(3):735-745. 6. Aitken MJL, Lee HJ, Post SM. Emerging treatment options for patients with p53pathway-deficient CLL. Ther Adv Hematol. 2019;10:2040620719891356. 7. Brander D, Islam P, Barrientos JC. Tailored treatment strategies for chronic lymphocytic leukemia in a rapidly changing era. Am Soc Clin Oncol Educ Book. 2019;39:487-498. 8. Byrd JC, Brown JR, O'Brien S, et al. Ibrutinib versus ofatumumab in previously treated chronic lymphoid leukemia. N Engl J Med. 2014;371(3):213-223. 9. Brown JR, Hillmen P, O'Brien S, et al. Extended follow-up and impact of highrisk prognostic factors from the phase 3 RESONATE study in patients with previously treated CLL/SLL. Leukemia. 2018; 32(1):83-91. 10. O'Brien S, Jones JA, Coutre SE, et al. Ibrutinib for patients with relapsed or refractory chronic lymphocytic leukaemia

2362

Contributions Together with BeiGene authors (AC, SF, VR, and JH), CST, PH, TR, PG, BSK, and JRB were responsible for study design, and CST, JRB, JCP, SBK, AC, SF, VR, and JH contributed to data interpretation and analysis; all investigators and their respective research teams reviewed patient records and contributed to data collection; BeiGene authors (JCP, SBK, SF, VR, AC, and JH) confirmed assay validation and data accuracy and compiled data for summation and analysis; CST, JCP, SBK, SF, and VR contributed to the first draft of the manuscript; all authors further contributed to final manuscript writing; CST had final responsibility to submit for publication. All authors had full access to all of the data, carefully reviewed the manuscript, and approved the final version. Acknowledgments The authors thank the patients who participated in the study, their supporters, and the investigators and clinical research staff from the study centers. Medical writing and editorial assistance were provided, under the direction of the authors, by Bio Connections. Funding This work, including medical writing and editorial assistance, was supported by BeiGene USA, Inc. BeiGene was involved in the study design, compilation of data, and statistical analysis.

with 17p deletion (RESONATE-17): a phase 2, open-label, multicentre study. Lancet Oncol. 2016;17(10):1409-1418. 11. Calquence® [package insert]. Gaithersburg, MD: AstraZeneca; November 2019. 12. Sharman JP, Egyed M, Jurczak W, et al. Acalabrutinib with or without obinutuzumab versus chlorambucil and obinutuzmab for treatment-naive chronic lymphocytic leukaemia (ELEVATE TN): a randomised, controlled, phase 3 trial. Lancet. 2020;395(10232):1278-1291. 13. Venclyxto® [summary of product characteristics]. Ludwigshafen, Germany: AbbVie Deutschland GmbH & Co. KG; April 2020. 14. Zydelig® [summary of product characteristics]. County Cork, Ireland: Gilead Sciences, Inc.; April 2019. 15. 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. 16. 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. 17. Furman RR, Sharman JP, Coutre SE, et al. Idelalisib and rituximab in relapsed chronic lymphocytic leukemia. N Engl J Med. 2014; 370(11):997-1007. 18. O'Brien SM, Lamanna N, Kipps TJ, et al. A phase 2 study of idelalisib plus rituximab in treatment-naive older patients with chronic lymphocytic leukemia. Blood. 2015; 126 (25):2686-2694. 19. Guo Y, Liu Y, Hu N, et al. Discovery of zanubrutinib (BGB-3111), a novel, potent, and selective covalent inhibitor of Bruton's tyrosine kinase. J Med Chem. 2019 62(17):7923-7940. 20. Tam CSL, Trotman J, Opat S, et al. Phase 1 study of selective BTK inhibitor zanubrutinib in B-cell malignancies and safety and efficacy evaluation in CLL. Blood. 2019;

134(11):851-859. 21. Xu W, Yang S, Zhou K, et al. Treatment of relapsed/refractory chronic lymphocytic leukemia/small lymphocytic lymphoma with the BTK inhibitor zanubrutinib: phase 2, single-arm, multicenter study. J Hematol Oncol. 2020;13(1):48. 22. Brukinsa® [package insert]. Beijing, China: BeiGene Co. Ltd.; November 2019. 23. Hallek M, Cheson BD, Catovsky D, et al. iwCLL Guidelines for diagnosis, indications for treatment, response assessment, and supportive management of CLL. Blood. 2018;131(25):2745-2760. 24. Oken MM, Creech RH, Tormey DC, et al. Toxicity and response criteria of the Eastern Cooperative Oncology Group. Am J Clin Oncol. 1982;5(6):649-655. 25. Mu S, Tang Z, Novotny W, et al. Effect of rifampin and itraconazole on the pharmacokinetics of zanubrutinib (a Bruton's tyrosine kinase inhibitor) in Asian and nonAsian healthy subjects. Cancer Chemother Pharmacol. 2020;85(2):391-399. 26. Schulz KF, Altman DG, Moher D, et al. CONSORT 2010 statement: updated guidelines for reporting parallel group randomized trials. Ann Int Med. 2010;152:726-732. 27. Cheson BD, Byrd JC, Rai KR, et al. Novel targeted agents and the need to refine clinical end points in chronic lymphocytic leukemia. J Clin Oncol. 2012;30(23):28202822. 28. Barrington SF, Mikhaeel NG, Kostakoglu L, et al. Role of imaging in the staging and response assessment of lymphoma: consensus of the International Conference on Malignant Lymphomas Imaging Working Group. J Clin Oncol. 2014;32(27):30483058. 29. Herman SE, Niemann CU, Farooqui M, et al. Ibrutinib-induced lymphocytosis in patients with chronic lymphocytic leukemia: correlative analyses from a phase II study. Leukemia. 2014;28(11):2188-2196. 30. Hock BD, McIntosh ND, McKenzie JL, et

haematologica | 2021; 106(9)

Zanubrutinib in treatment naïve del(17p) CLL/SLL

al. Incidence of cutaneous squamous cell carcinoma in a New Zealand population of chronic lymphocytic leukaemia patients. Intern Med J. 2016;46(12):1414-1421. 31. Royle JA, Baade PD, Joske D, et al. Second cancer incidence and cancer mortality among chronic lymphocytic leukaemia patients: a population-based study. Br J Cancer. 2011;105(7):1076-1081. 32. Fischer K, Bahlo J, Fink AM, et al. Longterm remissions after FCR chemoimmunotherapy in previously untreated patients with CLL: updated results of the CLL8 trial. Blood. 2016;127(2):208-215. 33. Burger JA, Tedeschi A, Barr PM, et al. Ibrutinib as initial therapy for patients with chronic lymphocytic leukemia. N Engl J Med. 2015;373(25):2425-2437. 34. Shanafelt TD, Wang XV, Kay NE, et al. Ibrutinib-rituximab or chemoimmunotherapy for chronic lymphocytic leukemia. N Engl J Med. 2019;381(5):432-443. 35. Woyach JA, Ruppert AS, Heerema NA, et al. Ibrutinib regimens versus chemoimmunotherapy in older patients with untreated CLL. N Engl J Med. 2018; 379:2517-2528. 36. Moreno C, Greil R, Demirkan F, et al. Ibrutinib plus obinutuzumab versus chlorambucil plus obinutuzumab in first-line

haematologica | 2021; 106(9)

treatment of chronic lymphocytic leukaemia (iLLUMINATE): a multicentre, randomised, open-label, phase 3 trial. Lancet Oncol. 2019;20(1):43-56. 37. Farooqui MZ, Valdez J, Martyr S, et al. Ibrutinib for previously untreated and relapsed or refractory chronic lymphocytic leukaemia with TP53 aberrations: a phase 2, single-arm trial. Lancet Oncol. 2015;16 (2):169-176. 38. Ahn IE, Farooqui MZH, Tian X, et al. Depth and durability of response to ibrutinib in CLL: 5-year follow-up of a phase 2 study. Blood. 2018;131(21):2357-2366. 39. Oscier DG, Wade R, Orchard J, et al. Prognostic factors in the UK LRF CLL4 trial. Blood. 2006;108(11):299. 40. Catovsky D, Richards S, Matutes E, et al. Assessment of fludarabine plus cyclophosphamide for patients with chronic lymphocytic leukaemia (the LRF CLL4 Trial): a randomised controlled trial. Lancet. 2007; 370(9583):230-239. 41. Strati P, Keating MJ, O'Brien SM, et al. Outcomes of first-line treatment for chronic lymphocytic leukemia with 17p deletion. Haematologica. 2014;99(8):1350-1355. 42. Fabbri G, Khiabanian H, Holmes AB, et al. Genetic lesions associated with chronic lymphocytic leukemia transformation to

Richter syndrome. J Exp Med. 2013; 210(11):2273-2288. 43. Tam CS, Opat S, Zhu J, et al. Pooled analysis of safety data from monotherapy studies of the Bruton tyrosine kinase (BTK) inhibitor, zanubrutinib (BGB-3111) in B-cell malignancies. Presented at: 24th European Hematology Association Congress; June 13-16, 2019; Amsterdam, the Netherlands. 44. Hillmen P, Brown JR, Eichhorst BF, et al. ALPINE: zanubrutinib versus ibrutinib in relapsed/refractory chronic lymphocytic leukemia/small lymphocytic lymphoma. Future Oncol. 2020;16(10):517-523. 45. Tam CS, LeBlond V, Novotny W, et al. A head-to-head phase III study comparing zanubrutinib versus ibrutinib in patients with Waldenström macroglobulinemia. Future Oncol. 2018;14(22):2229-2237. 46. Jain N, Keating M, Thompson P, et al. Ibrutinib and venetoclax for first-line treatment of CLL. N Engl J Med. 2019; 380(22):2095-2103. 47. Lampson BL, Tyekucheva S, Crombie JL, et al. Preliminary safety and efficacy results from a phase 2 study of acalabrutinib, venetoclax and obinutuzumab in patients with previously untreated chronic lymphocytic leukemia (CLL). Blood. 2019; 134(Suppl 1):S32.

2363

ARTICLE Ferrata Storti Foundation

Chronic Lymphocytic Leukemia

Phase II study of acalabrutinib in ibrutinibintolerant patients with relapsed/refractory chronic lymphocytic leukemia Kerry A. Rogers,1 Philip A. Thompson,2 John N. Allan,3 Morton Coleman,3 Jeff P. Sharman,4 Bruce D. Cheson,5 Daniel Jones,1 Raquel Izumi,6 Melanie M. Frigault,6 Cheng Quah,6 Rakesh K. Raman,6 Priti Patel,6 Min Hui Wang6 and Thomas J. Kipps7

Haematologica 2021 Volume 106(9):2364-2373

The Ohio State University, Columbus, OH; 2MD Anderson Cancer Center, Houston, TX; Weill Cornell Medicine, New York, NY; 4Willamette Valley Cancer Institute, Eugene, OR; 5 Georgetown University Hospital, Washington, DC; 6AstraZeneca, South San Francisco, CA; and 7UC San Diego Moores Cancer Center, San Diego, CA, USA 1 3

ABSTRACT

Correspondence: KERRY A. ROGERS kerry.rogers@osumc.edu Received: September 25, 2020. Accepted: February 19, 2021. Pre-published: March 18, 2021.

-cell receptor signaling inhibition by targeting Bruton tyrosine kinase (BTK) is effective in treating chronic lymphocytic leukemia. The BTK inhibitor ibrutinib may be intolerable for some patients. Acalabrutinib is a more selective BTK inhibitor that may be better tolerated by patients who are intolerant to ibrutinib. A phase II study of acalabrutinib was conducted in patients with relapsed/refractory chronic lymphocytic leukemia who were ibrutinib-intolerant and had continued disease activity. Intolerance was defined as having discontinued ibrutinib due to persistent grade 3/4 adverse events or persistent/recurrent grade 2 adverse events despite dose modification/interruption. Patients received oral acalabrutinib 100 mg twice daily until disease progression or intolerance. Sixty patients were treated. The overall response rate to acalabrutinib was 73% and three patients (5%) achieved complete remission. At a median follow-up of 35 months, the median progression-free and overall survival were not reached; 24-month estimates were 72% and 81%, respectively. The most frequent adverse events with acalabrutinib were diarrhea (53%), headache (42%), contusion (40%), dizziness (33%), upper respiratory tract infection (33%), and cough (30%). The most common reasons for acalabrutinib discontinuation were progressive disease (23%) and adverse events (17%). Most patients with baseline samples (49/52; 94%) and all with on-treatment samples (3/3; 100%) had no detectable BTK and/or PLCG2 mutations. Acalabrutinib is effective and tolerable in most patients with relapsed/refractory chronic lymphocytic leukemia who are intolerant of ibrutinib. Acalabrutinib may be useful for patients who may benefit from BTK inhibitor therapy but are ibrutinib intolerant. ClinicalTrials.gov identifier: NCT02717611.

https://doi.org/10.3324/haematol.2020.272500

Introduction ©2021 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.

2364

Targeted Bruton tyrosine kinase (BTK) inhibitors are highly effective for the treatment of chronic lymphocytic leukemia (CLL).1 These agents block signaling by inhibiting BTK, a key kinase in the B-cell receptor signaling pathway.2-4 The efficacy of BTK inhibition in CLL was demonstrated by ibrutinib, the first BTK inhibitor approved for treatment of CLL.5 Ibrutinib is not always tolerated by patients with CLL. In a large, retrospective study of ibrutinib-treated CLL, toxicity was the most common reason for treatment discontinuation, accounting for 63.1% of discontinuations in the front-line setting and 50.2% of discontinuations among patients with relapsed/refractory CLL.6 The most common toxicities leading to discontinuation were arthralgia (41.6%), atrial fibrillation (25.0%), and rash (16.7%) in the front-line setting and atrial fibrillation (12.3%), infection (10.7%), and pneumonitis (9.9%) in relapsed/refractory CLL. These toxicities may be due to BTK inhibition or off-tar-

haematologica | 2021; 106(9)

Acalabrutinib in ibrutinib-intolerant R/R CLL patients

get effects of ibrutinib on other kinases.2,5,7 Some toxicities can be managed with supportive care and some require ibrutinib discontinuation, especially if more severe. For example, current guidelines recommend careful monitoring in the case of atrial fibrillation, and potential use of non-warfarin anticoagulation, although consideration should be given to alternate therapies if the atrial fibrillation is uncontrolled.8 Rates of ibrutinib discontinuation due to adverse events during extended follow-up in a clinical trial population are approximately 20%.9 In CLL patients treated outside of clinical trials at academic and community sites, discontinuation rates due to adverse events were as high as 50%, which may better capture tolerability in a general practice setting.6 This means that patients who cannot take ibrutinib due to toxicity may not be able to realize the potential benefit of BTK inhibition on their disease, thereby reducing therapeutic options available for CLL treatment. Acalabrutinib is an oral covalent inhibitor of BTK approved for treatment of patients with CLL.10 Acalabrutinib binds to BTK at the cysteine 481 residue, which is the same binding site as that for ibrutinib.11 Compared with ibrutinib, acalabrutinib is a more selective BTK inhibitor.12,13 Fewer off-target effects potentially provide an improved safety profile compared with ibrutinib.5,14 A low frequency of adverse events of interest, specifically atrial fibrillation and severe bleeding, has been reported with acalabrutinib.11 In a phase III trial in patients with relapsed/refractory CLL (ASCEND), atrial fibrillation occurred in eight of 154 patients (5%) receiving acalabrutinib monotherapy, seven of whom had a history of ongoing hypertension. Bleeding and infections (any grade), also events of clinical interest, occurred in 40 (26%) and 87 (57%) patients, respectively. In that trial, 11% of patients receiving acalabrutinib monotherapy discontinued this treatment because of adverse events.15 Given the improved selectivity of acalabrutinib relative to ibrutinib, we hypothesized that acalabrutinib would be effective and tolerable in patients with CLL who discontinued ibrutinib due to adverse events. This hypothesis is supported by a previous study in which the overall response rate (ORR; i.e., partial response [PR] or better) to acalabrutinib was 61% in patients with relapsed/refractory CLL who were previously unable to continue ibrutinib treatment because of adverse events.16 However, the previous study did not objectively define events classified as ibrutinib-intolerant and analyzed a cohort of patients added to the open-label, phase II, dose-expansion portion of the phase I/II study.16 We therefore conducted a dedicated phase II study of acalabrutinib in patients with relapsed/refractory CLL who were intolerant to ibrutinib treatment as defined by specific criteria, including event grade, persistence, and recurrence.

Methods Study design and participants This multicenter, single-agent, phase II study (ClinicalTrials.gov identifier NCT02717611; ACE-CL-208) enrolled adults with CLL who were intolerant to ibrutinib and for whom purine analogue-based therapy was not an option. Ibrutinib intolerance was defined as: (i) having discontinued ibrutinib treatment due to grade 3 or 4 adverse events that persisted despite optimal supportive care; or (ii) having experienced

haematologica | 2021; 106(9)

grade 2 adverse events related to ibrutinib treatment that persisted for at least 2 weeks or recurred at least twice, whether the dose of ibrutinib was reduced or interrupted, despite optimal supportive care. Patients had to have had at least one prior attempt at ibrutinib treatment for CLL and not be appropriate for treatment or retreatment with purine analogue-based therapy (e.g., fludarabine). After discontinuing ibrutinib, patients had to meet the International Workshop on Chronic Lymphocytic Leukemia (iwCLL) 2008 criteria for progressive disease (PD)17 as a sign of continued disease activity and not have received other CLL therapy. To meet the eligibility criteria, patients’ most recent systemic anticancer therapy was required to be ibrutinib; those who received an alternative anticancer therapy after ibrutinib discontinuation were excluded. Patients were excluded if they had an ongoing grade 3 or 4 adverse event attributed to ibrutinib. Other patients who were excluded were those with evidence of active Richter transformation or any evidence of PD on ibrutinib; patients who had previously received a BCL-2 inhibitor; patients who had significant cardiovascular disease, such as uncontrolled or symptomatic untreated arrhythmias, congestive heart failure, or myocardial infarction within 6 months of screening, or any class 3 or 4 cardiac disease as defined by the New York Heart Association functional classification or QTc >480 ms at screening (except for controlled, asymptomatic atrial fibrillation during screening, which was allowed); and patients who were receiving anticoagulation with warfarin or equivalent vitamin K antagonists within 7 days of the first study drug dose. Patients taking other anticoagulants could be included. All patients signed written informed consent before enrollment into the study, which was approved by the institutional review board/independent ethics committee of each participating institution and conducted in accordance with the principles of the Declaration of Helsinki and International Conference on Harmonization Guidelines for Good Clinical Practice.

Procedures Eligible patients were treated with acalabrutinib 100 mg orally twice a day on days 1 to 28 of 28-day cycles until disease progression, as long as treatment was tolerated. Response was assessed according to modified iwCLL 2008 criteria,17 with the first assessment occurring 3 months after starting acalabrutinib. Adverse events were collected and graded according to Common Terminology Criteria for Adverse Events version 4.03. An exploratory analysis of molecular resistance to BTK inhibitors was performed retrospectively using deep sequencing of BTK and PLCG2 in patients with pretreatment samples18,19 (details in the Online Supplementary Methods).

Outcomes The primary endpoint was investigator-assessed ORR according to iwCLL 2008 criteria.17 ORR was defined as the proportion of patients achieving a best overall response of either complete remission (CR), complete remission with incomplete bone marrow recovery (CRi), nodular partial remission (nPR), or PR at or before initiation of subsequent anticancer therapy. Secondary efficacy endpoints were duration of response (DOR; defined as the time from the initial response [CR, CRi, nPR, or PR] until documented PD), progression-free survival (PFS; defined as the time from first dose to first documented PD or death), time to next treatment (TTNT; defined as the time from first dose to institution of subsequent anticancer therapy for CLL or death), and overall survival (OS; defined as the time from first dose to death). Safety was assessed via laboratory assessments and adverse events by their frequency, and causal attribution.

2365

K.A. Rogers et al.

Details of the statistical analysis can be found in the Online Supplementary Methods.

Results Patients, treatment, and disposition In total, 60 patients were enrolled between March 23, 2016, and August 2, 2017. Their median age was 69.5 years (range, 43-88 years) and the median time from diagnosis to first dose of study drug was 103.2 months (range, 10.3-307.9 months). Seventeen (28%) patients had del(17p) and 31 (52%) had Rai stage III or IV disease. The baseline patient and disease characteristics are shown in Table 1. The median number of prior therapies was two (range, 1-10). All patients had taken ibrutinib previously, with 50 (83%) having received ibrutinib as monotherapy and ten (17%) having received ibrutinib in combination with another agent (Online Supplementary Table S1). Forty-three (72%) patients had been exposed to an anti-CD20 monoclonal antibody and 36 (60%) had received prior systemic chemotherapy (Online Supplementary Table S1). The median duration of ibrutinib treatment was 5.7 months (range, <1-55.5). Out of total of 60 patients, 15 (25%) received ibrutinib for <2 months. Of these 15 patients, only two discontinued acalabrutinib (due to squamous cell carcinoma of the lung and endometrial cancer [n=1 each]). As ibrutinib treatment occurred before study entry, treatment response to ibrutinib was not fully captured for the entire patient population (safety, however, was captured). The most common adverse events leading to ibrutinib discontinuation were atrial fibrillation (23%), diarrhea (12%), arthralgia (10%), and rash (10%) (Online Supplementary Table S2). After ibrutinib discontinuation, the median time from taking the last dose of ibrutinib to starting acalabrutinib was 7.5 months (range, 0.831.1). At a median follow-up of 34.6 months (range, 1.1-47.4), 29 (48%) patients remained on acalabrutinib; 45 patients (75%) had at least 1 year of treatment. The median time exposed to acalabrutinib was 32 months (range, 0.3-47.4). Of the 31 patients who discontinued acalabrutinib, the most common reason for discontinuation was disease progression (n=14, 23%) followed by adverse events (n=10, 17%); other reasons were patient or physician decision (n=3 and n=3, respectively), and comorbid anorexia (n=1) (Figure 1). For the 14 patients who discontinued due to disease progression, 11 patients had an Eastern Cooperative Oncology Group performance status of 1, seven had Rai stage III-IV disease, and their median age was 72 years. Four of these 15 patients had del(17p), four had del(11q), and 12 had unmutated IGHV.

Efficacy The ORR to acalabrutinib treatment was 73% (n=44/60; 95% confidence interval [95% CI]: 60-84%) (Figure 2). The ORR in patients with del(17p) was similar (71% [n=12/17]; 95% CI: 44-90%). The ORR including patients with PR with lymphocytosis (PRL) was 78% (n=47/60; 95% CI: 66-88%), comprising three (5%) patients with a CR, two (3%) with CRi, 39 (65%) with a PR, and three (5%) with a PRL. Of the 13 patients not achieving a response, four (7%) had stable disease (SD), one (2%) had PD; six (10%) patients were not evaluable for response 2366

due to discontinuing treatment before the first response assessment at 3 months, and two (3%) were not available for response assessment. For the six patients who were not evaluable, three discontinued due to adverse events and three discontinued due to patient or physician decision (1 and 2 patients, respectively) (Figure 1). The median DOR was not reached; the estimated 24-month DOR was 81% (n=44, 95% CI: 66-90%) and 78% (n=47, 95% CI: 63-88%) when patients with PRL were included, and the estimated 36-month DOR was 65% (95% CI: 46-79%) and 64% (95% CI: 45-77%) when patients with PRL were included (Figure 3A and B, respectively). The median PFS was not reached; estimated 24-month and 36-month PFS rates were 72% (95% CI: 58-82%) and 58% (95% CI: 42-71%), respectively (Figure 4). The median OS was not reached. Estimated 24-month and 36month OS rates were 81% (95% CI: 68-89%) and 78% (95% CI: 65-87%), respectively (Figure 4). Sixteen (27%) patients started a subsequent treatment for CLL, and the median TTNT was 44 months (95% CI: 27-not estimable) (Online Supplementary Figure S1). The efficacy (ORR, DOR, PFS) of acalabrutinib was also assessed by duration of previous ibrutinib treatment and by duration of treatment hold (time from ibrutinib discontinuation to start of acalabrutinib). These assessments were exploratory, and no statistical analyses were performed. The ORR was 64% (n=20/31; 95% CI: 45-81%) in patients who received prior ibrutinib treatment for ≥6 months and 83% (n=24/29; 95% CI: 64-94%) in those who received ibrutinib for <6 months. DOR and PFS on acalabrutinib in patients who received prior ibrutinib treatment for ≥6 months trended towards being shorter (no statistical analyses were performed) (Online Supplementary Figure S2 A, C, and E). Duration of treatment hold did not appear to affect ORR, DOR, or PFS during acalabrutinib treatment (Online Supplementary Figure S2 B, D, and F).

Table 1. Patient’s baseline.

Characteristic

N=60

Age in years, median (range) Men, n (%) ECOG PS ≤1, n (%) Number of prior systemic therapies, n (%) 1 2 3 ≥4 β2-microglobulin >3 mg/L, n/N (%) Genetic risk features, n/N (%) Unmutated IGHV del(11q)a del(17p)a Rai stage III-IV, n (%) Lymph nodes ≥5 cm, n (%) Laboratory values, median (range) Lymphocyte count, 109/L Neutrophil count, 109/L Hemoglobin, g/dL Platelet count, 109/L

69.5 (43-88) 38 (63) 58 (97) 14 (23) 18 (30) 11 (18) 17 (28) 46/58 (79) 46/58 (79) 14/60 (23) 17/60 (28) 31 (52) 19 (32) 12.3 (0.9-172.4) 3.3 (0.4-20.1) 12.2 (7.5-17.3) 117.5 (37-350)

a By fluorescence in situ hybridisation testing. ECOG PS: Eastern Cooperative Oncology Group performance status; IGHV: immunoglobulin heavy chain gene.

haematologica | 2021; 106(9)

Acalabrutinib in ibrutinib-intolerant R/R CLL patients

Safety The most frequent adverse events of any grade occurring with acalabrutinib were diarrhea (n=32, 53%), headache (n=25, 42%), contusion (n=24, 40%), dizziness (n=20, 33%), upper respiratory tract infection (n=20, 33%), and cough (n=18, 30%) (Table 2 and Online Supplementary Table S3). The most frequent grade ≥3 adverse events were pneumonia (n=9, 15%), neutropenia (n=7, 12%), increased lymphocyte count (including lymphocytosis and lymphocyte count increased; n=8, 13%), and thrombocytopenia (including platelet count decreased and thrombocytopenia; n=5, 8%) (Table 2). Serious adverse events of any grade were experienced by 31 (52%) patients. Treatment-related severe adverse events of any grade that were deemed related to acalabrutinib by the investigator were experienced by ten (17%) patients. There were five dose reductions in four patients; one patient had two dose reductions due to vaginal yeast infection. All four patients had successful adverse event management with dose reduction and continued on study. However, one of these patients later discontinued acalabrutinib due to adverse events. Ten (16.7%) patients had an adverse event leading to

acalabrutinib discontinuation, including pneumonia (n=2, 1 grade 3 event and 1 death), diarrhea (n=1, grade 2), headache (n=1, grade 1), endometrial cancer (n=1, grade 3), stomatitis (n=1, grade 2), subdural hematoma (n=1, grade 2), cerebrovascular accident (n=1, grade 2), increased transaminases (n=1, grade 4), and squamous cell carcinoma of lung (n=1, grade 2). Among these events, the investigator considered diarrhea, headache, stomatitis, and subdural hematoma to be related to the acalabrutinib treatment. Only one patient discontinued acalabrutinib due to the same adverse event (diarrhea) that resulted in prior ibrutinib discontinuation; grade 3 or 4 diarrhea led to ibrutinib discontinuation and grade 2 diarrhea led to acalabrutinib discontinuation. To better understand acalabrutinib tolerability following ibrutinib discontinuation, the incidence of ibrutinibintolerance adverse events was examined during acalabrutinib treatment. Among 60 enrolled patients, 27 ibrutinib-intolerance adverse events occurred in 24 (40%) patients during acalabrutinib treatment. Of these, most events (67% [n=18/27 events in 18 patients) were lower grade on acalabrutinib than on prior ibrutinib treatment; 30% (n=8/27 events in 6 patients) were of an unchanged

Figure 1. Trial profile. AE: adverse events.

haematologica | 2021; 106(9)

2367

K.A. Rogers et al.

grade (Online Supplementary Table S4). Only one event (4% [n=1/27] events in 1 patient) was of a higher grade during acalabrutinib treatment than during prior ibrutinib treatment. This event was increased liver function test (grade 2 on ibrutinib and grade 3 during acalabrutinib treatment). Two patients treated with acalabrutinib had recurrence of the same ibrutinib-intolerance adverse event of atrial fibrillation, both of whom had atrial fibrillation events with a lower severity grade on acalabrutinib treatment (grade 3/2 and grade 2/1 during ibrutinib/acalabrutinib treatment). One of the two patients had a medical history of atrial fibrillation and hypertension and discontinued acalabrutinib treatment due to pneumonia that started the same day as atrial fibrillation. The second patient had a medical history of atrial fibrillation; at the time of data analysis, treatment with acalabrutinib was ongoing in the presence of ongoing atrial fibrillation (the dose was not changed and no other action was taken). Among 60 enrolled patients, during ibrutinib treatment, 41 patients had the following ibrutinib-intolerance adverse events: arthralgia, atrial fibrillation, bleeding, diarrhea, or rash (Table 3, Online Supplementary Table S5). Of the 74 ibrutinib-intolerance adverse events in the 60 enrolled patients, 42 (57%) did not recur during acalabrutinib treatment. Eighteen (30%) patients treated with acalabrutinib had recurrence of the same ibrutinib-intolerance adverse event. The most common ibrutinib-intolerance adverse events recurring with acalabrutinib were diarrhea (n=5) and bleeding events (n=5), all of which had the same or lower severity grade with acalabrutinib treatment. Among the five patients with recurrent bleeding events, only one patient had the same type of bleeding event reported with ibrutinib and acalabrutinib (recurrent hematuria); the other four patients had bleeding events with ibrutinib that were different from those reported on acalabrutinib (Online Supplementary Table S4). Eleven deaths occurred during the study; the causes included pneumonia (n=3), Richter transformation (n=2), bronchopulmonary aspergillosis, ventricular fibrillation, squamous cell carcinoma of lung, multiple organ dysfunction syndrome, disease progression, and death (n=1

each). Of the seven deaths due to adverse events, two occurred while the patient was receiving study treatment (1 each, pneumonia and subdural hematoma), and the remainder occurred following acalabrutinib discontinuation. Of note, an 85-year-old male patient with multiple cardiac morbidities died of ventricular fibrillation 27 days after acalabrutinib was discontinued due to stomatitis. The other event of interest which resulted in death occurred in a 76-year-old female patient with a medical history of herpes zoster who died of bronchopulmonary aspergillus 7 days after discontinuation of acalabrutinib due to pneumonia.

Analysis of mutations associated with resistance to BTK inhibitors To determine whether mutations associated with resistance to BTK inhibitors were present before acalabrutinib treatment, purified B-cell samples at the start of acalabrutinib treatment were tested for mutations in BTK and PLCG2. Mutations in BTK and PLCG2 have been associated with clinical disease progression during ibrutinib treatment and with resistance to acalabrutinib because both bind to BTK at the same site.18-22 Samples were available for 55 of 60 (92%) patients. Pretreatment samples were available for 52 patients; samples were taken at later time-points for three patients (cycle 1, day 28, n=1; cycle 6, day 28, n=2). Three (5%) were found to have a mutation in at least one gene. Two patients had multiple mutations associated with ibrutinib resistance in BTK or BTK and PLCG2, while one patient had a PLCG2 mutation of uncertain significance (Online Supplementary Table S6).18,19,23,24 All mutations were found in pretreatment samples. Of the two patients with mutations associated with BTK and/or PCLG2, one was electively taken off acalabrutinib after just over 2 months of exposure due to the presence of these mutations and was not evaluable for evaluation of response to acalabrutinib. The other patient received acalabrutinib and had PD after 15 months, having achieved a best response of SD during acalabrutinib treatment and at CLL progression, the major BTK C481S

Figure 2. Response to acalabrutinib. Patients who discontinued study treatment for before evaluation response (n=6) or who were not available for response assessment (n=2) were classified as not evaluable. CR: complete remission; Cri: CR with incomplete bone marrow recovery; ORR: overall response rate; PD: progressive disease; PR: partial remission; PRL: partial remission with lymphocytosis; SD: stable disease.

2368

haematologica | 2021; 106(9)

Acalabrutinib in ibrutinib-intolerant R/R CLL patients

clone identified at baseline expanded from 30.7% to 90.2% allele fraction (Online Supplementary Tables S6 and S7). The patient with the D993N missense variant in PLCG2 (which has been associated with ibrutinib resistance, but not shown to alter drug sensitivity in vitro to date), achieved a CR with acalabrutinib at treatment cycle 18 and had remained on therapy for 25 months at the time of data cutoff. At the time of this analysis, five patients had relapsed on acalabrutinib and peripheral blood mononuclear cell samples were collected at treatment termination. Three of the five patients were confirmed to have no BTK or PLCG2 mutations; one of these patients experienced a best overall response of PRL on acalabrutinib treatment (DOR, 11.53 months), the second patient had a best overall response of PR (DOR, 15.67 months), and the third patient had SD (Online Supplementary Table S7). The fourth patient who achieved a PR (DOR, 14.29 months) had low levels of BTK C481S and T474I mutations as well as a predominant PLCG2 1140N mutation at treatment termination, none of which was detectable at baseline. The fifth patient, described above, had a pre-existing clone with BTK C481S expansion during treatment that was detectable at progression (Online Supplementary Table S7).

Discussion This phase II study of acalabrutinib in patients who were ibrutinib-intolerant demonstrated that acalabrutinib is effective and tolerable in a large proportion of this population. The ORR of 73% with a median PFS that was not reached demonstrates durable disease control in this population of relapsed/refractory CLL patients. A similar response rate was reported with ibrutinib in the front-line setting in a population of elderly patients with a similar median age (71 years) and follow-up duration (22.1 months).25 In this study, 10% of patients were not evaluable for response because they discontinued treatment before the first response assessment. As responses to BTK inhibitors tend to improve with longer treatment duration, it is possible that with additional follow-up, the ORR will increase and additional CR will be observed.9 It was not unexpected that acalabrutinib was effective in these patients, as the prior ibrutinib exposure was short for most patients (median, <6 months) and most were assumed to have a response to BTK inhibition (based on disease progressing after discontinuing ibrutinib due to adverse events). The safety of acalabrutinib in this study is perhaps more helpful than the observed response rate in understanding the impact of this agent. At a median follow-up

Figure 3. Duration of response to acalabrutinib. (A, B) The median duration of response was not reached when patients with partial remission with lymphocytosis were excluded (A) or when they were included (B). CI: confidence interval; DOR: duration of response; PR: partial remission; PRL: partial remission with lymphocytosis.

haematologica | 2021; 106(9)

2369

K.A. Rogers et al.

of 35 months, 48% of patients remained on acalabrutinib. The most common reason for discontinuation was disease progression (23%) and the rate of acalabrutinib discontinuation due to adverse events was 17%. This rate of discontinuation due to adverse events is low considering that 100% of patients had discontinued ibrutinib due to adverse events, suggesting that acalabrutinib is tolerable in a large proportion of patients who are intolerant of ibrutinib. Comparing the full spectrum of adverse events between ibrutinib and acalabrutinib in this study is difficult because the ibrutinib experience was not captured prospectively. The study was not intended to compare toxicity between two drugs, but rather to determine acalabrutinib tolerability in patients who discontinued ibrutinib due to toxicity. When reviewing events of arthralgia, atrial fibrillation, bleeding, diarrhea, and rash leading to ibrutinib intolerance, 24/41 patients experienced recurrence during acalabrutinib treatment, and recurrence was at a similar (25%) or lower (75%) grade of severity in all patients. Most adverse events (64%) limiting ibrutinib treatment were not experienced during acalabrutinib treatment. Additionally, all adverse events causing ibrutinib intolerance and recurring with acalabrutinib treatment (27 events in total) were reviewed to determine dif-

ferences in maximal severity grade experienced. Of these adverse events, only one occurred at a higher grade, while 18 occurred at a lower grade with acalabrutinib, demonstrating that the severity of intolerance adverse events during acalabrutinib treatment may be decreased. This reduction in rate includes hemorrhage events, which have previously been observed to be a class effect of BTK inhibitors,26 but in this study were observed to occur at a lower grade with acalabrutinib than with ibrutinib. Only one patient discontinued acalabrutinib for the same adverse event (diarrhea) that was also the cause reported for ibrutinib discontinuation. Clinical strategies for patients with ibrutinib intolerance, such as switching to alternative kinase inhibitors or combining different therapeutic agents, have been evaluated. Real-world data have suggested that ibrutinib-intolerant patients could be treated successfully with an alternative kinase inhibitor.27 Early phase clinical trial data have also demonstrated the efficacy and safety of an alternative kinase inhibitor, umbralisib, in ibrutinib-intolerant patients. There is a potential clinical benefit in switching patients with ibrutinib intolerance to another BTK inhibitor so that venetoclax remains a future treatment option. However, depending on the type and severity of the adverse events and their potential for harm of

Figure 4. Progression-free survival and overall survival with acalabrutinib. (A, B) The medians were not reached for progression-free survival (A) or overall survival (B). CI: confidence interval; OS: overall survival; PFS: progression-free survival.

2370

haematologica | 2021; 106(9)

Acalabrutinib in ibrutinib-intolerant R/R CLL patients

recurrence, switching to another drug class such as venetoclax should be considered.28 The safety profile and efficacy of different therapeutic agents and combination strategies has been evaluated in head-to-head trials. One such trial is ASCEND, a phase III study of acalabrutinib monotherapy versus rituximab plus idelalisib (I-R) or rituximab plus bendamustine (B-R), which demonstrated improved PFS for acalabrutinib compared with either rituximab combination; there were fewer serious adverse events and fewer adverse events leading to discontinuation with acalabrutinib monotherapy compared with IR.29 In that study, fatal adverse events occurred in 6/154 (4%), 5/118 (4%), and 2/35 (6%) patients receiving acalabrutinib monotherapy, I-R, and B-R, respectively. This study was not designed to test whether acalabru-

tinib is effective in patients with therapeutic resistance to ibrutinib and the analysis of mutations associated with resistance to ibrutinib was exploratory. Among patients with evaluable samples at baseline (92%), most (95%) had no mutation in BTK/PLCG2, as determined by deep sequencing of sorted B cells. One patient harboring a PLCG2 D993N mutation at baseline achieved a response to acalabrutinib. This is an uncommon PLCG2 potentially gain-of-function mutation and may not confer resistance to acalabrutinib or ibrutinib. However, acalabrutinib may not be effective in patients who develop progression on ibrutinib with typical resistance mutations. Of the five patients with paired baseline and progression samples, only one acquired mutations in BTK and PLCG2 after a best overall response of PR and a DOR of

Table 2. Adverse events occurring in ≥10% of patients (all grades) or ≥5% of patients (grade ≥3 in severity).

Adverse event

All grades

Grade 1

32 (53) 25 (42) 24 (40) 20 (33) 20 (33) 18 (30) 15 (25) 15 (25) 14 (23) 14 (23) 13 (22) 12 (20) 10 (17) 10 (17) 10 (17) 10 (17) 10 (17) 10 (17) 10 (17) 9 (15) 9 (15) 8 (13) 8 (13) 8 (13) 8 (13) 8 (13) 8 (13) 8 (13) 7 (12) 7 (12) 6 (10) 6 (10) 6 (10) 6 (10) 6 (10)

18 (30) 20 (33) 20 (33) 18 (30) 3 (5) 9 (15) 10 (17) 0 8 (13) 6 (10) 0 7 (12) 0 3 (5) 4 (7) 9 (15) 7 (12) 6 (10) 0 3 (5) 6 (10) 4 (7) 5 (8) 2 (3) 4 (7) 6 (10) 0 3 (5) 2 (3) 4 (7) 6 (10) 3 (5) 6 (10) 4 (7) 3 (5)

Diarrhea Headache Contusion Dizziness Upper respiratory tract infection Cough Nausea Neutropeniaa Arthralgia Fatigue Pneumonia Pyrexia Lymphocyte count increasedb Thrombocytopeniac Back pain Constipation Dyspnea Rash Sinusitis Anemia Upper-airway cough syndrome Fall Hematuria Hypertension Night sweats Peripheral edema Urinary tract infection Weight increased Abdominal pain Influenza-like illness Chills Depression Hyperhidrosis Insomnia Nasal congestion

All treated patients (N=60) Grade 2 Grade 3 11 (18) 4 (7) 4 (7) 1 (2) 17 (28) 9 (15) 5 (8) 3 (5) 5 (8) 7 (12) 4 (7) 5 (8) 2 (3) 2 (3) 5 (8) 0 3 (5) 4 (7) 10 (17) 3 (5) 3 (5) 2 (3) 2 (3) 4 (7) 4 (7) 2 (3) 7 (12) 5 (8) 4 (7) 2 (3) 0 3 (5) 0 2 (3) 3 (5)

3 (5) 1 (2) 0 1 (2) 0 0 0 7 (12) 1 (2) 1 (2) 7 (12) 0 8 (13) 3 (5) 1 (2) 1 (2) 0 0 0 3 (5) 0 2 (3) 1 (2) 2 (3) 0 0 1 (2) 0 1 (2) 1 (2) 0 0 0 0 0

Grade 4

Grade 5

0 0 0 0 0 0 0 5 (8) 0 0 0 0 0 2 (3) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 2 (3) 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

All data presented as n (%). aIncludes neutropenia and decreased neutrophil count. bIncludes lymphocytosis and increased lymphocyte count. cIncludes decreased platelet count and thrombocytopenia.

haematologica | 2021; 106(9)

2371

K.A. Rogers et al.

Table 3. Ibrutinib-intolerance adverse events and recurrence after acalabrutinib treatment.

Adverse event Atrial fibrillation Diarrhea Rash Bleedingc,d Arthralgia Total

Number of patients with ibrutinib intolerancea 16 7 7 6 7e 41

Acalabrutinib experience for same patients Total

Lower grade

Same grade

Higher grade

2 5 3 5 2 24

2 3 3 3 1 18

0 2 0 2 1 6

0 0 0 0 0 1

a Among 60 patients meeting the study enrollment criteria, 41 patients had a medical history of one or more (43 events in total) of the following categories of ibrutinib-intolerance events: atrial fibrillation, diarrhea, rash, bleeding, or arthralgia. bIncludes patients with atrial flutter (n=2). cEvents categorized as bleeding included ecchymosis, hemorrhage, epistaxis, contusion, hematuria, and subdural hematoma. dAll but one patient experienced a different type of bleeding event with acalabrutinib compared with ibrutinib treatment. e Includes one patient with arthritis.

14.29 months. One patient had low levels of the BTK C481S and the T4741 gatekeeper resistance mutation at baseline as well as the PLCG2 D1140N C2 domain mutation detected at progression. The PLCG2 D1140N mutation was predominant (indicating many CLL cells in the sample were without a BTK mutation), whereas with ibrutinib, treatment resistance mutations in this PLCG2 domain were more commonly secondary mutations after BTK C481X development.19 This study was designed to determine whether acalabrutinib is effective in patients intolerant to ibrutinib or unable to continue ibrutinib treatment due to adverse events. However, it is acknowledged that the study had a few limitations, the most significant being that the ibrutinib experience was not prospectively or rigorously captured. This not only means that a significant portion of these patients’ responses to ibrutinib were unknown, but also that not all of the details were captured for the adverse events on ibrutinib. In addition, subjective reporting of adverse events by patients prior to enrollment who sought to have access to the study drug could have influenced the patients’ enrollment. It is therefore possible that some adverse events occurring at a low grade with ibrutinib may have occurred at a greater severity with acalabrutinib. To partially overcome this limitation, we applied two different approaches to assessing the occurrence of known adverse events with ibrutinib during acalabrutinib treatment. However, only a prospective or randomized study could fully capture differences in toxicities between the two drugs. The other important limitation is in understanding differential CLL resistance to acalabrutinib. PD was the most common reason for acalabrutinib discontinuation, with a relatively high rate of 23%. The direct comparison of acalabrutinib with ibrutinib is ongoing via a phase III randomized non-inferiority clinical trial in patients with previously treated, high-risk CLL (NCT02477696). In summary, the results of this study demonstrate that acalabrutinib is a safe and effective option for patients with relapsed/refractory CLL who are not able to tolerate ibrutinib. Acalabrutinib is an important therapeutic option in this population and will allow more CLL patients to benefit from BTK inhibitor treatment. Disclosures KAR has received research funding from Genentech, AbbVie, and Janssen, serves on advisory boards for Acerta Pharma, AstraZeneca, and Pharmacyclics, and has received travel sup-

2372

port from Acerta Pharma. PAT has received research funding from Pharmacyclics, AbbVie, Genentech, Pfizer, and Amgen, and has served as a consultant to Pharmacyclics, AbbVie, Genentech, Amgen, Janssen-Cilag, and Gilead. JNA has received research funding from Genentech, Janssen, and Celgene, and has served as a consultant to Pharmacyclics, AbbVie, Genentech, AstraZeneca, Sunesis, and Janssen. MC has stock and other ownership interests in Gilead Sciences and Immunomedics, has served as a consultant to Celgene, Gilead, and Pharmacyclics, serves on the speakers bureau for Gilead, Janssen China R&D, and Pharmacyclics, and has received research funding from Bristol-Myers Squibb, Celgene, GlaxoSmithKline, Merck, Millennium Pharmaceuticals, and Pharmacyclics. JPS has received research funding from Sunesis, Gilead, Acerta Pharma, Pharmacyclics, AbbVie, BeiGene, and TG Therapeutics, and has served as a consultant to Acerta Pharma, Pharmacyclics, AbbVie, BeiGene, and TG Therapeutics. BDC has received research funding from Pharmacyclics/Janssen, AstraZeneca, AbbVie, TG Therapeutics, MorphoSys, and Roche-Genentech, and has served as a consultant to Pharmacyclics/Janssen, AstraZeneca, AbbVie, TG Therapeutics, Karyopharm, and MorphoSys. DJ has received research funding from Acerta Pharma and Pharmacyclics. RI is an employee of Acerta Pharma, has equity ownership in AstraZeneca, and has patents for acalabrutinib. MMF is an employee and stock shareholder of AstraZeneca. CQ, RKR, and MHW are employees of Acerta Pharma. TJK has received research funding from AbbVie and HoffmanLaRoche. Contributions The clinical study was designed by RI and Ahmed Hamdy (both of Acerta Pharma) in collaboration with John C. Byrd (The Ohio State University Comprehensive Cancer Center). Data collection and interpretation were done by the authors, investigators, and study sponsor. Statistical analyses were performed by MHW and DJ, who also oversaw the mutation testing. KAR developed the first draft of the manuscript, for which all authors reviewed and provided important intellectual contributions; all authors approved the final version for publication. All authors had full access to the data and vouch for the completeness and accuracy of the data. Acknowledgments The authors acknowledge the Acerta Pharma study team for their commitment to this study. They also thank the patients who participated in this study as well as their friends and family who supported them.

haematologica | 2021; 106(9)

Acalabrutinib in ibrutinib-intolerant R/R CLL patients

Funding KAR is a scholar in clinical research of and received grant support from the Leukemia & Lymphoma Society. This grant provides salary support for clinical research and for writing and other study activities. The study was funded by Acerta Pharma (South San Francisco, CA, USA) a member of the AstraZeneca Group. Acerta Pharma provided the study drug. Medical writing assistance, funded by Acerta Pharma, was provided by Tracy Diaz, PhD, and Cindy Gobbel, PhD, of Peloton Advantage, LLC, an OPEN Health company.

References 1. Pal Singh S, Dammeijer F, Hendriks RW. Role of Bruton's tyrosine kinase in B cells and malignancies. Mol Cancer. 2018;17(1):57. 2. Herman SE, Gordon AL, Hertlein E, et al. Bruton tyrosine kinase represents a promising therapeutic target for treatment of chronic lymphocytic leukemia and is effectively targeted by PCI-32765. Blood. 2011;117(23):6287-6296. 3. Woyach JA, Bojnik E, Ruppert AS, et al. Bruton's tyrosine kinase (BTK) function is important to the development and expansion of chronic lymphocytic leukemia (CLL). Blood. 2014;123(8):1207-1213. 4. Herman SEM, Montraveta A, Niemann CU, et al. The Bruton tyrosine kinase (BTK) inhibitor acalabrutinib demonstrates potent on-target effects and efficacy in two mouse models of chronic lymphocytic leukemia. Clin Cancer Res. 2017;23(11): 2831-2841. 5. Byrd JC, Furman RR, Coutre SE, et al. Targeting BTK with ibrutinib in relapsed chronic lymphocytic leukemia. N Engl J Med. 2013;369(1):32-42. 6. Mato AR, Nabhan C, Thompson MC, et al. Toxicities and outcomes of 616 ibrutinibtreated patients in the United States: a realworld analysis. Haematologica. 2018;103(5): 874-879. 7. Bose P, Gandhi VV, Keating MJ. Pharmacokinetic and pharmacodynamic evaluation of ibrutinib for the treatment of chronic lymphocytic leukemia: rationale for lower doses. Expert Opin Drug Metab Toxicol. 2016;12(11):1381-1392. 8. National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines in Oncology: Chronic Lymphocytic Leukemia/Small Lymphocytic Lymphoma version 2.2021. December 3, 2020. Available from: https://www.nccn.org/ professionals/physician_gls/default.aspx. Accessed December 21, 2020. 9. O'Brien S, Furman RR, Coutre S, et al. Single-agent ibrutinib in treatment-naive and relapsed/refractory chronic lymphocytic leukemia: a 5-year experience. Blood. 2018;131(17):1910-1919. 10. Calquence approved in the US for adult patients with chronic lymphocytic

haematologica | 2021; 106(9)

Data-sharing Acerta Pharma, a member of the AstraZeneca Group, is committed to data transparency and will consider data-sharing requests on a case-by-case basis. Any requests for de-identified patients’ data can be submitted to Acerta Pharma 3 months post-publication and ending 5 years following article publication with the intent-toachieve aims of the original proposal. In addition, Acerta Pharma will provide the study protocol, statistical analysis plan, and informed consent form, as well as post results on clinicaltrials.gov, as required.

leukaemia [press release]. 2019. Available from: https://www.astrazeneca.com/content/astraz/media-centre/pressreleases/2019/calquence-approved-in-theus-for-adult-patients-with-chronic-lymphocytic-leukaemia-21112019.html. Accessed December 21, 2020. 11. Byrd JC, Harrington B, O'Brien S, et al. Acalabrutinib (ACP-196) in relapsed chronic lymphocytic leukemia. N Engl J Med. 2016;374(4):323-332. 12. Bye AP, Unsworth AJ, Desborough MJ, et al. Severe platelet dysfunction in NHL patients receiving ibrutinib is absent in patients receiving acalabrutinib. Blood Adv. 2017;1 (26):2610-2623. 13. Barf T, Covey T, Izumi R, et al. Acalabrutinib (ACP-196): a covalent bruton tyrosine kinase inhibitor with a differentiated selectivity and in vivo potency profile. J Pharmacol Exp Ther. 2017;363(2):240-252. 14. Byrd JC, Brown JR, O'Brien S, et al. Ibrutinib versus ofatumumab in previously treated chronic lymphoid leukemia. N Engl J Med. 2014;371(3):213-223. 15. Ghia P, Pluta A, Wach M, et al. ASCEND: phase III, randomized trial of acalabrutinib versus idelalisib plus rituximab or bendamustine plus rituximab in relapsed or refractory chronic lymphocytic leukemia. J Clin Oncol. 2020;38(25):2849-2861. 16. Awan FT, Schuh A, Brown JR, et al. Acalabrutinib monotherapy in patients with chronic lymphocytic leukemia who are intolerant to ibrutinib. Blood Adv. 2019;3(9): 1553-1562. 17. Hallek M, Cheson BD, Catovsky D, et al. Guidelines for the diagnosis and treatment of chronic lymphocytic leukemia: a report from the International Workshop on Chronic Lymphocytic Leukemia updating the National Cancer Institute-Working Group 1996 guidelines. Blood. 2008;111(12): 5446-5456. 18. Woyach JA, Ruppert AS, Guinn D, et al. BTK(C481S)-mediated resistance to ibrutinib in chronic lymphocytic leukemia. J Clin Oncol. 2017;35(13):1437-1443. 19. Jones D, Woyach JA, Zhao W, et al. PLCG2 C2 domain mutations co-occur with BTK and PLCG2 resistance mutations in chronic lymphocytic leukemia undergoing ibrutinib treatment. Leukemia. 2017;31(7):1645-1647. 20. Byrd JC, Wierda WG, Schuh A, et al.

Acalabrutinib monotherapy in patients with relapsed/refractory chronic lymphocytic leukemia: updated phase 2 results. Blood. 2020;135(15):1204-1213. 21. Woyach JA, Furman RR, Liu TM, et al. Resistance mechanisms for the Bruton's tyrosine kinase inhibitor ibrutinib. N Engl J Med. 2014;370(24):2286-2294. 22. Woyach J, Huang Y, Rogers K, et al. Resistance to acalabrutinib in CLL is mediated primarily by BTK mutations. Blood. 2019;134(Suppl 1):504. 23. Albitar A, Ma W, DeDios I, et al. Using highsensitivity sequencing for the detection of mutations in BTK and PLCgamma2 genes in cellular and cell-free DNA and correlation with progression in patients treated with BTK inhibitors. Oncotarget. 2017;8(11): 17936-17944. 24. Liu TM, Woyach JA, Zhong Y, et al. Hypermorphic mutation of phospholipase C, gamma2 acquired in ibrutinib-resistant CLL confers BTK independency upon Bcell receptor activation. Blood. 2015;126(1): 61-68. 25. O'Brien S, Furman RR, Coutre SE, et al. Ibrutinib as initial therapy for elderly patients with chronic lymphocytic leukaemia or small lymphocytic lymphoma: an open-label, multicentre, phase 1b/2 trial. Lancet Oncol. 2014;15(1):48-58. 26. Series J, Garcia C, Levade M, et al. Differences and similarities in the effects of ibrutinib and acalabrutinib on platelet functions. Haematologica. 2019;104(11):22922299. 27. Mato AR, Nabhan C, Barr PM, et al. Outcomes of CLL patients treated with sequential kinase inhibitor therapy: a real world experience. Blood. 2016;128(18): 2199-2205. 28. Mato AR, Schuster SJ, Lamanna N, et al. A phase 2 study to assess the safety and efficacy of umbralisib (TGR-1202) in patients with chronic lymphocytic leukemia (CLL) who are intolerant to prior BTK or PI3Kδ inhibitor therapy. Hematol Oncol. 2019;37(s2):88-89. 29. Ghia P, Pluta A, Wach M, et al. ASCEND: Phase III, randomized trial of acalabrutinib versus idelalisib plus rituximab or bendamustine plus rituximab in relapsed or refractory chronic lymphocytic leukemia. J Clin Oncol. 2020;38(25):2849-2861.

2373

ARTICLE Ferrata Storti Foundation

Haematologica 2021 Volume 106(9):2374-2383

Compications in Hematology

Elastography improves accuracy of early hepato-biliary complications diagnosis after allogeneic stem cell transplantation Pierre-Edouard Debureaux,1* Pierre Bourrier,2* Pierre Emmanuel Rautou,3,4 Anne-Marie Zagdanski,2 Morgane De Boutiny,2 Simona Pagliuca,1 Aurélien Sutra de Galy,1 Marie Robin,1 Régis Peffault de Latour,1,4 Aurélie Plessier,3 Flore Sicre de Fontbrune,1 Aliénor Xhaard,1 Pedro Henrique Prata,1 Dominique Valla,3,4 Gérard Socie1,4# and David Michonneau1,4# 1

Hematology and Transplantation Unit, Saint Louis Hospital, APHP, Paris; 2Radiology Unit, Saint Louis Hospital, APHP, Paris; 3DHU Unit, Pôle des Maladies de l’Appareil Digestif, Service d'Hépatologie, Centre de Référence des Maladies Vasculaires du Foie, Hôpital Beaujon, AP-HP, Clichy and 4Université de Paris, INSERM U976, Paris, France *PED and PB contributed equally as co-first authors. # GS and DM contributed equally as co-senior authors

ABSTRACT

Correspondence: DAVID MICHONNEAU david.michonneau@aphp.fr Received: March 5, 2020. Accepted: July 27, 2020. Pre-published: July 30, 2020. https://doi.org/10.3324/haematol.2019.245407

2374

ignificant morbidity and mortality have been associated with liver complications after allogeneic hematopoietic stem cell transplantation (allo-HSCT). Causes and consequences of these hepato-biliary complications are various and might be life-threatening. A high misdiagnosis rate has been reported because of a weak correlation between clinical, laboratory and imaging data. Liver elastography, a liver stiffness measure, is able to assess liver fibrosis and portal hypertension in most liver diseases, but data after allo-HSCT are scarce. Our aim was to determine the interest of sequential liver stiffness measurements for the diagnosis of early hepatic complications after allo-HSCT. Over a 2-year time period, 161 consecutive adult patients were included and 146 were analyzed. Ultrasonography and elastography measurements were performed before transplantation, at day+7 and day+14 by three different experienced radiologists unaware of the patients’ clinical status. Eightyone (55%) patients had liver involvements within the first 100 days after allo-HSCT. Baseline elastography was not predictive for the occurrence of overall liver abnormalities. A significant increase in two-dimensional real-time shearwave elastography (2D-SWE) was found in patients with sinusoidal obstruction syndrome (SOS). Fifteen patients (10%) fulfilled European Society for Blood and Marrow Transplantation (EBMT) score criteria and twelve (8%) reached Baltimore criteria for SOS diagnosis, but only six (4%) had a confirmed SOS. 2D-SWE at day+14 allowed early detection of SOS (AUROC=0.84, P=0.004) and improved sensibility (75%), specificity (99%) and positive predictive value (60%) over the Seattle, Baltimore or EBMT scores. A 2D-SWE measurement above 8.1 kPa at day+14 after allo-HSCT seems a promising, non-invasive, and reproducible tool for early and accurate diagnosis of SOS.

Introduction Over the past two decades, overall survival rate after allo-HSCT has improved.1 However, transplantation-related mortality (TRM) remains a significant cause of death, with a reported 15% to 35% rate in the current era.1 Hepato-biliary complications lead to significant morbidity and TRM after allo-HSCT.2 They include liver graft-versus-host disease (GvHD), sinusoidal obstruction syndrome (SOS), drug-induced hepatotoxicity, cholangitis lenta, malignant infiltration, iron overload, hemodynamic modification and biliary obstruction.2 Previous studies have reported a higher TRM rate in patients who had a high level of bilirubinemia

haematologica | 2021; 106(9)

Elastography in sinusoidal obstruction syndrome

(above 4 mg/dL or 68 μmol/L) after allo-HSCT.3 More recently, even after reduced-intensity conditioning (RIC), 20% to 26% of patients were reported to have a bilirubin peak above 4 mg/dL, which was also associated with poor outcomes.1,4 Clinical, biological and imaging data are weekly correlated and can lead to delayed or wrong diagnoses.5 Although liver biopsy is useful to establish diagnosis, it has limitations including its feasibility in severely-ill patients and bleeding risk in case of severe thrombocytopenia.5 Several clinical scores have been developed to diagnose SOS after allo-HSCT6,7 but currently lack of specificity and sensibility.8 We previously observed that among patients who fulfilled clinical criteria for SOS diagnosis, less than a half had a final diagnosis of SOS after liver biopsy.5 Pitfall associated with ultrasonography and Doppler are due to heterogeneity and lack of reproducibility8,9 and to the late onset of some radiological signs (e.g., reverse flow in the portal vein) that can occur in SOS. There are currently no early ultrasonography specific signs that are able to discriminate these various hepatic complications.8 Liver stiffness measurement using elastography is widely used and recommended for the assessment of liver fibrosis, cirrhosis and portal hypertension.11,12 Different techniques have been described: transient elastography (TE with FibroScan®), point shear wave with acoustic radiation force impulse (ARFI), and two-dimensional real-time shear wave (2D-SWE).1,13 The objective of this study was to determine the feasibility and interest of sequential measures of liver elastography for the diagnosis of early hepatic complications after allo-HSCT.

Methods

Liver test and definition of liver involvements Liver involvement was considered if increased serum aspartate aminotransferase (AST) or alanine aminotransferase (ALT) level above twice the upper limit of normal values or hyperbilirubinemia (above 17 μmol/L) occurred in two consecutive measures. All medical records were retrospectively reviewed to determine the final liver diagnosis. GvHD was graded according to the modified Glucksberg’s classification.17 When patients had no other organ involvement than suspected liver GvHD, a biopsy was performed to ascertain the diagnosis (n=3). SOS diagnosis was suspected when EBMT, Baltimore or modified Seattle clinical criteria were present in patients.18–20 Diagnosis was retained only if proven on liver biopsy (n=3), or using ultrasonography and Doppler criteria, as described in EBMT classification.20 If not proven on biopsy or ultrasonography, SOS diagnosis was considered only in the absence of infectious disease, drug toxicity or GvHD (n=3), as recommended by the European Association for the Study of the Liver (EASL) guidelines.8,21 Drug-induced liver injury (DILI) was defined according to EASL guidelines.22

Statistical analysis Two-group comparisons were performed with Mann-Whitney U test and multiple comparisons were performed with KruskalWallis test followed by a Dunn’s correction for multiple comparisons. Two-way ANOVA test followed by Dunnet correction was used for multiple comparisons of data with normal distribution and equal variance. ROC curves were built for continuous variable and area under the ROC curve (AUROC) was calculated for SOS diagnosis using all ultrasound and Doppler criteria, 2D-SWE and TE measurements at baseline, day+7 and day+14. Best cutoff value was determined using Youden index. Scores performance were calculated using an intention to diagnose approach using 3x2 table, as previously described to assess performance of diagnostic tests.21,23 All statistical tests were two-tailed with a significance level of 0.05.

Patients Between July 2017 and July 2019, 212 patients underwent an allo-HSCT in the Saint Louis Hospital (Paris, France). A total of 161 patients were included. This study has been conducted in compliance with the Declaration of Helsinki. All patients gave their written consent for the registration of clinical and biological data (CNIL number 2093819), were collected and processed anonymously in a dedicated study (CNIL number 2211540), with authorization of the IRB 00003888 (study number 20-697).

Ultrasonography and elastography Ultrasonography, Doppler, and elastography were performed at baseline, at day+7, and at day+14. Two methods were used for elastography for all patients: transient elastography with Fibroscan® (Echosens, Paris, France) and 2D-SWE (Aixplorer, SuperSonic Imaging SA, Aix-en-Provence, France) with a 3.5 MHz convex ultrasound probe (SCX-6-1) and a 7.5 Mhz linear ultrasound probe (SL-10-2). For all ultrasonography and Doppler examination, the following criteria, were assessed: liver and splenic measurements, measurement of the gallbladder wall, ascites, portal vein diameter, portal vein direction flow and maximal flow velocity, spectral waveforms of the hepatic veins. Based on Lassau et al.,6 and European Society for Blood and Marrow Transplantation (EBMT) classification, an ultrasound-Doppler score based on seven criteria was performed: (i) hepatomegaly, (ii) splenomegaly, (iii) gall bladder wall thickening, (iv) dilatation of main portal vein, (v) ascites, (vi) decrease mean velocity of portal vein, (vii) hepatofugal flow or no flow in portal vein. An additional ultrasonography and elastography could be performed at the discretion of the physician.

haematologica | 2021; 106(9)

Results Population overview Over two years, 146 patients out of 161 consecutive patients were analyzed (Figure 1). Fifteen patients were excluded for analysis due to incomplete ultrasonography evaluations. Main patient, disease and transplant features are summarized in Table 1 and Table 2. Median follow-up was 9.2 months (range, 3-19). Five donors had prior hepatitis B virus (HBV) hepatitis, but none had viral replication at the time of stem cell collection. None of the patients had detectable HBV DNA or hepatitis C virus (HCV) RNA. Two patients had Child-Pugh A cirrhosis (previous HBV infection and telomeropathy, one each). Eighteen (12%) and seven (5%) patients had prior cholecystectomy or splenectomy, respectively and could not be evaluated for all the ultrasonography criteria. During follow-up, 32 patients (22%) died, including 16 (11%) early deaths (before day+100). The leading cause of early death was TRM (93%), including three patients with SOS and two with liver GvHD who had all received a RIC regimen.

Incidence of hepatic involvements after allogeneic hematopoietic stem cell transplantation Eighty-one (55%) patients had a hepatic involvement defined by an elevation of liver enzymes and/or hyperbilirubinemia during the first 100 days after allo-HSCT (Figure 2). Hepatic GvHD was diagnosed in 11 patients 2375

P.-E. Debureaux et al.

(including three cases that were proven after liver biopsy) and were associated with skin or gut GvHD in all cases but one. Liver tests improved after GvHD treatment in six cases. SOS diagnosis was suspected when patients had clinical criteria based on classifications (EBMT, Seattle and Baltimore). SOS diagnosis was retrospectively retained when proven on liver biopsy (n=3) or in patients with usual ultrasonography and Doppler criteria (n=3), in the absence of other causes (GvHD, infectious or drugrelated) and according to the clinical evolution. Druginduced liver injury was observed in 48 patients after exposure to drugs within the conditioning regimen (n=32), azole antifungal therapy (n=12) or cyclosporine A (n=4), with normal imaging, no clinical sign for SOS or GvHD, and no positive biological test for infectious disease. All improved after drug withdrawal. A liver biopsy was performed in ten (7%) patients when liver blood test abnormalities were not explained by clinical, laboratory or imaging results. Diagnosis was established after pathology analysis in nine patients: SOS (n=3), GvHD

(n=3), DILI (n=1), HEV infection (n=1) and cholangitis lenta (n=1). One biopsy was not conclusive. Of the eight patients in whom liver biopsy was performed using the transjugular route, six had a hepatic venous pressure gradient (HVPG) <10 mmHg and did not have a SOS, while the two patients with an HVPG >10 mmHg had a confirmed SOS in biopsy.

Elastography baseline values are not associated with the occurrence of liver involvements after allogeneic hematopoietic stem cell transplantation Median basal values before transplantation were 5.4 kPa (interquartile range [IQR], 4.1-6.8) for transient elastography and 6.0 kPa (IQR, 4.9-7.8) for 2D-SWE (Figure 3A). Mean elastography measures from the three radiologists were not significantly different, suggesting that they were not dependent of the operator (P=0.39; Online Supplementary Figure S1). TE measure was obtained in 413 out of 432 procedures (two failures at baseline, five at day+7 and 12 at day+14). TE measures did not reach qual-

Figure 1. Patients flowchart. Out of 212 patients, 161 patients consented to the study and were transplanted between July 2017 and July 2019, 146 had a baseline evaluation followed by sequential ultrasonography and elastography measures at day+7 and day+14. Eighty-one patients developed liver blood tests disorders, among which 15 fulfilled European Society for Blood and Marrow Transplantation (EBMT) criteria for sinusoidal obstruction syndrome (SOS) diagnosis, including six with a proven SOS.

2376

haematologica | 2021; 106(9)

Elastography in sinusoidal obstruction syndrome

Table 1. Patients’ characteristics before transplant.

Patients’ characteristics

N=146, No. (%)

Sex (Male/Female) 86 (59%) / 60 (41%) Age at allogeneic HSCT (years, median, range) 53 (21-72) Median follow-up (months, range) 9.2 (2.5-19) Body mass index (median, range) 24.5 (16.5-42) Hepatic comorbidities Cirrhosis 2 (1%) Alcohol misuse 6 (4%) Prior resolved HBV infection (anti-Hbc and anti-HB antibodies) 9 (6%) Prior abdominal radiotherapy 2 (1%) Hematologic disorders-related hepatomegaly 10 (7%) AST or ALT > 1.5 ULN before allogeneic HSCT 6 (4%) Bilirubin > 2 mg/dL (34 μmol/L) before allogeneic HSCT 1 (1%) Ferritin > 1000 ng/mL 74 (51%) Treatment before allogeneic HSCT Gemtuzumab ozogamycin 12 (8%) Oxaliplatin 5 (3%) Prior autologous HSCT* 7 (5%) Prior allogeneic HSCT 5 (3%) Diagnosis Acute myeloid leukemia 51 (35%) Acute lymphoid leukemia 12 (8%) Myelodysplastic syndrome 27 (18%) Lymphoma 12 (8%) Cutaneous T-cell lymphoma 8 (5%) Myeloproliferative neoplasm 20 (14%) Severe aplastic anemia 11 (8%) Myeloma 1 (1%) Others** 4 (3%) Disease risk index score 1 10 (7%) 2 89 (61%) 3 31 (21%) 4 3 (2%) Not applicable 13 (9%) OMS status 0-1 130 (89%) 2 4 (3%) Missing 12 (8%) Fibroscan® transient elastography (kPa) (median, interquartile) 5.4 (4.1-6.8) 2D shear wave elastrography (kPa) (median, interquartile) 6.0 (4.9-7.8) HSCT: hematopoietic stem cell transplantation, HBV: hepatitis B virus, kPa: kilo Pascal; DRI: disease risk index; allo: allogeneic; OMS: ECOG Zubrod scrore; *one patient received two auto HSCT before allo-HSCT; **two chronic lymphoid leukemia, one plasmacytoid dendritic cell leukemia, one prolymphocytic leukemia.

ity criteria (IQR/M <0.3) from international guidelines16 and were excluded from analysis in 10.8% of procedures (n=45). A 2D-SWE measure was obtained in 414 procedures out of 432 (seven failures at baseline, six at day+7 and five at day+14). Overweight (body mass index [BMI] >29 kg/m2) was associated with 2D-SWE failure (P=0.004) but not with TE failure (P=0.21). Ascites did not influence the risk of 2D-SWE (P=0.1) or TE (P=0.09) failure. Patients with history of liver abnormalities before transplantation had significantly higher baseline values than other patients for TE (7.9 kPa vs. 5.25 kPa, respectively, P=0.0003) and 2D-SWE (9 kPa vs. 5.7 kPa, respectively, P=0.004). However, the mean baseline value of elastography was not significantly different between patients who haematologica | 2021; 106(9)

Table 2. Allogeneic hematopoietic stem cell transplantation characteristics.

Transplantation characteristics Donor HLA-matched-related HLA-matched-unrelated HLA-mismatched unrelated Haplo-identical Umbilical cord blood Source Peripheral blood stem cells Bone marrow Umbilical cord blood Donor Age (median, range) Male/Female CMV seropositivity Conditioning Myeloablative Sequential Reduced intensity Non myeloablative TBI-based conditioning Anti-thymocyte globulin GvHD prophylaxis CSA+MTX CSA+MMF CSA+MMF+Cy CSA Other GvHD grading Grade II-IV Grade III-IV

N=146, No. (%) 39 (27%) 77 (53%) 5 (3%) 18 (12%) 1 (1%) 132 (90%) 13 (9%) 1 (1%) 30 (18-67) 91 (62%)/55 (38%) 70 (48%) 28 (19%) 6 (4%) 100 (68%) 12 (8%) 16 (11%) 95 (65%) 37 (25%) 82 (56%) 20 (14%) 6 (4%) 1 (1%) 61 (42%) 20 (14%)

TBI: total body irradiation; GvHD: graft-versus-host disease; CSA: cyclosporin, MTX: methotrexate, MMF: mycophenolate mofetil; Cy: post transplantation cyclophosphamide; CMV: cytomegalovirus.

developed liver involvements after transplantation and those who did not, for both TE (6.4 kPa vs. 6.0 kPa, respectively, P=0.71) and 2D-SWE (7.0 kPa vs. 6.8 kPa, respectively, P=0.71) (Figure 3B).

Sinusoidal obstruction syndrome diagnosis is associated with an increase in two-dimensional real-time shear wave elastography In order to determine if repeated measures of elastography could improve or precede clinical diagnosis of liver disease, additional measures of TE and 2D-SWE were systematically performed at day+7 and day+14 after alloHSCT. Eight (5%) patients were not evaluable at all time points. Three patients had missing exam at day+7 related to intensice care unit (ICU) transfer (n=1) for acute renal failure (without hepatic disorder) or to missing imaging (without hepatic disorder) (n=2). Five patients had missing examination at day+14 related to ICU transfer (n=3) with early death before day+14 (one SOS, one GvHD/congestive heart failure and one with invasive fungal infection) or because of missing data (n=2) for patients with hepatic GvHD (one confirmed by hepatic biopsy and the others with gut GvHD confirmed with pathologic gut biopsy and improvement after immunosuppressive treatment). Liver stiffness value measured using TE or 2D-SWE was neither different at day+7 nor at day+14 between patients developing or not hepatic complications after allo-HSCT (Figure 3C and D). Similarly, change in liver stiffness value 2377

P.-E. Debureaux et al.

Figure 2. Description of liver blood tests involvement during the first 100 days after allogeneic hematopoietic stem cell transplantation. (A) Pie chart representing frequency of hepatic biological involvements (cytolysis, hyperbilirubinemia, or both). (B) Distribution of diagnoses according to hepatic biological disorders (cytolysis, hyperbilirubinemia, or both). Others are malignant infiltration (n=1) and aplastic anemia-related autoimmune hepatitis (n=1).

measured using TE or 2D-SWE between baseline and day+7 or day+14 was not associated with hepatic complications after allo-HSCT. We then compared the liver stiffness change from a baseline value to day+7 or day+14 according to each type of liver disease and observed a significant increase only in patients who developed SOS using 2D-SWE at day+14 (+4.15 kPa vs. -0.5kPa in patients without liver involvement, P=0.02) (Figure 3E and F). We also observed a significant increase in patients who developed SOS versus other liver complications using 2D-SWE at day +14 (+4.15 kPa vs. -0.57 kPa, P=0.018).

Two-dimensional real-time shear wave elastography improves positive predictive value of sinusoidal obstruction syndrome diagnosis scores Fifteen patients (10%) fulfilled EBMT criteria for SOS diagnosis,20 but only six (4%) had a confirmed SOS diagnosis after retrospective review of medical history (as described in the Methods section) including two lateonset SOS (Online Supplementary Table S1). The other diagnoses were liver GvHD (n=3), sepsis (n=2), cyclosporine cholestasis (n=1), HEV infection (n=1), liver GvHD with congestive heart failure (n=1), and aplastic anemia-related hepatitis with drug induced liver injury (n=1). According to the Seattle score, twenty-seven (18%) patients had criteria for SOS diagnosis,24 which was confirmed in only five. Twelve patients (8%) fulfilled SOS criteria according to the Baltimore score,19 including four confirmed SOS (Figure 4A and B). Median time between allo-HSCT and SOS diagnosis was 14 days (range, 6-22). Two (6%) and four (3.5%) patients with SOS received myeloablative conditioning (MAC) and RIC regimens, respectively. Three of the four SOS patients with RIC regimen ultimately died. Patients were followed until day+100 after allo-HSCT and no other late-onset SOS was detected. Eight patients, including the two late-onset SOS, had a supplementary ultrasound-Doppler. The higher ultrasonography-Doppler score in patient with SOS diagnosis were 2 (n=3), and 3 (n=3, including the 2 late SOS), respectively. The two late-onset SOS patients had an increase of TE value at 36 kPa (vs. 12.4 kPa at baseline) and 72 kPa (vs. 5.4 kPa at baseline). No 2D-SWE was available for these two patients. Patients without SOS (n=6) had a median 2D-SWE value at 4.3 kPa and a median TE value 2378

at 6.15 kPa. Among patients with EBMT criteria, most patients with a SOS diagnosis presented an increase of TE and 2D-SWE measures at day+7 or day+14. In four patients, increased 2D-SWE was recorded at a median of 6 days (range, 1-8) before clinical signs of SOS. 2D-SWE increased 1 day after SOS diagnosis in one patient. Among all ultrasonography, Doppler and elastography measures performed at day+7 or day+14, 2D-SWE value at day+14 was the best marker for SOS diagnosis, with a best cutoff value estimated with Youden index at 8.1 kPa (P=0.004) and a concordance index calculated at 0.84 (AUROC=0.84 [0.69-0.95]) (Figure 4C). By comparison, TE was less efficient (best cutoff at 8.2 kPa, AUROC=0.78 [0.61-0.91]) (Figure 4D). Ultrasonography and Doppler criteria for SOS were more frequently observed in patients with EBMT criteria but were unable to distinguish between SOS and non-SOS patients (Figure 4E). In order to determine the efficacy of the 2D-SWE measure to improve the predictive value of the current score for SOS diagnosis, we used 3x2 tables as a diagnostic tool to calculate sensibility, specificity and positive predictive value of 2D-SWE combined with clinical criteria.21,25 Classical 2x2 tables (after exclusion of non-evaluable measures of 2D-SWE) and 3x2 tables were built to determine the diagnostic value of 2DSWE measures (Table 3). A worse-scenario approach was calculated, where all non-evaluable measures are classified in the wrong category (absence of SOS diagnosis for patients with SOS and positive SOS diagnosis for patients without SOS). When combined with the EBMT, Baltimore or Seattle scores, a 2D-SWE value higher than 8.1 kPa improved the ability the EBMT score to efficiently diagnose SOS, with a better sensibility (75%) and specificity (99%) for SOS diagnosis (Figure 4F). The addition of 2DSWE improved the positive predictive value of the Seattle score from 19% to 38%, of the Baltimore score from 33% to 50% and of the EBMT score from 40% to 60%.

Discussion After allo-HSCT, liver injury is a major cause of early mortality.3,4,23,24 Accurate and early diagnosis is the cornerstone for personalized treatment. This study presents a large real-life cohort of 146 consecutive adult patients who haematologica | 2021; 106(9)

Elastography in sinusoidal obstruction syndrome

Figure 3. Baseline and evolution of transient elastography and two-dimensional real-time shear wave elastography after allogeneic hematopoietic stem cell transplantation. (A) Individual baseline measure for transient elastography (TE) (5.4 kPa [range, 4.1-6.8]) and two-dimensional real-time shear wave elastography (2DSWE) (6.0kPa [range, 4.9-7.8]) with median and interquartile value. Stars represent a patient with fibrosis before transplantation. (B) No significant difference was observed between baseline value (median and interquartile range) of patients with (red) or without (black) liver involvements (unpaired t test with Holm-Sidak corrections for multiple testing). (C, and D) Sequential evolution of elastography measures compared with baseline with mean and 95% Confidence Interval (CI). No significant difference was observed between patients with (red line) or without (black line) liver involvements for TE (C) and shear wave (D) at day +7 or at day+14 (unpaired t-test with Holm-Sidak correction). For each group, difference between baseline, day+7 and day+14 were not significant (Two-way ANOVA with Dunnet correction for multiple comparison). (E and F) Comparison of elastography delta value (difference between measure and baseline) according to liver diagnosis (box represents median and interquartile values, whiskers are minimum and maximum values) for TE (E) and shear wave (F) (Two-way ANOVA with Dunnet correction multiple comparison). 2D-SWE values were significantly increased in patients with sinusoidal obstruction syndrome (SOS) by comparison with those without liver involvements.

haematologica | 2021; 106(9)

2379

P.-E. Debureaux et al.

Table 3. Analyzed results with recalculated 3x2 tables including non-evaluable measures.

3x2 table Truepositive 2D-SWE > 8.1 kPa EBMT EBMT + 2D-SWE > 8.1 kPa Baltimore Baltimore + 2D-SWE >8.1 kPa Seattle Seattle + 2D-SWE >8.1 kPa

4 6 3 4 3 5 3

2x2 table FalseFalsepositive negative 29 9 1 8 2 22 3

Non-evaluable results at patient level for 2D-SWE TrueNonNonnegative evaluable evaluable (positive) (negative)

0 0 0 2 2 1 1

97 131 135 132 132 118 131

1 NA 1 NA 1 NA 1

11 NA 1 NA 1 NA 2

Score performance Se

PPV

NPV

80% 100% 75% 67% 50% 83% 60%

71% 94% 99% 94% 98% 84% 96%

9% 40% 60% 33% 50% 19% 38%

99% 100% 99% 99% 98% 99% 99%

2D-SWE: two-dimensional real-time shear wave elastography; EBMT: European Bone Marrow Transplantation; NA: not applicable; Se: sensibility; Sp: specificity; PPV: positive predictive value; NPV: negative predictive value.

underwent allo-HSCT. In our study, elastography was significantly and specifically increased in patients who developed a SOS. In association with standard SOS scores, 2DSWE improved the specificity of these scores and their positive predictive value (thus decreasing the risk of over treating patients with expensive drugs with potential sideeffects). In this patient cohort, liver biological abnormalities were frequent after transplantation. Around 39% of patients developed a hyperbilirubinemia, an incidence similar to what has previously been described after alloHSCT.3,4 Hyperbilirubinemia has been shown to be predictive of TRM3 and liver injury is associated with a high mortality rate after transplantation.26 We previously reported that clinical and biological features are poorly predictive of liver lesions and that liver biopsy is a useful approach to improve diagnosis of liver involvements in selected patients.5 Improving accuracy of liver involvements after allo-HSCT is challenging and we currently lack reproducible and specific markers easily available for all patients. Elastography has been widely developed as a non-invasive and quantitative tool for liver involvements, especially fibrosis and cirrhosis.27 The baseline values measured in this study were quite similar to those of healthy subjects previously published for TE (5.49+/-1.59 kPa)28 or 2D-SWE (5.19+/-1.03 kPa).29 However, a significant proportion of patients had higher basal value before transplantation, especially in those with pre-existing liver involvements. In a smaller mixed cohort of patients who underwent autologous (n=37) or allo-HSCT (n=30), baseline transient elastography values could predict the occurrence of hyperbilirubinemia after transplantation.30 In addition, it has been shown that transient elastography and point shear wave values before allo-HSCT were higher in patients developing severe or life-threatening liver complications after allo-HSCT.31 Our study did not identify baseline values as a marker for subsequent liver injury after allo-HSCT, but baseline measures might help the interpretation of post-transplantation measures. This study was initially designed to explore elastography for the diagnosis of all early liver complications occurring during the first 100 days after transplantation. In our study, the liver stiffness measure was not significantly increased in patients with GvHD, DILI or infectious disease. SOS was the only complication in which 2D-SWE 2380

measures were significantly increased as compared to baseline values. Only six cases of SOS were confirmed in this cohort, a low incidence that limits the interpretation of our data on 2D-SWE measures for early diagnosis of SOS. In rat models of SOS, point shear wave velocity with ARFI was increased in animals with SOS as compared to controls and was correlated with a high SOS histological score, inflammation and congestion, but not with fibrosis.32 Inflammation or congestion due to viral hepatitis,33 biliary obstruction,34 cardiac failure,35 and acute lymphoid leukemia relapse36 have been reported to be associated with elevated TE. The TE value mainly depends on tissue stiffness with a region of interest (ROI) of 40-50 mm². By contrast, 2D-SWE measure depends on tissue stiffness and viscosity with ROI of 100 mm² or more.16 As SOS is a heterogeneous vascular liver disease characterized by progressive sinusoid vessel obstruction that could affect viscosity, it could explain why 2D-SWE was more efficient than TE to diagnose SOS. The incidence of an unreliable TE measure (> 10%) could decrease the diagnosis performance in comparison to 2D-SWE. Finally, 2DSWE was not operator-dependent in our study as in others.37 Many techniques of elastography have been proposed to predict hepatic complications30,31 or to diagnose SOS after allo-HSCT.36,38,39 In a pediatric cohort of 22 patients, sequential measures of TE after allo-HSCT showed increased TE measures 3 to 6 days before clinical signs in five SOS cases (based on Seattle or Baltimore classification).38 In a second pediatric series of 25 patients, sequential 2D-SWE velocity at day+5 and day+14 in five SOS patients was higher than in the control population.39 In the adult setting, one single center study including 78 patients after allo-HSCT evaluated the usefulness of sequential measures of TE for SOS diagnoses.36 TE was significantly increased in four patients with SOS. Currently, SOS diagnosis relies on clinical criteria that have low specificity. To our knowledge, there is no published evaluation of the sensibility, specificity or predictive value of EBMT, Seattle or Baltimore criteria, using liver biopsy as a gold standard method for SOS diagnosis.8 In our cohort, one third of patients with EBMT criteria had a liver biopsy and the systematic retrospective review of their medical history helped us to better estimate the predictive value of each clinical score.40 For patients who did haematologica | 2021; 106(9)

Elastography in sinusoidal obstruction syndrome

Figure 4. Two-dimensional real-time shear wave elastography measures improve sinusoidal obstruction syndrome diagnosis. (A and B) Venn diagram representing the distribution of patients with European Society for Blood and Marrow Transplantation (EBMT), Seattle and/or Baltimore Criteria in this cohort (A) and those with final diagnosis of sinusoidal obstruction syndrome (SOS) in each group (B). (C and D) ROC curves for day+14 two-dimensional real-time shear wave elastography (2DSWE) and transient elastography (TE) measures, and box plot of individual values (box represents median and interquartile values, whiskers are minimum and maximum values) in patients with (blue) or without SOS (yellow) for shear wave (A) and TE (B) compared with Mann-Whitney U test. (E) Radar plot of individual ultrasound criteria and box plot of ultrasonography scores (calculated by adding one point per criteria) for patients with no EBMT criteria (gray lines) and with EBMT criteria without (orange lines) or with SOS diagnosis (blue lines) compared with Kruskal Wallis test and Dunn’s correction for multiple testing. (F) Representation of sensibility, specificity, and positive predictive rate (PPR) (color circle, large and red circle are associated with greatest PPR). Addition of 2D-SWE measure with a cutoff value at 8.1 kPa improved the PPR value of the Seattle score from 19% to 38%, of the Baltimore score from 33% to 50% and of the EMBT score from 40% to 60%.

haematologica | 2021; 106(9)

2381

P.-E. Debureaux et al.

not have a biopsy, the final diagnosis relied on a retrospective analysis of all medical data, of clinical evolution with treatments, and on consensual diagnosis criteria if available. However, the lack of gold standard criteria for most of diagnoses, such as GvHD, SOS or DILI, could bias conclusions as there is currently no reliable tool to avoid misdiagnoses. An unreliable TE or 2D-SWE value is classically estimated to be included between 5% and 10%, a range very similar to what was observed in our cohort.41 Using an intention to diagnose approach helped us to estimate the suitability of the liver stiffness measurement in addition to clinical criteria for SOS diagnosis. Our results confirm the interest of liver stiffness measures and strongly suggest that 2D-SWE significantly improved the positive predictive value of clinical scores for SOS diagnosis after allo-HSCT. Increasing the positive predictive value of current criteria for SOS is critical to avoid useless and potentially toxic treatment in patient with comorbidities. Thus, 2D-SWE appears as a promising, non-invasive, quantitative, safe, and reproducible technique allowing an

References 1. Gooley TA, Chien JW, Pergam SA, et al. Reduced mortality after allogeneic hematopoietic-cell transplantation. N Engl J Med 2010;363(22):2091-2101. 2. McDonald GB. Hepatobiliary complications of hematopoietic cell transplantation, 40 years On. Hepatology 2010;51(4):14501460. 3. Gooley TA, Rajvanshi P, Schoch HG, McDonald GB. Serum bilirubin levels and mortality after myeloablative allogeneic hematopoietic cell transplantation. Hepatology 2005;41(2):345-352. 4. Hogan WJ. Hepatic injury after nonmyeloablative conditioning followed by allogeneic hematopoietic cell transplantation: a study of 193 patients. Blood 2004; 103(1):78-84. 5. Ruggiu M, Bedossa P, Rautou PE, et al. Utility and safety of liver biopsy in patients with undetermined liver blood test anomalies after allogeneic hematopoietic stem cell transplantation: a monocentric retrospective cohort study. Biol Blood Marrow Transplant 2018;24(12):2523-2531. 6. Lassau N, Leclère J, Auperin A, et al. Hepatic veno-occlusive disease after myeloablative treatment and bone marrow transplantation: value of gray-scale and Doppler US in 100 patients. Radiology 1997;204(2):545-552. 7. Nishida M, Kahata K, Hayase E, et al. Novel Ultrasonographic scoring system of sinusoidal obstruction syndrome after hematopoietic stem cell transplantation. Biol Blood Marrow Transplant J Am Soc Blood Marrow Transplant 2018; 24(9):1896-1900. 8. EASL Clinical Practice Guidelines: vascular diseases of the liver. J Hepatol 2016; 64(1):179-202. 9. Teefey SA, Brink JA, Borson RA, Middleton WD. Diagnosis of venoocclusive disease of the liver after bone marrow transplantation: value of duplex sonography. AJR Am J Roentgenol 1995;164(6):1397-1401. 10. McCarville MB, Hoffer FA, Howard SC, Goloubeva O, Kauffman WM. Hepatic veno-occlusive disease in children undergoing bone-marrow transplantation: usefulness of sonographic findings. Pediatr Radiol

2382

early and accurate diagnosis of SOS. 2D-SWE measures combined with classical scores such as Baltimore or EBMT criteria could significantly help discriminating SOS from other post-transplantation early liver injury. Prospective multi-center trials would be necessary to confirm these results and to broadly evaluate the impact of liver stiffness measures on management and treatment of liver involvements after allo-HSCT. Disclosures No conflicts of interest to disclose. Contributions PED and DM collected, analyzed data and performed statistics analysis; PB, AMZ and MDB performed ultrasonography, Doppler and elastrography measurements; PER, SP, ASG, MR, RPL, AP, FSF, AX, PHP and DV provided data and commented manuscript; PED, GS and DM wrote the manuscript; GS and DM conceived the project and supervised the work; all authors approved the manuscript.

2001;31(2):102-105. 11. Shiina T, Nightingale KR, Palmeri ML, et al. WFUMB guidelines and recommendations for clinical use of ultrasound elastography: part 1: basic principles and terminology. Ultrasound Med Biol 2015;41(5):11261147. 12. European Association for Study of Liver, Asociacion Latinoamericana para el Estudio del Higado. EASL-ALEH Clinical Practice Guidelines: non-invasive tests for evaluation of liver disease severity and prognosis. J Hepatol 2015;63(1):237-264. 13. Dietrich CF, Trenker C, Fontanilla T, et al. New ultrasound techniques challenge the diagnosis of sinusoidal obstruction syndrome. Ultrasound Med Biol 2018; 44(11):2171-2182. 14. Bacigalupo A, Ballen K, Rizzo D, et al. Defining the intensity of conditioning regimens: working definitions. Biol Blood Marrow Transplant J Am Soc Blood Marrow Transplant 2009;15(12):1628-1633. 15. Armand P, Gibson CJ, Cutler C, et al. A disease risk index for patients undergoing allogeneic stem cell transplantation. Blood 2012;120(4):905-913. 16. Dietrich CF, Bamber J, Berzigotti A, et al. EFSUMB guidelines and recommendations on the clinical use of liver ultrasound elastography, update 2017 (long version). Ultraschall Med Stuttg Ger 1980 2017; 38(4):e16-e47. 17. Glucksberg H, Storb R, Fefer A, et al. Clinical manifestations of graft-versus-host disease in human recipients of marrow from HL-A-matched sibling donors. Transplantation 1974;18(4):295-304. 18. McDonald GB, Hinds MS, Fisher LD, et al. Veno-occlusive disease of the liver and multiorgan failure after bone marrow transplantation: a cohort study of 355 patients. Ann Intern Med 1993;118(4):255-267. 19. Jones RJ, Lee KS, Beschorner WE, et al. Venoocclusive disease of the liver following bone marrow transplantation. Transplantation 1987;44(6):778-783. 20. Mohty M, Malard F, Abecassis M, et al. Revised diagnosis and severity criteria for sinusoidal obstruction syndrome/venoocclusive disease in adult patients: a new classification from the European Society for Blood and Marrow Transplantation. Bone

Marrow Transplant 2016;51(7):906-912. 21. Cohen JF, Korevaar DA, Altman DG, et al. STARD 2015 guidelines for reporting diagnostic accuracy studies: explanation and elaboration. BMJ Open 2016; 6(11): e012799. 22. Andrade RJ, Aithal GP, Björnsson ES, et al. EASL clinical practice guidelines: druginduced liver injury. J Hepatol 2019; 70(6):1222-1261. 23. Schuetz GM, Schlattmann P, Dewey M. Use of 3x2 tables with an intention to diagnose approach to assess clinical performance of diagnostic tests: meta-analytical evaluation of coronary CT angiography studies. BMJ 2012;345e6717. 24. McDonald GB, Sharma P, Matthews DE, Shulman HM, Thomas ED. Venocclusive disease of the liver after bone marrow transplantation: diagnosis, incidence, and predisposing factors. Hepatol Baltim Md 1984;4(1):116-122. 25. Schuetz GM, Schlattmann P, Dewey M. Use of 3×2 tables with an intention to diagnose approach to assess clinical performance of diagnostic tests: meta-analytical evaluation of coronary CT angiography studies. BMJ;345. 26. Sakai M, Strasser SI, Shulman HM, McDonald SJ, Schoch HG, McDonald GB. Severe hepatocellular injury after hematopoietic cell transplant: incidence, etiology, and outcome. Bone Marrow Transplant 2009;44(7):441-447. 27. Dighe M, Bruce M. Elastography of diffuse liver diseases. Semin Roentgenol 2016; 51(4):358-366. 28. Roulot D, Czernichow S, Le Clésiau H, Costes J-L, Vergnaud A-C, Beaugrand M. Liver stiffness values in apparently healthy subjects: influence of gender and metabolic syndrome. J Hepatol 2008;48(4):606-613. 29. Petzold G, Hofer J, Ellenrieder V, Neesse A, Kunsch S. Liver stiffness measured by 2dimensional shear wave elastography: prospective evaluation of healthy volunteers and patients with liver cirrhosis. J Ultrasound Med Off J Am Inst Ultrasound Med 2019;38(7):1769-1777. 30. Auberger J, Graziadei I, Clausen J, Vogel W, Nachbaur D. Non-invasive transient elastography for the prediction of liver toxicity following hematopoietic SCT. Bone

haematologica | 2021; 106(9)

Elastography in sinusoidal obstruction syndrome

Marrow Transplant 2013;48(1):159-160. 31. Karlas T, Weiße T, Petroff D, et al. Predicting hepatic complications of allogeneic hematopoietic stem cell transplantation using liver stiffness measurement. Bone Marrow Transplant 2019;1-9. 32. Park SH, Lee SS, Sung J-Y, et al. Noninvasive assessment of hepatic sinusoidal obstructive syndrome using acoustic radiation force impulse elastography imaging: a proof-of-concept study in rat models. Eur Radiol 2018;28(5):2096-2106. 33. Arena U, Vizzutti F, Corti G, et al. Acute viral hepatitis increases liver stiffness values measured by transient elastography. Hepatology 2007;47(2):380-384. 34. Millonig G, Reimann FM, Friedrich S, et al. Extrahepatic cholestasis increases liver stiffness (FibroScan) irrespective of fibrosis. Hepatology 2008;48(5):1718-1723. 35. Colli A, Pozzoni P, Berzuini A, et al.

haematologica | 2021; 106(9)

Decompensated chronic heart failure: increased liver stiffness measured by means of transient elastography. Radiology 2010;257(3):872-878. 36. Colecchia A, Ravaioli F, Sessa M, et al. Liver stiffness measurement allows early diagnosis of veno-occlusive disease/sinusoidal obstruction syndrome in adult patients who undergo hematopoietic stem cell transplantation: results from a monocentric prospective study. Biol Blood Marrow Transplant 2019;25(5):995-1003. 37. Poynard T, Pham T, Perazzo H, et al. Realtime shear wave versus transient elastography for predicting fibrosis: applicability, and impact of inflammation and steatosis. a non-invasive comparison. PLOS ONE 2016;11(10):e0163276. 38. Colecchia A, Marasco G, Ravaioli F, et al. Usefulness of liver stiffness measurement in predicting hepatic veno-occlusive dis-

ease development in patients who undergo HSCT. Bone Marrow Transplant 2017;52 (3):494-497. 39. Reddivalla N, Robinson AL, Reid KJ, et al. Using liver elastography to diagnose sinusoidal obstruction syndrome in pediatric patients undergoing hematopoetic stem cell transplant. Bone Marrow Transplant 2018;1-8. 40. Volin L, Niittyvuopio R, Heiskanen J, et al. Diagnosis of veno-occlusive disease/sinusoidal obstruction syndrome of the liver: problems of interpretation. Bone Marrow Transplant 2016;51(12):1633-1635. 41. Kim DW, Suh CH, Kim KW, Pyo J, Park C, Jung SC. Technical performance of twodimensional shear wave elastography for measuring liver stiffness: a systematic review and meta-analysis. Korean J Radiol 2019;20(6):880-893.

2383

ARTICLE Ferrata Storti Foundation

Myeloproliferative Disorders

Altered T-cell subset repertoire affects treatment outcome of patients with myelofibrosis Ivo Veletic,1,* Sanja Prijic,1,2,* Taghi Manshouri,1 Graciela M. Nogueras-Gonzalez,3 Srdan Verstovsek,1 and Zeev Estrov1 1

Haematologica 2021 Volume 106(9):2384-2396

Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 2Clinical Department of Laboratory Diagnostics, University Hospital Center Zagreb, Zagreb, Croatia and 3Department of Biostatistics, The University of Texas MD Anderson Cancer Center, Houston, TX, USA *

IV and SP contributed equally to the study as co-first authors.

ABSTRACT

Correspondence: ZEEV ESTROV zestrov@mdanderson.org Received: February 6, 2020. Accepted: July 16, 2020. Pre-published: July 30, 2020. https://doi.org/10.3324/haematol.2020.249441

2384

henotypic characterization of T cells in myelofibrosis is intriguing because of increased inflammation, markedly elevated pro-inflammatory cytokines, and altered distribution of T-cell subsets. Constitutive activation of Janus kinase-2 (JAK2) in the majority of patients with myelofibrosis contributes to the expression of the programmed cell death protein-1 (PD1) and T-cell exhaustion. We wondered whether T-cell activation affects treatment outcome of patients with myelofibrosis and sought to determine whether the JAK1/2 inhibitor ruxolitinib affects the activation of T-cell subsets. T cells from 47 myelofibrosis patients were analyzed and the percentages of either helper (CD4+) or cytotoxic (CD8+) naïve, central memory, effector memory, or effector T cells; and fractions of PD1-expressing cells in each subset were assessed. Higher numbers of T cells co-expressing CD4/PD1 and CD8/PD1 were found in myelofibrosis patients than in healthy controls (n=28), and the T cells were significantly skewed toward an effector phenotype in both CD4+ and CD8+ subsets, consistent with a shift from a quiescent to an activated state. Over the course of ruxolitinib treatment, the distribution of aberrant T-cell subsets significantly reversed towards resting cell phenotypes. CD4+ and CD8+ subsets at baseline correlated with monocyte and platelet counts, and their PD1+ fractions correlated with leukocyte counts and spleen size. Low numbers of PD1+/CD4+ and PD1+/CD8+ cells were associated with complete resolution of palpable splenomegaly and improved survival rate, suggesting that low levels of exhausted T cells confer a favorable response to ruxolitinib treatment.

Introduction Primary or secondary myelofibrosis (MF) is characterized by a significant immune deregulation.1,2 In the vast majority of patients with MF, Janus kinase (JAK)-2 is constitutively activated.3 As a result, MF neoplastic cells produce high levels of inflammatory cytokines and pentraxins that contribute to the induction of progressive bone marrow (BM) fibrosis, debilitating constitutional symptoms, and poor prognosis in MF patients.4,5 Cytokines, such as interleukin (IL)-1, IL-6 and IL8, modulate T-cell activation and immune function through the activation of JAK2 and its downstream signal transducer and activator of transcription (STAT) pathways.6,7 The JAK1/2 inhibitor ruxolitinib alleviates constitutional symptoms in MF patients, primarily by profound suppression of inflammation.8 Although it is known that JAK-STAT signaling modifies T-helper cell activity and inflammatory responses and JAK1/2 inhibition impairs the cytotoxic function of T cells in vitro,911 the effects of aberrant JAK2 signaling and its modulation of T cells in patients with MF remain elusive. A few recent studies showed increased T-cell response to neo-antigens in patients with myeloproliferative neoplasms.12-15 However, persistent tumorinduced activation prompts T cells to enter a dysfunctional state, referred to as T-

haematologica | 2021; 106(9)

T cells affect the prognosis of myelofibrosis

cell exhaustion.16 Programmed cell death protein 1 (PD1) is an essential T-cell checkpoint receptor that is overexpressed as T cells undergo persistent activation, thus providing a pathway to control T-cell response.17 In several neoplasms, PD1 was found to play a critical role in regulating T-cell function.18 Recent studies showed that myeloid cells from subjects with myeloproliferative neoplasms express high levels of PD1 ligand 1 (PDL1),19,20 and that the overexpression of PDL1 is induced by constitutively activated JAK2.19 Elevated levels of PDL1 on tumor cells frequently correlate with increased JAK2 activity,21 and high PDL1 expression confers sensitivity to anti-PD1 therapy.22 Conversely, decreased PDL1 levels were detected in rare solid tumors harboring loss-of-function JAK2 mutations, and have been associated with poor outcomes of PD1/PDL1 blockade therapy.23 Because JAK2 is constitutively activated in MF, we sought to determine whether the T-cell activation status is altered in MF patients and whether and how it is affected by JAK inhibitor therapy. We also sought to elucidate the relationship between T-cell activation and exhaustion given the implications of this association in treating MF patients, developing immune-harnessing MF strategies, and rationally guiding clinical trials. To achieve these goals, we systematically analyzed the activation status of peripheral T-cell subsets of patients with MF at baseline and over the course of treatment with ruxolitinib. Furthermore, we tested the association of PD1-co-expressing helper (CD4+) and cytotoxic (CD8+) T-cell subpopulations with disease progression and assessed the effect of PD1+ T-cell fractions on the clinical outcome of MF patients.

Methods Specimen assessed in this study We obtained corresponding BM and peripheral blood (PB) specimens from 47 patients with MF who were enrolled in a phase I/II clinical trial of ruxolitinib at the University of Texas MD Anderson Cancer Center (MDACC) (ClinicalTrials.gov identifier, NCT00509899) between June 2007 and April 2015.24 Specimens were collected prior to treatment and for up to 7 years after treatment, once Institutional Review Board (IRB)approved informed consent had been obtained. The patients were diagnosed with primary MF, post-essential thrombocythemia MF or post-polycythemia vera MF; the diagnoses were established in accordance with the 2008 World Health Organization classification.25 The patients did not receive antineoplastic medications for at least 14 days before starting treatment. Ruxolitinib was administered orally (10-25 mg twice a day or 50-200 mg once a day) according to the clinical trial protocol, which was designed to assess the efficacy and safety of ruxolitinib. For control studies, PB specimens were obtained from 28 healthy individuals (13 males and 15 females) agematched with MF patients. The clinical and laboratory research was conducted in accordance with the Declaration of Helsinki and approved by the MDACC IRB1 committee (protocol identities LAB01-473 and LAB05-0321).

Cell fractionation and immunostaining To detect and quantify cell surface proteins, we performed multiparameter flow cytometry analyses of low-density cells that were previously stored in dimethylsulfoxide. Briefly, BM and PB specimens were collected into Vacutainer tubes contain-

Figure 1. T-cell subset gating strategy. Representative dot plots demonstrating the flow cytometry gating strategy that was used to analyze individual T-cell subsets. (A) First, leukocytes were isolated based on CD45 (PE-CF594) positivity and side scatter area (SSC-A). (B) Singlets were then gated using forward scatter area (FSCA) and height (FSC-H). (C) Within the singlet leukocyte population, lymphocytes were defined based on FSC-A and SSC-A parameters. (D) Subsequently, T cells were separated from B cells and natural killer (NK) cells based on positivity for CD3 (APC) and HLA-DR (PerCP-Cy5.5). (E) Gamma delta (γ/δ) T cells were excluded from further analysis using T-cell receptor gamma delta (TCRγ/δ; PE-Cy7) and SSC-A. (F) Alpha beta (α/β) T cells were further gated to differentiate between CD4+ and CD8+ T cells based on CD4 (V500) and CD8 (APC-H7). (G) Naïve (T ), central memory (T ), effector memory (T ) and effector T cells (T ) of CD4/CD8 subsets were obtained using quadrant gates based on CD45RO (FITC) and CD62L (BV421). (H) Cells positive for programmed cell death 1 (PD1) were obtained using density plots for PD1 (PE) with the corresponding isotype control as reference. Red dots depict T cells within each of the previously defined gates; only the CD4+ subset is displayed in (G) and (H). N

haematologica | 2021; 106(9)

EFF

2385

I. Veletic et al.

ing sodium heparin (BD, Franklin Lakes, NJ, USA). Low-density cells were fractionated using Ficoll-Histopaque 1077 (SigmaAldrich, St. Louis, MO, USA), washed in phosphate-buffered saline (Invitrogen, Carlsbad, CA, USA), spun down, dissolved in 10% dimethylsulfoxide solution (Sigma-Aldrich, St. Louis, MO, USA) supplemented with fetal bovine serum (Invitrogen, Carlsbad, CA, USA), and frozen in liquid nitrogen. Prior to flow cytometry analysis, cells were thawed, washed and re-suspended in fetal bovine serum. After trypan blue viability assessment, live cells (106) were incubated with the appropriate antibodies or their corresponding isotype controls, and their cell surface protein expression was assessed using the Gallios multichannel flow cytometer (Beckman Coulter, Brea, CA, USA). The antibodies used and their isotype controls are listed in Online Supplementary Table S1.

Flow cytometry analysis of T cells A universal gating strategy was applied to identify individual T-cell subsets. Singlet lymphocytes in the CD45+ cell population were identified based on size and lack of granularity (Figure 1AC). Subsequently, T cells were gated by using anti-CD3 and antiHLA-DR antibodies and further subdivided using anti-TCR γ/δ antibodies (Figure 1D and E). The CD4+ and CD8+ subpopulations of the α/β + T cells were further separated into naïve (T ), central memory (T ), effector (T ), and effector memory (T ) subsets, using anti-CD62L and anti-CD45RO antibodies (Figure 1F and G). The percent of PD1+ T cells was assessed in each subset (Figure 1H). All flow cytometry data were analyzed using FlowJo software v10.5 (Treestar, San Carlos, CA, USA). N

EFF

corrplot v0.84, ggplot2 v3.3.0, ggpubr v0.2.5, ggeffects v0.14.3, and survminer v0.4.6.

Data sharing statement De-identified original data are available from the corresponding author (zestrov@mdanderson.org).

Results Clinical features of myelofibrosis patients and outcome of ruxolitinib treatment To evaluate the distribution and characteristics of T-cell subsets in patients with MF, we analyzed PB and BM specimens of 47 MF patients (27 with PMF, 13 with post-polycythemia vera MF, and 7 with post-essential thrombocythemia MF) and 28 agematched healthy controls (Online Supplementary Table S2). The median daily dose of ruxolitinib was 50 mg (range, 20-200 mg) at the start of the clinical trial. The dose was reduced because of anemia and/or thrombocytopenia in seven patients. The median duration of treatment was 38.9 months. Among the patients who had their specimens analyzed in this study, two discontinued treatment because of myelosuppression and four because of transformation to acute myeloid leukemia. Infection (pneumonia and/or sepsis) was the cause of death in five of the 16 (31.3%) patients who died while on trial.

Analysis of myelofibrosis patients’ T-cell subsets The percentages of CD4+ and CD8+ cells and their T , T , T , and T subsets were assessed in BM or PB specimens from 41 MF patients and 28 healthy individuals. Because analyses of T-cell subsets using PB (n=35) and BM (n=16) specimens from the same MF patients revealed similar results (Online Supplementary Figure S3), we have not presented the data separately. Whereas CD4+ and CD8+ cell distributions in MF patients were not different from those in healthy individuals (Figure 2Ai and Bi), marked differences were found in both CD4-derived (Figure 2Aii and Bii) and CD8-derived (Figure 2Aiii and Biii) T-cell subsets. We detected a 2.93-fold and a 3.45-fold (P<0.001 for both) reduction in the number of T cells, and a 3.45-fold and a 4.03-fold (P<0.001 for both) reduction in T cells within the CD4+ and CD8+ cell subsets, respectively, in MF-derived T cells as compared to normal controls. Conversely, we detected an increase in the number of T cells within both CD4+ and CD8+ cell subsets (mean fold changes, 2.75 and 1.86, respectively; P<0.001 for both), and in the number of T cells within the CD4+ cell fraction (mean fold change, 1.51; P=0.005) but not within the CD8+ cell fraction. Whereas CD4+ and CD8+ resting subsets (T and T ) in MF patients correlated significantly and positively with one another, two effector subsets (T and T ) exhibited negative correlation between both one another and the resting subsets (Figure 2C), indicating that one effector population prevails within each patient’s CD4 or CD8 subset. Overall, the increase in effector T-cell phenotype suggests that in patients with MF T cells shift from a quiescent to an activated state. Compared to CD4+, MF CD8+ T cells shift more towards a terminally activated state, suggesting a predominant effector-mediated cytotoxic response in MF. N

EFF

Statistical analyses The Student t-test was used to assess whether T-cell subsets of normal individuals were significantly different from those of patients with MF. A paired t-test was used to determine whether ruxolitinib treatment significantly affected T-cell subset distribution. A linear mixed-effects model with repeated measures was developed to determine whether there were differences in T-cell subsets at sequential time-points. In order to correct for clinical response or progression over time, several model specifications that included clinical variables obtained at the time of sample collection were compared using mean Akaike and Bayesian information criteria and R-squared values, and the best performing model was selected for use in the longitudinal analyses. Significance of overall change in time and each predictor were assessed using the Kenward-Roger adjusted F-test. Correlations between continuous clinical variables and T-cell subsets were assessed using the Pearson coefficient and between-group differences were calculated using the Welch t-test. The percentage of each T-cell subset was dichotomized into high and low groups using the optimal cutoff value of maximally selected rank statistics. The patients’ overall survival was estimated by the KaplanMeier method and a log-rank test was used to compare the survival probabilities. A univariate Cox proportional hazard regression model was fitted to assess the association between clinical variables and overall survival. To assess the predictive value of T-cell subsets, a multivariate Cox proportional hazard model was applied, adjusted for the clinical variables that were found to be significant in the univariate analyses. The Wald test was used to assess the significance of each covariate in Cox models. Statistical analysis was performed using Stata/SE v15.1 (Stata Corp, College Station, TX, USA) and R v3.6.3 (R Foundation for Statistical Computing, Vienna, Austria) statistical software with tidyverse v1.3.0, lme4 v1.1-23, pbkrtest v0.4-8.6, and survival v3.1-8 packages. Graphs were created using GraphPad Prism v7.03 (GraphPad Software, La Jolla, CA, USA) and R packages

2386

EFF

Long-term effects of ruxolitinib treatment on T-cell subsets Because treatment with ruxolitinib reduces plasma levels of cytokines and chemokines and significantly reduces spleen size in most MF patients,24,26,27 we sought to assess the effect of ruxolitinib treatment on the distribution of T-cell subsets. Analysis of the corresponding PB or BM specimens obtained from 25 MF patients before and during ruxolitinib treatment demonstrated an overall shift towards a CD8+ phenotype over the course of

haematologica | 2021; 106(9)

T cells affect the prognosis of myelofibrosis

EB i

iii

C Figure 2. Distribution of T-cell subsets in myelofibrosis. T-cell subsets from patients with myelofibrosis (MF; n=41) and healthy controls (CTRL; n=28) were analyzed by flow cytometry. (A) Representative dot plots from a single MF patient and CTRL are depicted. Minimal difference was observed in CD4+ and CD8+ differentiation subsets within the total T cells (i). In contrast, a significant shift in distribution towards effector populations was seen among activation subsets within CD4+ (ii) and CD8+ (iii) T-cell populations. Whereas the numbers of naïve (T ) and central memory T cells (T ) were reduced, effector memory (T ) and effector T cells (T ) were significantly increased, apart from CD8+ T cells. (B) Quantification of differentiation subsets (i), and CD4+ (ii) and CD8+ (iii) activation subsets from MF patients (red) and CTRL (gray). (C) Correlation matrix showing significant colorcoded relationships (lower half), Pearson coefficients (r), and P-values (upper half) between each pair of Tcell subsets. Numbers in dot plots denote percent of cells per gate. Bars represent means with standard deviation. The Student t-test was used to compare the two groups. The P-values in the correlation matrix were adjusted by the Benjamini-Hochberg method. P<0.05 was considered statistically significant. PB, peripheral blood; n.s., not significant. N

EFF

haematologica | 2021; 106(9)

2387

I. Veletic et al.

time (Figure 3). As shown in a representative patient after 2 years of treatment (Figure 3Ai) and in all patients over the mean treatment period (Figure Bi), the percent of CD4+ cells decreased, whereas the percent of CD8+ cells significantly increased (mean differences -5.1% vs. 13.9%, P=0.042 and P=0.025, respectively). Within the CD4+ cell subsets (Figure 3Aii and Bii), we observed a 2-fold increase in the percent of T and T cells (mean fold changes 2.05 and 2.26, respectively; P<0.001 for both), and 1.2N

iii

fold (P=0.01) and 1.48-fold (P<0.001) decreases in the percent of T and T cells, respectively, during ruxolitinib treatment. Comparable effects were observed within the CD8+ cell subsets (Figure 3Aiii and Biii): the percent of T and T cells increased by 2.22-fold and 1.64-fold (P=0.002 and P=0.03, respectively), the percent of CD8+ T cells decreased by 1.21-fold (P=0.001), whereas the percent of CD8+ T cells remained unaffected by ruxolitinib treatment. EM

EFF

Figure 3. Effects of ruxolitinib treatment on T-cell subsets of patients with myelofibrosis. T-cell subsets from patients with myelofibrosis (MF) (n=25) were analyzed before and over the course of treatment with ruxolitinib. (A) Representative dot plots from a MF patient at treatment baseline (Bsl) and after 2 years of treatment (Ruxo). Naïve (T ) and central memory T cells (T ) were increased, whereas effector memory (T ) and effector T cells (T ) were reduced. Only CD4+ T cells did not change significantly with treatment as compared to baseline. (B) Quantification of CD4/CD8 differentiation subsets (i) and activation subsets (ii-iii) at treatment baseline (red) and during the overall follow-up period (light blue). Follow-up values were calculated as a mean of all the analyzed timepoints over the course of treatment for each patient. (C) Longitudinal analysis of T-cell subsets over 6 years of ruxolitinib treatment. The figures shows the mean percentages (green) and linear predictions (dark blue) based on linear mixed-effects model with repeated measures in differentiation (i) and activation subsets (ii-iii). Numbers in dot plots denote the percent of cells per gate. A paired t-test was used to compare the two groups. In the longitudinal plots error bars denote the standard error, P-values represent the statistical significance of change from baseline over time, and asterisks indicate the significance of change in each year of treatment. Linear mixed models were corrected for spleen size, grade of bone marrow fibrosis, and JAK2V617F allele burden. The Pvalues were computed using the KenwardRoger adjusted F-test. P values <0.05 were considered statistically significant. *P<0.05; **P<0.01; ***P<0.001. PB: peripheral blood; BM: bone marrow; n.s.: not significant. N

EFF

iii

2388

haematologica | 2021; 106(9)

T cells affect the prognosis of myelofibrosis

iii

Figure 4. PD1-expressing fractions of T-cell subsets in myelofibrosis. The fractions of PD1+ cells were assessed within differentiation and activation T-cell subsets from patients with myelofibrosis (MF; n=41) and healthy controls (CTRL; n=28). (A) Representative flow cytometry dot plots from a MF patient and a CTRL are depicted. PD1+ T-cell fractions were increased across both CD4/CD8 differentiation subsets (i) and all CD4+ (ii) and CD8+ (iii) activation subsets, except CD4+ naïve T cells (T ). (B) Quantification of PD1+ cell fractions in CD4+ and CD8+ T cells (i) and activation subsets within the CD4-derived (ii) and the CD8derived (iii) subsets in MF patients and CTRL subjects. (C) Correlation matrix showing significant colorcoded relationships (lower half), Pearson coefficients (r), and P-values (upper half) between each pair of PD1+ T-cell subsets. Violet dots depict PD1+ cells; numbers in dot plots denote percent of cells per gate. Bars represent means with standard deviation. The Student t-test was used to compare the two groups. The P-values in the correlation matrix were adjusted by the Benjamini-Hochberg method. P values <0.05 were considered statistically significant. T : naïve T cells; T : central memory T cells; T : effector memory T cells; T : effector T cells; BM: bone marrow; PB, peripheral blood; n.s.: not significant. N

EFF

haematologica | 2021; 106(9)

2389

I. Veletic et al.

To investigate dynamic changes in T-cell surface marker expression over the course of ruxolitinib treatment, consecutive cell surface marker profiles were analyzed using mixed linear models with repeated measures. To account for the progressive nature of MF, our final model also included spleen size, grade of BM fibrosis, and JAK2V617F allele burden (Online Supplementary Tables S3 and S4). After correcting for these variables, we observed no significant change in CD4+ and CD8+ T cells over the course of ruxolitinib treatment (Figure 3Ci), suggesting that the increase in cytotoxic T cells that we observed over the whole treatment period is a result of disease progression rather than an effect of JAK inhibition. In contrast, after correction we still observed time-dependent shifts from effector to resting Tcell subsets (Figure 3Cii-iii), confirming our hypothesis that long-term ruxolitinib treatment mitigates T-cell overactivation. Whereas significant T and T cell increases were observed in the second, third and fifth years of therapy, in both CD4+ and CD8+ subsets T cells consistently decreased over the same period. Similar changes were also observed in the fourth year of treatment, although they reached statistical significance only in the T and CD4+ T subsets. Remarkably, both T subsets showed no significant change during treatment, except for the CD8+ subset during the fifth year of therapy, suggesting that long-term ruxolitinib treatment prevents terminal activation of T cells in MF, but has little effect on the effector memory arm of T-cell activation. To determine whether baseline distributions of T-cell differentiation and activation subsets affect the overall survival, datasets were further analyzed using the Kaplan-Meier method and no significant differences were found (Online Supplementary Figure S1). In summary, these data suggest that ruxolitinib treatment shifts the activation state of T-cell subsets from terminal effector towards resting phenotype in a timedependent manner. N

EFF

PD1-expressing fractions within the T-cell subsets of myelofibrosis patients Because it was recently reported that MF myeloid cells express high levels of PDL1,19 we sought to evaluate PD1expressing fractions within T-cell subsets of MF patients. The proportion of cells co-expressing PD1 in CD4+ and CD8+ T cells of MF patients (n=35) was higher by 55.9% (P=0.028) and 86.8% (P=0.001), respectively, compared to T cells of healthy controls (n=28) (Figure 4Ai and Bi). Specifically, PD1+ fractions were increased within both CD4+ and CD8+ T , T , and T cells (mean fold-changes, 1.49, 2.97, and 3.05 in CD4+ cells; 1.77, 2.64, and 2.83 in CD8+ cells, respectively; P=0.013 in CD4+ T , P<0.001 in the rest), and within CD8+ T cells (mean foldchange, 1.74; P=0.007) (Figure 4Aii-iii and Bii-iii). Importantly, most PD1+ fractions correlated positively between one another (Figure 4C) while no significant correlation was observed with any of the T-cell subsets, suggesting that PD1+ cells are prevalent among MF T cells irrespective of their differentiation or activation state. In addition, we analyzed how ruxolitinib affects PD1+ CD4/CD8 and activation subsets over the whole follow-up period and in each year of treatment, corrected for the parameters of disease progression (spleen size, BM fibrosis grade, and JAK2V617F allele burden). Overall, no significant differences were observed in PD1+ fractions over the course of ruxolitinib treatment (Online Supplementary Figure S2). CM

EFF

Association between T-cell subsets, PD1+ fractions, and disease progression 28

Because MF is a progressive myeloproliferative neoplasm, and T cells are known to interact with clonal neoplastic cells,29 we analyzed the correlation between T-cell subsets and PB cell counts of untreated MF patients (n=41). We found that the number of CD8+

2390

cells correlated positively with monocyte counts and negatively with platelet counts (r=0.317 and r=-.335; P=0.043 and P=0.032, respectively); however, CD4+ cell subsets had a negative and positive correlation with monocyte and platelet counts (r=-0.371 and r=0.375, respectively; P=0.017 and P=0.016, respectively) (Figure 5Ai). Given that both monocytosis and thrombocytopenia are associated with disease progression,30 it is likely that CD8+ cells expand with disease propagation in untreated MF patients. Conversely, increased PD1 levels of both CD4+ and CD8+ cells correlated with total leukocyte counts (r=0.628 and r=0.547, respectively; P<0.001 for both) and palpable spleen size (r=0.435 and r=0.465; P=0.005 and P=0.002, respectively Figure 5Aii), suggesting that the increase in PD1+ T-cell fractions, typically associated with T-cell exhaustion, correlates with disease progression, regardless of PD1 distribution across those T-cell subsets. To investigate the effect of disease progression on subset levels at baseline and following treatment with ruxolitinib, we stratified patients based on spleen size, BM fibrosis grade, and JAK2V617F allele burden, and compared their total, T and PD1+ subsets, using healthy controls as a reference (Figure 5B). Although we found a 29.5% larger CD8+ T-cell population in MF patients with a palpable spleen larges than 20 cm at treatment baseline (n=11), this effect did not reach statistical significance (P=0.087). Interestingly, however, we also found 17.1% fewer CD4+ cells in this group of patients than in the control group (P=0.027). Moreover, MF patients with advanced-stage disease prior to treatment did not exhibit the significant repolarization of CD4/CD8 populations over time shown by patients with early-stage disease, further indicating that CD8 predominance is not a ruxolitinib effect but a result of disease progression. MF patients with splenomegaly greater than 20 cm had 1.4-fold larger baseline CD4+ T subsets and CD4+ PD1+ fractions (P=0.045 and P=0.029, respectively) compared to patients with smaller spleens. Of note, both PD1+ CD4+ and CD8+ subsets of these patients were significantly higher than normal (P=0.003 and P=0.013, respectively), similar to patients with MF-3 grade fibrosis (P=0.042 and P<0.001, respectively) and patients with mutant JAK2 allele burden above 50% (P=0.038 and P=0.002, respectively). Overall, the CD8+ T subset showed little difference based on the analyzed parameters of disease progression, supporting the idea that CD8+ resting cells in MF rapidly transit to T-effectors as they become activated. Remarkably, patients with high mutant JAK2 allele burden had significantly lower numbers of CD8+ T (P=0.022). EM

Association between T-cell subsets, PD1+ fractions, and clinical response to ruxolitinib Because a reduction in spleen size is typically associated with a good response to ruxolitinib and favorable treatment outcome,24,26,27 we tested the association between pretreatment T-cell subsets and spleen size 6 months into therapy. We found that complete resolution of palpable splenomegaly was associated with an increased percent of CD4+ cells and a decreased percent of CD8+ cells (mean differences 14.4% and -23%; P=0.038 and P=0.049, respectively) (Figure 6A). Furthermore, complete resolution of palpable splenomegaly was associated with a low percent of PD1+ fractions in both CD4+ and CD8+ cell subsets (mean differences, -30.7% and -31.7%; P=0.012 and P=0.036, respectively) (Figure 6B), suggesting that MF patients with low levels of exhausted (PD1+) T cells likely respond favorably to ruxolitinib treatment.

Effect of PD1+ T-cell fractions on survival rates of myelofibrosis patients Data from 41 MF patients were further analyzed using the Kaplan-Meier method (34 patients [82.9%] had died) to determine whether the distribution of PD1+ T-cell fractions affects

haematologica | 2021; 106(9)

T cells affect the prognosis of myelofibrosis

Figure 5. Association between myelofibrosis T-cell subsets and PD1+ fractions with disease progression. Correlations between T-cell subsets and PD1+ fractions with peripheral blood cell counts and spleen size were assessed at treatment baseline. Pre- and on-treatment T cells were also compared based on selected parameters of disease progression/reduction. (A) Correlation of CD4+ and CD8+ T-cell subsets with absolute monocyte and platelet counts (i), and correlation of PD1+/CD4+ and PD1+/CD8+ T-cell percent with absolute leukocyte counts and spleen size (ii). (B) Comparison of CD4+ and CD8+ total T cells, effector memory T cell (T ) subsets, and PD1+ fractions based on the baseline spleen size, bone marrow (BM) fibrosis grade, and JAK2V617F allele burden. Data are shown for healthy control (CTRL; gray), baseline (Bsl; red), and ruxolitinib-treated (Ruxo; light blue) groups. Follow-up values were calculated as a mean of all the analyzed time- points over the course of treatment. Spleen size was defined by physical examination (the measured distance of palpable spleen edge from the left costal margin in the left midclavicular line) with the cutoff value of 20 cm. BM fibrosis grade was assessed in accordance with European consensus criteria. Mutant JAK2 allele burden was quantified using quantitative polymerase chain reaction analysis and dichotomized using the cutoff value of 50%. Regression lines are shown in blue; gray shaded areas denote 95% confidence intervals. Dots with error bars represent means with standard deviation. The Pearson coefficient was used to determine the degree of correlation. A Welch or paired t-test was used to compare the two groups. P values <0.05 were considered statistically significant. *P<0.05; **P<0.01; ***P<0.001. n.s.: not significant. EM

haematologica | 2021; 106(9)

2391

I. Veletic et al.

Figure 6. Association of myelofibrosis T-cell subsets and PD1+ fractions with clinical response to ruxolitinib treatment. (A, B) Comparison of T-cell subset (A) and PD1+ fraction (B) distribution based on spleen response of patients with myelofibrosis 6 months into ruxolitinib treatment. Spleen response was classified as either complete resolution (CR; n=9) or persistent splenomegaly (PS; n=29). CR was defined as no palpable splenomegaly after 6 months of treatment in patients with ≥5 cm of palpable spleen at treatment baseline. P values <0.05 were considered statistically significant. *P<0.05; **P<0.01; ***P<0.001.

2392

haematologica | 2021; 106(9)

T cells affect the prognosis of myelofibrosis

the outcome of ruxolitinib treatment. Using the log-rank test, we found that an increase of CD4+/PD1+ cells above 3.31% or of CD8+/PD1+ cells above 6.12% was associated with a poor overall survival (P=0.014 and P=0.003, respectively) (Figure 7Ai). However, significant differences in survival were maintained only across CD8+, and not CD4+, T , T , T , and T cells (P=0.002, P=0.013, P=0.009, and P=0.007, respectively) (Figure N

EFF

7Aii-iii), suggesting that increased PD1+ T-cell fractions, typically associated with T-cell exhaustion, are a better predictor of outcome within the cytotoxic T-cell subset. A univariate Cox proportional hazard regression analysis of the clinical variables revealed that disease subtype (primary MF vs. secondary MF), BM fibrosis grade (MF-3 vs. MF-1/2), transformation to acute myeloid leukemia, or transfusion dependence were significant

iii

Figure 7. Association of baseline PD1+ T-cell fractions with the clinical outcome of ruxolitinib-treated patients with myelofibrosis. Associations between the survival of patients with myelofibrosis (MF) and percent of PD1+ T cells in differentiation and activation subsets at the treatment baseline were analyzed using the KaplanMeier method and Cox models. (A) Kaplan-Meier survival analysis based on percent of PD1+ cells in CD4/CD8 differentiation subsets (i), and in CD4+ (ii) and CD8+ (iii) activation subsets. (B) Results of multivariate analysis of survival using high versus low PD1+/CD4+ and PD1+/CD8+ baseline T-cell subsets are shown. Cutoff values for dichotomization of each subset into high and low groups were determined using the maximally selected rank statistic. P-values for differences in overall survival were calculated using the log-rank test. Each multivariate model also included disease subtype (primary [PMF] vs. secondary MF [PPV/PET MF]), grade of bone marrow fibrosis (MF-3 vs. MF-1/2), transformation to acute myeloid leukemia (AML), and transfusion dependence as confounding predictors. Rhombi depict the hazard ratio (HR) of each predictor, and lines represent 95% confidence intervals (CI). P values <0.05 were considered statistically significant. n.s.: not significant.

haematologica | 2021; 106(9)

2393

I. Veletic et al.

confounding factors for overall survival in this cohort of patients. We therefore performed a multivariate analysis of survival after adjusting the model for these four predictors (Figure 7B). This analysis confirmed that PD1-overexpressing fractions of CD8+ cells, but not CD4+ cells, before initiation of ruxolitinib treatment, independently predicted overall survival (hazard ratio: 2.48; P=0.03). Taken together, our results demonstrated that increased PD1+/CD8+ T-cell subsets were significantly associated with a high risk of death in ruxolitinib-treated MF patients.

Discussion In the current study, we found that T-cell subsets of patients with MF shifted from a quiescent to an activated state and that treatment with ruxolitinib reduced the activation of both helper CD4 and cytotoxic CD8 T cells in a time-dependent manner (Figure 8). The activation pattern of CD8+ T cells was significantly similar to that of polycythemia vera,31 including decreased T and T , unaltered T and increased T subsets. However, in MF, CD4+ N

EFF

cells were also considerably skewed toward an effector cell phenotype, unlike polycythemia vera. Whereas CD8+ cells are activated by major histocompatibility complex (MHC) type I molecules, priming of CD4+ T cells is restricted to MHC class II on predominantly monocyte-derived antigen-presenting cells. Because MHC expression is induced by activated JAK2,32 our data point toward a predominant role of neoplastic monocytes in aberrant T-cell responses in MF. Furthermore, circulating monocyte-derived dendritic cells from patients with MF were extremely efficient in antigen uptake as compared to dendritic cells from healthy individuals, despite their reduced numbers and function.33 In comparison, T cells of acute myeloid leukemia patients at diagnosis are predominantly CD8+,34,35 whereas their activation seems to differ minimally from normal cells in both CD4+ and CD8+ subsets.36,37 Overall, the activation status of T-cell subsets in patients with MF is consistent with an ongoing antineoplastic immune response, characteristic of the “T-cell inflamed” immune signature.38 Over the last decade, broad clinical experience has been acquired in treating MF patients with ruxolitinib. Overall, decreased rates of infections and spleen reduction with ruxoli-

Figure 8. Schematic representation of the circulating T-cell subset repertoire in patients with myelofibrosis at baseline and after treatment with ruxolitinib. The left panel summarizes flow cytometry data from 47 patients with myelofibrosis (MF) analyzed in this study. Prior to treatment, T cells are skewed towards effector subsets (middle) and PD1-expressing T cells are increased (bottom) compared to those in 28 age-matched normal donors. In addition, disease progression shifts T-cell subsets towards CD8+ phenotypes (top). Ruxolitinib treatment reverts the resting:effector T-cell ratio to normal (middle), but has little effect on CD4/CD8 subsets or percentage of PD1+ cells. The right panel summarizes correlations of differentiation subsets (top) and PD1+ fractions (bottom) with the clinical parameters at treatment baseline (dashed line). Increases in CD8/PD1-coexpressing subsets are associated with a lack of spleen response. In addition, a CD8-predominant T-cell repertoire is found in patients with monocytosis and low platelet counts, whereas abundance of PD1-overexpressing CD8+ T cells is predictive of poor overall survival.

2394

haematologica | 2021; 106(9)

T cells affect the prognosis of myelofibrosis

tinib treatment have been associated with improved infectionfree survival.39 However, while neoplastic myeloid cells are thought to be the main target of JAK inhibitors, ruxolitinib also exerts a potent effect on non-malignant immune cells. The present study is the first to demonstrate that long-term treatment with ruxolitinib repolarizes activated T cells in MF patients. This finding is in agreement with previous in vitro studies demonstrating decreased cytokine production in T cells from ruxolitinibtreated MF patients,11 and decreased activation, proliferation and function of T cells from normal individuals.9 In addition, our observation of decreased CD4+ cell subsets as a result of ruxolitinib therapy confirms the findings of a previous study that showed JAK inhibitor-induced decreases in the number and function of helper T cells.10 Similar suppressive effects of ruxolitinib have been observed in NK cells from MF patients.40 Importantly, we observed increased PD1+ fractions among all assessed subsets, indicating that T cells are functionally exhausted in MF. These findings are in agreement with a recent study that showed increased PDL1 expression on myeloid cells from patients with JAK2-mutated myeloproliferative neoplasms.19 Our findings also corroborate PD1 expression patterns previously reported in circulating CD4+ and CD8+ cells of patients with MF.20 In the aforementioned study, however, lack of any T-cellspecific markers in the gating strategy employed makes the reported results difficult to interpret. In our study, we specifically analyzed CD45+/CD3+/αβ+ T cells and assessed PD1-expressing cell fractions across the different activation subsets. T-cell exhaustion is typically manifested by a progressive defect in production of interferon-γ, IL-2, and tumor necrosis factor; T cells incapable of releasing these cytokines have been implicated in promoting the differentiation of monocytes into fibrocytes.41 It remains to be established how T-cell dysfunction affects the population of neoplastic fibrocytes, which induce BM fibrosis in MF.42 Conversely, there was no significant difference in the expression of either PD1 in T cells or PDL1 in blasts of patients with newly diagnosed acute myeloid leukemia, chronic myelomonocytic leukemia or myelodysplastic syndromes,37,43,44 suggesting that the neoplastic clone in MF exerts stronger immunogenicity with a significantly dysfunctional capacity as compared with other myeloid mallignancies. In our cohort of MF patients, monocytosis and thrombocythopenia were associated with a predominantly CD8+ T-cell phenotype. In addition, high levels of CD8+ cells and increased PD1+ fractions within the CD8 compartment correlated with disease progression and poor outcome. Although, we did not

References 1. Barosi G. An immune dysregulation in MPN. Curr Hematol Malig Rep. 2014;9(4):331-339. 2. Hasselbalch HC, Bjorn ME. MPNs as inflammatory diseases: the evidence, consequences, and perspectives. Mediators Inflamm. 2015;2015:102476. 3. Vainchenker W, Kralovics R. Genetic basis and molecular pathophysiology of classical myeloproliferative neoplasms. Blood. 2017;129(6):667-679. 4. Hasselbalch HC. The role of cytokines in the initiation and progression of myelofibrosis. Cytokine Growth Factor Rev. 2013; 24(2):133-145. 5. Veletic I, Manshouri T, Newberry KJ, Garnett J, Verstovsek S, Estrov Z. Pentraxin3 plasma levels correlate with tumour burden and overall survival in patients with primary myelofibrosis. Br J Haematol.

haematologica | 2021; 106(9)

observe that ruxolitinib altered the percentage of PD1+ T cells, the survival of ruxolitinib-treated patients with low PD1 levels was significantly improved, suggesting that T-cell dysfunction is associated with a poor response to treatment with ruxolitinib. Remarkably, like in MF, in acute lymphoblastic leukemia, a disease in which the JAK-STAT pathway is often constitutively activated,45-47 low numbers of PD1+ T cells predicted an improved treatment outcome. Because the spleen is a T-cell reservoir, and spleen size correlated with PD1+ fractions, whereas a low CD8+/PD1+ T-cell percent correlated with a favorable response to ruxolitinib treatment, it is likely that T-cell exhaustion plays a role in the pathogenesis of MF and the response to JAK-inhibitor treatment. Collectively, our data suggest that both cytotoxic and helper T cells in MF are overly activated and harbor increased PD1+ fractions. Long-term JAK inhibition reverses terminal T-cell overactivation; nonetheless, high levels of PD1-expressing CD8+ T cells result in poor survival. A further in-depth analysis of the innate immune system, including the heterogeneous T-cell populations and their interaction with the MF neoplastic myeloid cells, is warranted.

Disclosures SV receives research funding from Incyte Corporation, Wilmington, DE, USA. The remaining authors declare that they have no competing financial interests. Contributions IV analyzed and interpreted data, performed the statistical analyses, created the figures, and wrote the manuscript; SP analyzed and interpreted data; SP and TM carried out the experiments; GMNG performed the statistical analyses; SV directed the project, supervised the study, and treated the patients included in the study; and ZE conceived, designed and supervised the study, interpreted data, and wrote the manuscript. All authors provided critical feedback and helped to develop the final manuscript. Acknowledgments The authors thank the Hanns A. Pielenz Foundation for financial support. The authors acknowledge Helen T. Chifotides for scientific editing assistance. This work was performed in part in the Flow Cytometry and Cellular Imaging Core Facility and used the Biostatistics Resource Group; both are supported by the National Cancer Institute, National Institutes of Health under award number P30 CA016672.

2018;185(2):382-386. 6. Villarino AV, Kanno Y, O’Shea JJ. Mechanisms and consequences of Jak-STAT signaling in the immune system. Nat Immunol. 2017;18(4):374-384. 7. Perner F, Perner C, Ernst T, Heidel FH. Roles of JAK2 in aging, inflammation, hematopoiesis and malignant transformation. Cells. 2019;8(8):854. 8. Elli EM, Borate C, Mendicino F, Palandri F, Palumbo GA. Mechanisms underlying the anti-inflammatory and immunosuppressive activity of ruxolitinib. Front Oncol. 2019;9:1186. 9. Heine A, Held SAE, Daecke SN, et al. The JAK-inhibitor ruxolitinib impairs dendritic cell function in vitro and in vivo. Blood. 2013;122(7):1192-1202. 10. Yajnanarayana SP, Stuebig T, Cornez I, et al. JAK1/2 inhibition impairs T cell function invitro and in patients with myeloprolifera-

tive neoplasms. Br J Haematol. 2015; 169(6):824-833. 11. Keohane C, Kordasti S, Seidl T, et al. JAK inhibition induces silencing of T helper cytokine secretion and a profound reduction in T regulatory cells. Br J Haematol. 2015;171(1):60-73. 12. Holmstrom MO, Riley CH, Svane IM, Hasselbalch HC, Andersen MH. The CALR exon 9 mutations are shared neoantigens in patients with CALR mutant chronic myeloproliferative neoplasms. Leukemia. 2016; 30(12):2413-2416. 13. Holmstrom MO, Hjortso MD, Ahmad SM, et al. The JAK2V617F mutation is a target for specific T cells in the JAK2V617F-positive myeloproliferative neoplasms. Leukemia. 2017;31(2):495-498. 14. Holmstroem MO, Riley CH, Skov V, Svane IM, Hasselbalch HC, Andersen MH. Spontaneous T-cell responses against the

2395

I. Veletic et al. immune check point programmed-death-ligand 1 (PD-L1) in patients with chronic myeloproliferative neoplasms correlate with disease stage and clinical response. Oncoimmunology. 2018;7(6):e1433521. 15. Jorgensen MA, Holmstrom MO, Martinenaite E, Riley CH, Hasselbalch HC, Andersen MH. Spontaneous T-cell responses against Arginase-1 in the chronic myeloproliferative neoplasms relative to disease stage and type of driver mutation. Oncoimmunology. 2018;7(9):e1468957. 16. Blank CU, Haining WN, Held W, et al. Defining ‘T cell exhaustion’. Nat Rev Immunol. 2019;19(11):665-674. 17. Sharma P, Allison JP. The future of immune checkpoint therapy. Science. 2015;348(6230): 56-61. 18. Sharpe AH, Pauken KE. The diverse functions of the PD1 inhibitory pathway. Nat Rev Immunol. 2018;18(3):153-167. 19. Prestipino A, Emhardt AJ, Aumann K, et al. Oncogenic JAK2(V617F) causes PD-L1 expression, mediating immune escape in myeloproliferative neoplasms. Sci Transl Med. 2018;10(429):eaam7729. 20. Wang J-C, Chen C, Kundra A, et al. Programmed cell death receptor (PD-1) ligand (PD-L1) expression in Philadelphia chromosome-negative myeloproliferative neoplasms. Leuk Res. 2019;79:52-59. 21. 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. 22. Keenan TE, Burke KP, Van Allen EM. Genomic correlates of response to immune checkpoint blockade. Nat Med. 2019;25(3): 389-402. 23. Shin DS, Zaretsky JM, Escuin-Ordinas H, et al. Primary resistance to PD-1 blockade mediated by JAK1/2 mutations. Cancer Discov. 2017;7(2):188-201. 24. Verstovsek S, Kantarjian H, Mesa RA, et al. Safety and efficacy of INCB018424, a JAK1 and JAK2 inhibitor, in myelofibrosis. N Engl J Med. 2010;363(12):1117-1127. 25. Vardiman JW, Thiele J, Arber DA, et al. The 2008 revision of the World Health Organization (WHO) classification of

2396

myeloid neoplasms and acute leukemia: rationale and important changes. Blood. 2009;114(5):937-951. 26. Verstovsek S, Mesa RA, Gotlib J, et al. A double-blind, placebo-controlled trial of ruxolitinib for myelofibrosis. N Engl J Med. 2012;366(9):799-807. 27. Verstovsek S, Mesa RA, Gotlib J, et al. Longterm treatment with ruxolitinib for patients with myelofibrosis: 5-year update from the randomized, double-blind, placebo-controlled, phase 3 COMFORT-I trial. J Hematol Oncol. 2017;10(1):55. 28. Cervantes F, Dupriez B, Passamonti F, et al. Improving survival trends in primary myelofibrosis: an international study. J Clin Oncol. 2012;30(24):2981-2987. 29. Chen DS, Mellman I. Elements of cancer immunity and the cancer-immune set point. Nature. 2017;541(7637):321-330. 30. Boiocchi L, Espinal-Witter R, Geyer JT, et al. Development of monocytosis in patients with primary myelofibrosis indicates an accelerated phase of the disease. Mod Pathol. 2013;26(2):204-212. 31. Cardoso EM, Esgalhado AJ, Patrao L, et al. Distinctive CD8(+) T cell and MHC class I signatures in polycythemia vera patients. Ann Hematol. 2018;97(9):1563-1575. 32. Brutkiewicz RR. Cell signaling pathways that regulate antigen presentation. J Immunol. 2016;197(8):2971-2979. 33. Romano M, Sollazzo D, Trabanelli S, et al. Mutations in JAK2 and Calreticulin genes are associated with specific alterations of the immune system in myelofibrosis. Oncoimmunology. 2017;6(10):e1345402. 34. Le Dieu R, Taussig DC, Ramsay AG, et al. Peripheral blood T cells in acute myeloid leukemia (AML) patients at diagnosis have abnormal phenotype and genotype and form defective immune synapses with AML blasts. Blood. 2009;114(18):3909-3916. 35. Rey J, Fauriat C, Kochbati E, et al. Kinetics of cytotoxic lymphocytes reconstitution after induction chemotherapy in elderly AML patients reveals progressive recovery of normal phenotypic and functional features in NK cells. Front Immunol. 2017;8:64. 36. Vidriales MB, Orfao A, Lopezberges MC, et al. Lymphoid subsets in acute myeloid leukemias - increased number of cells with NK phenotype and normal T-cell distribu-

tion. Ann Hematol. 1993;67(5):217-222. 37. Schnorfeil FM, Lichtenegger FS, Emmerig K, et al. T cells are functionally not impaired in AML: increased PD-1 expression is only seen at time of relapse and correlates with a shift towards the memory T cell compartment. J Hematol Oncol. 2015;8(93):93. 38. O’Donnell JS, Teng MWL, Smyth MJ. Cancer immunoediting and resistance to T cell-based immunotherapy. Nat Rev Clin Oncol. 2019;16(3):151-167. 39. Polverelli N, Palumbo GA, Binotto G, et al. Epidemiology, outcome, and risk factors for infectious complications in myelofibrosis patients receiving ruxolitinib: a multicenter study on 446 patients. Hematol Oncol. 2018;36(3):561-569. 40. Schoenberg K, Rudolph J, Vonnahme M, et al. JAK inhibition impairs NK cell function in myeloproliferative neoplasms. Cancer Res. 2015;75(11):2187-2199. 41. Niedermeier M, Reich B, Gomez MR, et al. CD4(+) T cells control the differentiation of Gr1(+) monocytes into fibrocytes. Proc Natl Acad Sci U S A. 2009;106(42):17892-17897. 42. Verstovsek S, Manshouri T, Pilling D, et al. Role of neoplastic monocyte-derived fibrocytes in primary myelofibrosis. J Exp Med. 2016;213(9):1723-1740. 43. Yang H, Bueso-Ramos C, DiNardo C, et al. Expression of PD-L1, PD-L2, PD-1 and CTLA4 in myelodysplastic syndromes is enhanced by treatment with hypomethylating agents. Leukemia. 2014;28(6):12801288. 44. Kronig H, Kremmler L, Haller B, et al. Interferon-induced programmed death-ligand 1 (PD-L1/B7-H1) expression increases on human acute myeloid leukemia blast cells during treatment. Eur J Haematol. 2014; 92(3):195-203. 45. Vainchenker W, Constantinescu SN. JAK/STAT signaling in hematological malignancies. Oncogene. 2013;32(21):2601-2613. 46. Yang K, Xu J, Liu QH, Li J, Xi YF. Expression and significance of CD47, PD1 and PDL1 in T-cell acute lymphoblastic lymphoma/ leukemia. Pathol Res Pract. 2019; 215(2):265271. 47. Hohtari H, Bruck O, Blom S, et al. Immune cell constitution in bone marrow microenvironment predicts outcome in adult ALL. Leukemia. 2019;33(7):1570-1582.

haematologica | 2021; 106(9)

ARTICLE

Myeloploriferative Disorders

Dynamics of mutations in patients with essential thrombocythemia treated with imetelstat

Ferrata Storti Foundation

Elisabeth Oppliger Leibundgut,1,2 Monika Haubitz,2 Bart Burington,3 Oliver G. Ottmann,4 Gary Spitzer,5 Olatoyosi Odenike,6 Michael A. McDevitt,7 Alexander Röth,8 David S. Snyder9 and Gabriela M. Baerlocher1,2 1 Department of Hematology and Central Hematology Laboratory, Inselspital, Bern University Hospital, Bern, Switzerland; 2Department for BioMedical Research (DBMR), University of Bern, Bern, Switzerland; 3Geron Corporation, Menlo Park, CA, USA (at time of study); 4Department of Haematology, Cardiff University, Cardiff, UK; 5Cadex Genomics, Redwood City, CA, USA; 6University of Chicago Medical Center, Chicago, IL, USA; 7Johns Hopkins University School of Medicine, Divisions of Hematology, and Hematological Malignancy, Baltimore, MD, USA (at time of study); 8Department of Hematology, West German Cancer Center, University Hospital Essen, University of Duisburg-Essen, Essen, Germany and 9Gehr Family Center for Leukemia Research, City of Hope National Medical Center, Duarte, CA, USA

Haematologica 2021 Volume 106(9):2397-2404

ABSTRACT

n a phase II study, the telomerase inhibitor imetelstat induced rapid hematologic responses in all patients with essential thrombocythemia who were refractory to or intolerant of prior therapies. Significant molecular responses were achieved within 3-6 months in 81% of patients with phenotypic driver mutations in JAK2, CALR and MPL. Here, we investigated the dynamics of additional somatic mutations in response to imetelstat. At study entry, 50% of patients carried one to five additional mutations in the genes ASXL1, CBL, DNMT3A, EZH2, IDH1, SF3B1, TET2, TP53 and U2AF1. Three patients with baseline mutations also had late-emerging mutations in TP53, IDH1 and TET2. Most clones with additional mutations were responsive to imetelstat and decreased with the driver mutation, including the poor prognostic ASXL1, EZH2 and U2AF1 mutations, while SF3B1 and TP53 mutations were associated with poorer molecular response. Overall, phenotypic driver mutation response was significantly deeper in patients without additional mutations (P=0.04) and correlated with longer duration of response. In conclusion, this detailed molecular analysis of heavily pretreated and partly resistant patients with essential thrombocythemia reveals a high individual patient complexity. Moreover, imetelstat demonstrates potential to inhibit efficiently co-incident mutations occurring in neoplastic clones in patients with essential thrombocythemia. (ClinicalTrials.gov number, NCT01243073).

Introduction Imetelstat is a 13-mer lipid-conjugated oligonucleotide that targets the RNA template of hTERC and can, therefore, inhibit activity of telomerase and cell proliferation in cancer cells.1 hTERT, the catalytic subunit of telomerase that is generally not found in somatic cells, is expressed in megakaryocytes of patients with essential thrombocythemia (ET), a myeloproliferative neoplasm (MPN).2 Previously, we demonstrated a dose-dependent inhibition of megakaryocytic colony-forming units from patients with ET but not from healthy individuals in vitro.3 In a phase II study of ET patients who were refractory to or intolerant of prior treatment, imetelstat induced rapid and durable hematologic responses in all patients, and molecular responses were achieved in the majority of patients within 3-6 months.4 In ET, JAK2 V617F, CALR and MPL mutations are phenotypic driver mutations present in around 90% of patients; the remaining cases are termed “triple negative”. Non-canonical gain-of-function mutations have been identified in the JAK2 and MPL genes in a minority of triple-negative patients.5,6 Additional recurrent somatic mutations occur at lower frequencies in a number

haematologica | 2021; 106(9)

Correspondence: ELISABETH OPPLIGER LEIBUNDGUT elisabeth.oppliger@insel.ch GABRIELA M. BAERLOCHER gabriela.baerlocher@hematology.ch Received: March 23, 2020. Accepted: July 21, 2020. Pre-published: July 30, 2020. https://doi.org/10.3324/haematol.2020.252817

2397

E. Oppliger Leibundgut et al.

of genes in MPN, and clonality has been demonstrated.7-9 In ET, mutations in ASXL1, TET2 and DNMT3A genes are most frequent and are all involved in epigenetic regulation. Less frequent recurrent mutations are detected in EZH2, TP53, IDH1, IDH2 and CBL, as well as in genes of the splicing machinery, such as SF3B1, SRSF2, U2AF1 and ZRSF2. So-called “adverse mutations” in SF3B1, SRSF2, U2AF1, TP53, IDH2 and EZH2 have been found to have negative effects on overall and myelofibrosis-free survival in ET, and TP53 mutations predict leukemic transformation.8-10,14 Furthermore, ASXL1 mutations have been identified as a genetic risk factor for transformation to myelofibrosis in ET patients, as they are most frequently found in post-ET myelofibrosis.11 Subsequently, genomic data were integrated in prognostic models to predict patients’ outcomes.13,14 An influence of additional non-driver mutations on treatment response in MPN has been reported for interferon-α, ruxolitinib and imetelstat. In patients with CALR-mutated ET treated with interferon-α, the presence of additional mutations in ASXL1, TET2, IDH2 and TP53 correlated with a poorer molecular response.15 TET2mutated clones were resistant to interferon-α therapy in JAK2-mutated patients with polycythemia vera.16 Resistance to ruxolitinib was reported in patients with myelofibrosis carrying three or more mutations.17 Furthermore, in the first clinical trial with imetelstat in myelofibrosis patients, treatment response was reported to be negatively influenced by ASXL1 mutations and favorably impacted by SF3B1 and U2AF1 mutations.18 In the present study, we assessed a panel of genes frequently mutated in MPN by next-generation sequencing at study entry and during treatment with imetelstat, and investigated the dynamics of additional mutations in ET patients and their association with hematologic and molecular response.

Methods Patients and response criteria A total of 18 patients with ET diagnosed according to the World Health Organization (WHO) 2008 criteria were treated with imetelstat in a phase II study.4 The study was approved by the institutional review board at each participating site. All patients provided written informed consent. Diagnoses were re-evaluated according to the WHO 2016 classification.19 Sequential blood samples were taken at baseline and at up to eight time-points during treatment with imetelstat through cycle 26 (28-day cycles), with approximately 12 weeks between samples. Mutational analysis was performed on all collected samples. Clinical and hematologic responses were assessed according to the European LeukemiaNet criteria.20 Molecular responses of phenotypic driver mutations were defined as follows: a major molecular response (MMR) was achieved when the mutant allele burden reduction was >50% from baseline value, and a partial molecular response (PMR) was present when a 25% to 49% reduction of the mutant allele burden was observed.

Genetic analysis DNA was extracted from granulocytes or leukocytes from peripheral blood samples. The molecular response of JAK2 V617F, CALR and MPL mutations was assessed using allele-specific realtime polymerase chain reaction, sequencing and fragment length analysis, respectively, as previously described.4

2398

Targeted next-generation sequencing of all relevant exons and adjacent intronic sequences of 15 recurrently mutated genes (ASXL1, CBL, DNMT3A, EZH2, IDH1, IDH2, JAK2, MPL, SOCS1, TET2, TP53, SF3B1, SRSF2, U2AF1 and ZRSR2) was performed using Ion Torrent™ semiconductor chip technology on the Ion Personal Genome Machine® PGM™ (Thermo Fisher Scientific Inc.). Genes were covered by two custom-designed amplicon libraries comprising 511 and 307 amplicons. In addition, a commercial panel for TP53 was used for confirmation of TP53 variants (Ion AmpliSeqTM Community Panel TP53, Thermo Fisher). For each primer pool, 10 ng of DNA were processed using the AmpliSeqTM chemistry for selective amplification of target sequences and library preparation according to the manufacturer’s instructions. Libraries were diluted and combined according to the Ion PGM chip size to obtain a minimum coverage of 500x for all amplicons. Templates were prepared on the Ion Chef™ and sequencing was performed on the Ion PGM instrument. Variants were called using IonTorrent VariantCaller v4.3 software based on the human reference genome (GRCh37/hg19). Analysis of TP53 was performed according to the manufacturer’s instruction. Annotation was done using the Mutalyzer, dbSNP, COSMIC, ClinVar, UniProt and IARC TP53 databases and the functional in silico prediction algorithms PolyPhen-2 and SIFT.21 Fragment analysis was used to screen for insertions and deletions in ASXL1 exon 12 (NG_027868.1), which are frequently missed by next-generation sequencing. Primers were designed according to Pratcorona et al.22 with small adaptations, and analysis was performed on a 3130 Genetic Analyzer using peak scanner software (Thermo Fisher). Sequences were confirmed by Sanger sequencing. The limits of detection for real-time polymerase chain reaction analysis of JAK2 and MPL mutations were 0.5%, whereas those for CALR and ASXL1, determined by fragment analysis, and all variants detected by next-generation sequencing were set at 2%.

Validation of genetic variants All novel variants were confirmed by Sanger sequencing. For ASXL1 and TET2 analysis, published primers were used,23,24 and primers for other genes were designed using Oligo7 and Primer3 software.25,26 Low-level variants (<10%) were confirmed by a second round of next-generation sequencing analysis.

Statistics Categorical patients’ characteristics were summarized by frequencies and percentages and continuous characteristics by medians, means and an unpaired Student t-test. The efficiency of imetelstat treatment was analyzed by a paired Student t-test comparing percentages of mutant allele burdens before treatment and at best response. Smooth estimates of allele burdens over time were generated using running medians and smoothing splines. Standard errors and confidence intervals were computed by bootstrap.

Results Characteristics of patients and phenotypic driver mutations Of 18 patients with ET enrolled in the study, nine (50%) were refractory and 14 (78%) were intolerant of at least one prior therapy. Thirteen patients had received more than one prior therapy and the median time since diagnosis was 7.2 years (range, 0.3-24.9) (Table 1). The median age of patients at study entry was 59.5 years (range, 21-83) (Online Supplementary Table S1). Upon treatment with imetelstat, all patients had a hematologic response, with 16 patients achieving complete hematologic responses.4 haematologica | 2021; 106(9)

Mutations in ET treated with imetelstat

Table 1. Characteristics of patients.

Pt.

Age at Sex dg

Years Prior HR since therapies, dg n.

Duration of Duration of imetelstat response, therapy, months months

Driver Reduction mutation in allele burden at BR, %

Best MR

Additional mutations at study entry

Late-emerging mutations*

1 2 3 4

52 43 60 67

F M M F

20.7 4.6 3.3 19.9

3 2 3 3

CR CR CR CR

18.5 28.9 23.5 18.3

18.3+ 29.8+ 24+ 17.6

JAK2 V617F CALR Type 1 JAK2 V617F CALR Type 2

-96 -31 -82 -38

5 6 7 8 9

55 69 83 64 48

F M F M M

0.3 7.7 1.8 5.7 1.3

1 2 2 1 1

CR CR CR CR CR

25.2 18.7 13.2 14.5 10.8

23.9+ 17.5+ 2.4 16.5+ 9.7+

JAK2 V617F JAK2 V617F none CALR Type 1 JAK2 V617F

-94 -24 na -48 -100

1.9

15.9

14.5+

JAK2 V617F

-90

11 12

56 80

F M

12.1 11.8

2 2

CR CR

12.1 33.2

10.9 31.3+

na -55

13 14

46 77

F M

6.9 1

3 2

CR CR

36.7 23.3

29.5 21.7+

none CALR del1092-1124 MPL JAK2 V617F

MMR TP53 p.Arg249Lys PMR TET2 p.Tyr1608Leufs*6 TP53 p.Cys135Trp MMR PMR DNMT3A p.Tyr735Cys EZH2 p.Asp293Ala SF3B1 p.Lys666Arg TET2 p.Arg1465Ter TP53 p.His179Leu MMR No na SF3B1 p.Lys700Glu PMR MMR ASXL1 p.Gly646Trpfs*12 U2AF1 p.Gln157Pro CBL c.1432-1G>A MMR DNMT3A p.M880V TET2 p.Met1772Cysfs*48 na MMR

-66 -72

MMR MMR

15 16 17 18

47 21 61 59

M F F F

11.6 13.4 10.9 24.9

4 2 3 3

PR CR CR PR

7.8 24 16.6 6.9

3.8+ 22.4+ 10.7 4.2

CALR Type 1 JAK2 V617F JAK2 V617F MPL

-15 -96 -82 0

No MMR MMR No

DNMT3A p.Ala644Thr IDH1 p.Arg132His DNMT3A p.Arg688His DNMT3A c.2597+1G>A

ASXL1 p.Tyr591Ter DNMT3A p.Gly722Asp

TET2 p.Ser137Gly

Dg: diagnosis; HR: hematologic response; BR: best molecular response; MR: molecular response; MMR: major molecular response (>50% mutant allele burden reduction from baseline); PMR: partial molecular response (25% to 49% mutant allele burden reduction from baseline); No, no response; +, continued on treatment; * after best molecular response of the driver mutation.

Figure 1. Frequency and distribution of mutations by patient at study entry.

haematologica | 2021; 106(9)

2399

E. Oppliger Leibundgut et al.

Figure 2. Distribution of molecular response and driver mutations among patients with or without additional mutations. (A) Patients without additional mutations (n=8). (B) Patients with additional mutations (n=8). MMR: major molecular remission; PMR: partial molecular remission; NR: no remission; JAK2: JAK2 V617F; CALR: CALR mutation; MPL: MPL mutation.

With regard to phenotypic driver mutations, nine (50%) patients had a JAK2 V617F mutation, five (28%) patients had CALR mutations (type 1, n=3; type 2, n=1; a novel 33 bp deletion at position 1092, n=1) and two patients had MPL mutations (1 with W515L, 1 with W515K). Two patients (11%) were triple negative. Overall, there was a significant reduction of driver mutant allele burden, with a median decrease of 69% at best response during treatment (P<0.001). In detail, of 16 patients with a phenotypic driver mutation, ten (63%) reached a MMR (8 JAK2mutated, 1 CALR-mutated, 1 MPL-mutated), three (19%) reached a PMR (3 CALR-mutated), and three (19%) patients did not reach a PMR (1 JAK2-mutated, 1 CALRmutated, 1 MPL-mutated).

Additional mutations at study entry At study entry, a total of 18 different additional somatic mutations (11 missense, 3 frameshift, 2 nonsense, 2 splice site) were identified in nine patients (50%), affecting the DNMT3A (n=6), TET2 (n=3), ASXL1 (n=2), TP53 (n=2), SF3B1 (n=2), CBL (n=1), EZH2 (n=1) and U2AF1 (n=1) genes (Figure 1). Details on mutations and variant allele frequencies at diagnosis and best response are given in Online Supplementary Table S2. Among the patients with any driver mutation, 40-56% carried up to five additional mutations (5/9 patients with JAK2 V617F, 2/5 with CALR mutation, 1/2 with MPL mutation), and of two triple-negative patients one had an additional mutation.

Impact of additional mutations on molecular response and dynamics of mutant clones Patients with or without additional mutations had similar molecular responses to imetelstat therapy with five (63%) patients reaching MMR in each group; one and two patients without and with additional mutations reached PMR, respectively (Figure 2). All patients with additional mutations who reached MMR had a JAK2 V617F driver mutation. Regarding the reduction in mutant allele burden, phenotypic driver mutation response was significantly deeper in patients without additional mutations (P=0.04) (Figure 3). 2400

Different dynamics of mutations in response to imetelstat were observed in individual patients (Figure 4). In five patients (#1, #2, #9, #10, #17), additional mutant allele burdens decreased with the driver mutation. In contrast, in two patients with three and five additional mutations (#4, #14), differential responses to imetelstat treatment were observed; i.e., the allele burden of some additional mutations decreased in parallel with the driver mutation while others persisted or increased (i.e., mutations in TP53, SF3B1, and DNMT3A) despite driver mutation response, suggesting the presence of at least two clones or subclones. Lack of response was observed in two patients: in patient #18, a MPL mutation did not respond while the DNMT3Amutated clone expanded, and in patient #7 without a driver mutation (triple negative), a known hotspot mutation in SF3B1 persisted at a high level. In total, non-responsive mutations were detected in TP53, DNMT3A and SF3B1 genes. Three patients (#2, #14, #17) acquired additional mutations in TP53, IDH1 and TET2 with low allele burden (mean 5%) after best molecular response of the driver mutation, at 10, 9 and 13 months of imetelstat treatment, respectively (Table 1). All three patients already had one to three preexisting additional mutations in other genes at study entry.

Clinical outcome in relation to additional mutations Hematologic and molecular responses were equally reached independently of the presence of additional mutations (Table 2). Loss of response was, however, more frequent in patients with additional mutations. Namely, four patients with additional mutations lost their molecular response, including three patients with DNMT3A mutations and one patient with a TET2 mutation, but none of the patients without additional mutations lost their molecular response (P=0.025). Patients with a higher burden of additional mutations at study entry had a shorter duration of clinical response compared to patients with no or a lower burden of additional mutations (10.2 vs. 22.1 months, median; cut-off at 10% mutant allele burden, P=0.053). haematologica | 2021; 106(9)

Mutations in ET treated with imetelstat

Table 2. Clinical outcome data.

All (n=18) Hematologic response Hematologic complete response Major molecular responsea Partial molecular responsea Median duration of treatment, months, (range) Median duration of response, months (range) Thromboembolic event Transformation to myelofibrosis Loss of hematologic response Loss of molecular responseb

18 (100%) 16 (89%) 10 (63%) 3 (19%) 18.4 (6.9-36.7) 18.3 (2.4-31.3) 3 (17%) 3 (17%) 6 (33%) 4 (31%)

No additional mutations With additional mutations (n=9) (n=9) 9 (100%) 8 (89%) 5 (63%) 1 (13%) 23.5 (7.8-36.7) 22.4 (3.8-31.3) 1 (11%) 1 (11%) 2 (22%) 0

9 (100%) 8 (89%) 5 (63%) 2 (25%) 16.6 (6.9-28.9) 14.5 (2.4-29.8) 2 (22%)c 2 (22%) 4 (44%) 4 (57%)

P n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. 0.025

n=16 for this analysis: two patients with no driver mutation were excluded; bn=13 for this analysis: two patients with no driver mutation and three patients with no molecular response were excluded; cone patient with two thromboembolic events. n.s.=not statistically significant.

Figure 3. Mean phenotypic driver mutant allele burden over time. The solid line represents the driver mutant allele burden in patients without additional mutations (n=8). The dashed line represents the driver mutant allele burden in patients with additional mutations. Patients without additional mutations reached significantly lower mutant allele burdens (P=0.04).

Loss of response was associated with thromboembolic events (3/4 events unrelated to therapy), resistance to imetelstat treatment or progression to myelofibrosis. Transformation to myelofibrosis occurred in two patients with additional mutations during follow-up and 6 months after treatment termination (#4 and #17, respectively). The former, a CALR-mutated patient, carried five additional mutations in DNMT3A, EZH2, SF3B1, TET2 and TP53, and the latter, a JAK2-mutated patient, had an ASXL1 and a late-emerging TET2 mutation. On retrospective evaluation, none of the patients fulfilled the criteria for prefibrotic myelofibrosis according to the newer WHO 2016 criteria. haematologica | 2021; 106(9)

Discussion This is the first report on the mutational repertoire of refractory and/or intolerant ET patients after one to four prior therapies. Following treatment with imetelstat, a first-in-class, specific telomerase inhibitor, all patients achieved hematologic responses, and significant molecular responses were seen within 3-6 months, i.e., 63% and 19% of patients with driver mutations reached MMR or PMR, respectively. At study entry, 50% of patients carried one to five somatic mutations in addition to the phenotypic driver mutation, including one triple-negative case. This frequen2401

E. Oppliger Leibundgut et al.

Figure 4. Best molecular response of patients with additional mutations. Mutant allele burdens of each mutation before imetelstat treatment and at best response. Driver mutations are depicted in blue. Limit of detection at 2%, dashed line. BT: before treatment; BR: at best response.

cy is higher than mutation rates reported from other cohorts of ET patients,7,8,12 although still lower than the 86% and 98% overall rates of additional mutations detected in patients with myelofibrosis.17,18 The high frequency of additional mutations in our ET cohort might reflect the concept of genetic instability in MPN and subsequent clonal evolution in a subset of these highly pretreated and partially resistant patients who had been diagnosed a median of 7.2 years previously. This concept is further supported by the finding that more than half of patients carried more than one additional mutation, as has also been reported by others.27 In addition, only patients with additional mutations at study entry acquired even more somatic mutations late during treatment (n=3). The most frequently mutated gene was DNMT3A followed by TET2. DNMT3A mutations co-occurred with other somatic mutations in three of four patients, in line with published data.17 In ET, DNMT3A and TET2 mutations are often early events involved in disease initiation, which may precede the JAK2 V617F mutation and influence the phenotype.9,28,29 Of the other genes mutated at 2402

study entry, TP53, SF3B1, U2AF1 and EZH2 are part of a group of “adverse risk mutations” for ET, based on their significantly poor impact on overall, leukemia-free and myelofibrosis-free survival, and ASXL1 mutations are known as molecular risk factors for transformation to myelofibrosis.12,14 Patients with or without additional mutations had similar molecular responses to imetelstat with 63% of MMR in both groups; however, the presence of additional mutations had a negative effect on the depth of response, as mutant allele burden reductions were significantly deeper in patients without additional mutations. Of interest, all patients with additional mutations who gained a deep response (MMR) had JAK2 V617F driver mutations while patients with CALR or MPL driver mutations had poorer responses. In contrast, response depth in patients without additional mutations was not assigned to a specific driver mutation type. Further evidence from larger cohorts of patients is needed to support this observation. The majority of cells with additional mutations were suppressed by imetelstat, and additional mutations haematologica | 2021; 106(9)

Mutations in ET treated with imetelstat

tracked with the driver mutation. Of note, ASXL1 mutations were also responsive to imetelstat treatment, although one ASXL1-mutated patient later lost response and transformed to myelofibrosis with an acquired TET2 mutation. This is in contrast to a study on imetelstat in myelofibrosis patients that reported a lack of response among patients with ASXL1 mutations.18 With regard to the additional mutations, we observed several patterns of response. The parallel decrease of one or more mutations with the driver mutation in five of nine patients suggests that coexistence of mutations in the same clone or subclone was frequent in our cohort of patients. Unfortunately, we were not able to track the clonal architecture or coexistence of mutations within a cell due to the lack of additional cell material. Discrepant patterns of response were seen in patients with multiple mutations that were responsive or persistent, with DNMT3A, SF3B1 and TP53 mutations persisting or increasing over time, suggesting the presence of independent clones. It has been reported that DNMT3A mutations are often present in preleukemic clones and persist during therapy in myeloid malignancies, e.g., in acute myeloid leukemia.30,31 In this study, the four patients with DNMT3A mutations were 78, 80, 84 and 87 years old at study entry, and had had ET for 1, 2, 25 and 20 years, respectively. Since they were significantly older than the patients without DNMT3A mutations (mean age at study entry 82 years vs. 64 years, P<0.05), antecedent age-related clonal hematopoiesis (ARCH/CHIP) may be a contributing factor.32-34 Individuals with ARCH/CHIP have a high risk of developing a hematologic malignancy. Experiments in mouse models carrying loss-of-function mutations in DNMT3A or TET2 suggest a competitive advantage and enhanced self-renewal capacity of the mutant stem cells leading to clonal expansion.34 Of the other non-responsive mutations in this study, mutations affecting the splicing factor SF3B1 are uncommon events in ET, reported to occur in 5% or fewer.12,14,35 They have been considered as “adverse mutations” based on their negative impact on myelofibrosis-free and overall survival.12,14 TP53 mutations in MPN were described to be present for several years at low allelic burden and, after loss of the wild-type TP53 allele, clones expanded rapidly resulting in leukemic transformation.8,36 Hence, the presence of TP53 mutations may be a warning of leukemic transformation in MPN.10 The presence of additional mutations per se, specific mutations and the total number of additional mutations have been associated with inferior response to treatment with interferon-α and ruxolitinib.15-18,37 In contrast, imetelstat treatment led to a high proportion of MMR in patients with or without additional mutations, although the latter patients had more reduction of mutant allele burden.

References 1. Ouellette MM, Wright WE, Shay JW. Targeting telomerase-expressing cancer cells. J Cell Mol Med. 2011;15(7):14331442. 2. Florena AM, Tripodo C, Di Bernardo A, et al. Different immunophenotypical apoptotic profiles characterise megakaryocytes of essential thrombocythaemia and primary myelofibrosis. J Clin Pathol. 2009;62(4): 331-338.

haematologica | 2021; 106(9)

Furthermore, initial mutant allele burden may have an impact on response as high-level additional mutations at study entry correlated with shorter duration of response. Overall, this detailed molecular analysis of heavily pretreated and resistant ET patients reveals high individual patient complexity, with half of the patients harboring up to five additional somatic mutations at study entry. These results raise the question of whether additional mutations were acquired prior to diagnosis or whether mutational events were induced during treatment with prior therapies. Additional studies are needed to address this question. In conclusion, treatment with imetelstat led to rapid and sustained hematologic and molecular responses and additional mutant allele burdens were also reduced. However, additional mutations significantly reduced the depth of response and had an impact on duration of response. Of acquired mutations with known adverse prognosis and/or risk for transformation to myelofibrosis or acute myeloid leukemia, ASXL1, EZH2 and U2AF1 mutations were responsive to imetelstat, while SF3B1 and one of two TP53 mutations persisted. These data emphasize imetelstat’s potential to inhibit neoplastic clones in patients with ET. Disclosures EOL has received research funding and honoraria from Geron; BB has acted as a consultant for Geron; OO has received research funding from Geron; MMcD is a current employee of Abbvie; AR has received research funding and honoraria from Geron and honoraria from Janssen; DSS has participated in advisory board work for and received honoraria from Gilead; GB has received research funding and honoraria from Geron and honoraria from Janssen. The other authors have no competing interests. Contributions EOL and GMB designed the study and wrote the manuscript; MH performed research; EOL, MH, BB and GMB analyzed and interpreted data; OGO, GS, OO, MAM, AR, DSS and GMB provided clinical data; BB performed statistical analysis and all authors read and approved the manuscript. Acknowledgments The authors would like to thank the patients, caregivers and staff who participated in this study, and Ingrid Helsen, Dania Hiltbrunner and Barbara Hügli for technical assistance at the Laboratory of Hematopoiesis and Molecular Genetics, Department of BioMedical Research, University of Bern. Funding This investigator-initiated and -driven study was supported by research funding from Geron to GMB and EOL.

3. Baerlocher GM, Haubitz M, Braschler TR, et al. Imetelstat inhibits growth of megakaryocyte colony-forming units from patients with essential thrombocythemia. Blood Adv. 2019;3(22):3724-3728. 4. Baerlocher GM, Oppliger Leibundgut E, et al. Telomerase inhibitor imetelstat in patients with essential thrombocythemia. N Engl J Med. 2015;373(10):920-928. 5. Milosevic Feenstra JD, Nivarthi H, Gisslinger H, et al. Whole-exome sequencing identifies novel MPL and JAK2 muta-

tions in triple-negative myeloproliferative neoplasms. Blood. 2016;127(3):325-332. 6. Cabagnols X, Favale F, Pasquier F, et al. Presence of atypical thrombopoietin receptor (MPL) mutations in triple-negative essential thrombocythemia patients. Blood. 2016;127(3):333-342. 7. Nangalia J, Massie CE, Baxter EJ, et al. Somatic CALR mutations in myeloproliferative neoplasms with nonmutated JAK2. N Engl J Med. 2013;369(25):2391-2405. 8. Lundberg P, Karow A, Nienhold R, et al.

2403

E. Oppliger Leibundgut et al. Clonal evolution and clinical correlates of somatic mutations in myeloproliferative neoplasms. Blood. 2014;123(14):2220-2228. 9. Vainchenker W, Kralovics R. Genetic basis and molecular pathophysiology of classical myeloproliferative neoplasms. Blood. 2017; 129(6):667-679. 10. Harutyunyan A, Klampfl T, Cazzola M, Kralovics R. p53 lesions in leukemic transformation. N Engl J Med. 2011;364(5):488490. 11. Cerquozzi S, Tefferi A. Blast transformation and fibrotic progression in polycythemia vera and essential thrombocythemia: a literature review of incidence and risk factors. Blood Cancer J. 2015;5(11):e366. 12. Tefferi A, Lasho TL, Guglielmelli P, et al. Targeted deep sequencing in polycythemia vera and essential thrombocythemia. Blood Adv. 2016;1(1):21-30. 13. Grinfeld J, Nangalia J, Baxter EJ, et al. Classification and personalized prognosis in myeloproliferative neoplasms. N Engl J Med. 2018;379(15):1416-1430. 14. Tefferi A, Guglielmelli P, Lasho TL, et al. Mutation-enhanced international prognostic systems for essential thrombocythaemia and polycythaemia vera. Br J Haematol. 2020;189(2):291-302. 15. Verger E, Cassinat B, Chauveau A, et al. Clinical and molecular response to interferon-alpha therapy in essential thrombocythemia patients with CALR mutations. Blood. 2015;126(24):2585-2591. 16. Kiladjian JJ, Masse A, Cassinat B, et al. Clonal analysis of erythroid progenitors suggests that pegylated interferon alpha-2a treatment targets JAK2V617F clones without affecting TET2 mutant cells. Leukemia. 2010;24(8):1519-1523. 17. Patel KP, Newberry KJ, Luthra R, et al. Correlation of mutation profile and response in patients with myelofibrosis treated with ruxolitinib. Blood. 2015;126(6):790-797.

2404

18. Tefferi A, Lasho TL, Begna KH, et al. A Pilot Study of the telomerase inhibitor imetelstat for myelofibrosis. N Engl J Med. 2015; 373(10):908-919. 19. 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. 20. Barosi G, Birgegard G, Finazzi G, et al. Response criteria for essential thrombocythemia and polycythemia vera: result of a European LeukemiaNet consensus conference. Blood. 2009;113(20):4829-4833. 21. Adzhubei IA, Schmidt S, Peshkin L, et al. A method and server for predicting damaging missense mutations. Nat Methods. 2010;7(4):248-249. 22. Pratcorona M, Abbas S, Sanders MA, et al. Acquired mutations in ASXL1 in acute myeloid leukemia: prevalence and prognostic value. Haematologica. 2012;97(3):388392. 23. 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. 24. Langemeijer SM, Kuiper RP, Berends M, et al. Acquired mutations in TET2 are common in myelodysplastic syndromes. Nat Genet. 2009;41(7):838-842. 25. Rychlik W. OLIGO 7 primer analysis software. Methods Mol Biol. 2007;402:35-60. 26. Untergasser A, Cutcutache I, Koressaar T, et al. Primer3--new capabilities and interfaces. Nucleic Acids Res. 2012;40(15):e115. 27. Jones AV, Cross NC. Inherited predisposition to myeloproliferative neoplasms. Ther Adv Hematol. 2013;4(4):237-253. 28. Nangalia J, Nice FL, Wedge DC, et al. DNMT3A mutations occur early or late in patients with myeloproliferative neoplasms and mutation order influences phenotype. Haematologica. 2015; 100(11): e438-442.

29. Ortmann CA, Kent DG, Nangalia J, et al. Effect of mutation order on myeloproliferative neoplasms. N Engl J Med. 2015; 372(7):601-612. 30. Debarri H, Lebon D, Roumier C, et al. IDH1/2 but not DNMT3A mutations are suitable targets for minimal residual disease monitoring in acute myeloid leukemia patients: a study by the Acute Leukemia French Association. Oncotarget. 2015; 6(39):42345-42353. 31. Jeziskova I, Musilova M, Culen M, et al. Distribution of mutations in DNMT3A gene and the suitability of mutations in R882 codon for MRD monitoring in patients with AML. Int J Hematol. 2015;102(5):553-557. 32. Jaiswal S, Fontanillas P, Flannick J, et al. Age-related clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014;371(26):2488-2498. 33. Zink F, Stacey SN, Norddahl GL, et al. Clonal hematopoiesis, with and without candidate driver mutations, is common in the elderly. Blood. 2017;130(6):742-752. 34. Jaiswal S, Ebert BL. Clonal hematopoiesis in human aging and disease. Science. 2019;366(6465):eaan4673. 35. Boiocchi L, Hasserjian RP, Pozdnyakova O, et al. Clinicopathological and molecular features of SF3B1-mutated myeloproliferative neoplasms. Hum Pathol. 2019;86:111. 36. Kubesova B, Pavlova S, Malcikova J, et al. Low-burden TP53 mutations in chronic phase of myeloproliferative neoplasms: association with age, hydroxyurea administration, disease type and JAK2 mutational status. Leukemia. 2018;32(2):450-461. 37. Montalban-Bravo G, Takahashi K, Patel K, et al. Impact of the number of mutations in survival and response outcomes to hypomethylating agents in patients with myelodysplastic syndromes or myelodysplastic/myeloproliferative neoplasms. Oncotarget. 2018;9(11):9714-9727.

haematologica | 2021; 106(9)

ARTICLE

Non Hodgkin Lymphoma

A three-gene signature based on MYC, BCL-2 and NFKBIA improves risk stratification in diffuse large B-cell lymphoma Enrico Derenzini,1,2 Saveria Mazzara,3 Federica Melle,3 Giovanna Motta,3 Marco Fabbri,3 Riccardo Bruna,1 Claudio Agostinelli,4 Alessandra Cesano,5 Chiara Antonia Corsini,6 Ning Chen,5 Simona Righi,4 Elena Sabattini,4 Annalisa Chiappella,7 Angelica Calleri,3 Stefano Fiori,3 Valentina Tabanelli,3 Antonello Cabras,8 Giancarlo Pruneri,8 Umberto Vitolo,9 Alessandro Massimo Gianni,1 Alessandro Rambaldi,10 Paolo Corradini,7 Pier Luigi Zinzani,11 Corrado Tarella1,2 and Stefano Pileri3

Ferrata Storti Foundation

Haematologica 2021 Volume 106(9):2405-2416

1 Onco-Hematology Division, IEO European Institute of Oncology IRCCS, Milan and European Institute of Oncology IRCCS, Milan, Italy; 2Department of Health Sciences, University of Milan, Milan, Italy; 3Division of Diagnostic Hematopathology, IEO European Institute of Oncology IRCCS, Milan and European Institute of Oncology IRCCS, Milan, Italy; 4Hematopathology Unit, Department of Experimental, Diagnostic, and Specialty Medicine (DIMES), Bologna University School of Medicine, Bologna, Italy; 5NanoString Technologies Inc, Seattle, WA, USA; 6Laboratory of Hematology-Oncology, IEO European Institute of Oncology IRCCS, Milan and European Institute of Oncology IRCCS, Milan, Italy; 7Division of Hematology, Fondazione IRCCS Istituto Nazionale dei Tumori, University of Milan, Milan, Italy; 8Department of Pathology, Fondazione IRCCS Istituto Nazionale dei Tumori di Milano, Milan, Italy; 9Multidisciplinary Oncology Outpatient Clinic, Candiolo Cancer Institute, FPO-IRCCS, Candiolo, Italy; 10Hematology and Bone Marrow Transplant Unit, ASST-Papa Giovanni XXIII, Bergamo, Italy and 11Institute of Hematology and Medical Oncology “L. e A. Seragnoli”, Department of Experimental, Diagnostic, and Specialty Medicine (DIMES), Bologna University School of Medicine, Bologna, Italy

ABSTRACT

ecent randomized trials focused on gene expression-based determination of the cell of origin in diffuse large B-cell lymphoma could not show significant improvements by adding novel agents to standard chemoimmunotherapy. The aim of this study was the identification of a gene signature able to refine current prognostication algorithms and applicable to clinical practice. Here we used a targeted gene expression profiling panel combining the Lymph2Cx signature for cell of origin classification with additional targets including MYC, BCL-2 and NFKBIA, in 186 patients from two randomized trials (discovery cohort) (clinicaltrials gov. Identifier: NCT00355199 and NCT00499018). Data were validated in three independent series (two large public datasets and a real-life cohort). By integrating the cell of origin, MYC/BCL-2 double expressor status and NFKBIA expression, we defined a three-gene signature combining MYC, BCL-2 and NFKBIA (MBN-signature), which outperformed the MYC/BCL-2 double expressor status in multivariate analysis, and allowed further risk stratification within the germinal center B-cell/unclassified subset. The high-risk (MBN Sig-high) subgroup identified the vast majority of double hit cases and a significant fraction of activated B-cell-derived diffuse large B-cell lymphomas. These results were validated in three independent series including a cohort from the REMoDL-B trial, where, in an exploratory ad hoc analysis, the addition of bortezomib in the MBN Sig-high subgroup provided a progression free survival advantage compared with standard chemoimmunotherapy. These data indicate that a simple three-gene signature based on MYC, BCL-2 and NFKBIA could refine the prognostic stratification in diffuse large B-cell lymphoma, and might be the basis for future precision-therapy approaches.

haematologica | 2021; 106(9)

Correspondence: ENRICO DERENZINI enrico.derenzini@ieo.it STEFANO PILERI stefano.pileri@ieo.it Received: August 25, 2019. Accepted: August 3, 2020. Pre-published: August 13, 2020. https://doi.org/10.3324/haematol.2019.236455

2405

E. Derenzini et al.

Introduction The biologic complexity of diffuse large B-cell lymphoma (DLBCL) was first dissected in the early 2000s by gene expression profiling (GEP) studies, which subdivided DLBCL into two groups based on GEP signatures reminiscent of the respective cell of origin (COO). These studies showed that DLBCL with a gene signature related to activated B lymphocytes (ABC subgroup) had a significantly worse response to anthracycline-based therapies compared to those histogenetically related to germinal center B cells (GCB subtype), and were dependent on nuclear factor k-B (NF-kB) signaling.1-3 Since immunohistochemical algorithms failed to reproduce the results of GEP,4-10 the Lymphoma Leukemia Molecular Profiling Project (LLMPP) proposed a targeted GEP (T-GEP) panel (Lymph2Cx) desumed from previous studies on fresh/frozen tissue (FFT).11,12 This assay was applied on the NanoString platform to formalin-fixed, paraffin embedded (FFPE) tissue from DLBCL patients treated with R-CHOP,11,12 identifying three subgroups: GCB, ABC and unclassified, the latter representing about 15% of all cases and prognostically closer to the GCB.11,13 The reproducibility of this assay was confirmed in several studies.131-5 However, recent results from three independent phase III randomized trials16-18 based on COO classification were largely negative. Although these unsatisfactory results could be due to several reasons, including unexpected toxicities and suboptimal efficacy of these drugs in vivo, these data also indicate that the clinical development of predictive T-GEP signatures able to complement the COO for precision therapy approaches is an urgent unmet need. Besides the COO, current evidence indicates a negative prognostic value of double MYC and BCL-2 protein overexpression determined by immunohistochemistry (IHC).19-21 Furthermore those DLBCL with concurrent MYC and BCL-2 and/or BCL-6 genomic rearrangements are characterized by an even worse prognosis, being now classified as a separate entity, high-grade B-cell lymphoma (HG-BCL) with double/triple hits (w DH/TH).19,20,22 Recently large genomic studies integrating DNA and RNA sequencing data identified additional DLBCL subgroups beyond the COO and MYC/BCL-2 double expressor (DE) status,23-25 based on the mutational landscape, GEP signatures, copy number changes, and differences in outcome. Furthermore, recent studies identified GEP signatures able to define high-risk populations within the GCB/unclassified (GCB/U) subgroup.26,27 However, given their complexity, large-scale application of these prognostication algorithms could be difficult in daily clinical practice. The aim of this study was the implementation of a simple T-GEP panel able to complement and improve COO-based prognostic stratification for routine clinical application. We designed a panel of genes corresponding to those of the Lymph2Cx assay for COO determination plus additional candidates selected because of their potential prognostic and/or therapeutic interest including MYC, BCL-2 and central nodes of NF-kB, Janus kinase (JAK)/signal transducer and activator of transcription (STAT), and phosphatidylinositol-3 kinase (PI3K) signaling.3,28-33 This panel of genes was applied to 186 DLBCL enrolled in two recently reported large Italian trials (DLCL04 and R-HDS0305; clinicaltrials gov. Identifier: NCT00355199 and NCT00499018).34,35 We found that a three-gene signature based on MYC, BCL-2 and NFKBIA (MBN signature), identified a significant fraction of ABC cases and a sub2406

group of GCB/U cases (roughly 30%) enriched in HG-BCL w/DH, at increased risk of treatment failure. These data were validated in a real-life cohort and in silico in two large independent series, including one cohort of patients enrolled in the REMoDL-B trial,18,27 where the addition of bortezomib to chemoimmunotherapy provided a significant advantage for high-risk patients identified by the MBN signature.

Methods Study design Patients considered in this study had been enrolled in two prospective randomized phase III clinical trials investigating the role of first line autologous stem cell transplant (ASCT) consolidation in intermediate/high-risk DLBCL.34,35 Only cases of DLBCL not-otherwise specified (NOS) (including those originally diagnosed as DLBCL and nowadays included in the HG-BCL provisional category22) were selected for the present study (Figure 1). Patients’ characteristics and study algorithm are summarized in Table 1 and Figure 1. Results were validated in three independent cohorts, (two in silico validation datasets and one “real-life” cohort): a dataset from Sha and coworkers (n=928 patients: 469 treated with RCHOP and 459 with R-CHOP plus bortezomib [RB-CHOP]);27 a public dataset from Lenz et al.36 (n=233 patients treated with RCHOP); a “real-life” cohort including 102 consecutive DLBCLNOS cases with available FFPE tissue, treated with R-CHOP/RCHOP-like regimens in Bologna (S.Orsola-Malpighi Hospital), and in Milan (European Institute of Oncology) from 2007 to 2018. This study was approved by the Institutional Review Boards and Ethics Committees of the participating centers, in accordance with the Declaration of Helsinki.

Procedures Gene expression was measured on the NanoString nCounter Analysis System (NanoString Technologies, Seattle, USA). The TGEP panel contains 26 genes: 15 genes for COO subtyping;11 five housekeeping genes (UBXN4, ISY1, R3HDM1, WDR55, TRIM56); and six additional genes (MYC, BCL-2, STAT3, NFKBIA, PTEN, PIK3CA). Besides MYC and BCL-2, the additional genes were selected based on their known functions in key pathways involved in DLBCL lymphomagenesis and potential druggability.

Statistical analysis Survival data were analyzed retrospectively. We used KaplanMeier method37 for overall survival (OS) and progression-free survival (PFS) analyses. Multivariate and univariate analyses were constructed with the Cox proportional hazards regression model. A P-value ≤0.05 was considered statistically significant. Recursive Partitioning Analysis (RPA)38 was applied to classify patients into more homogenous prognostic groups based on survival. All analyses were performed using R 3.5.0 software.39 Correlations and differences in patient characteristics were analyzed with the χ2 and Fisher’s exact test.

Development of the three-gene prognostic signature (MBN signature) An expression ratio-based test was developed by selecting those genes significantly deregulated in the high risk subgroups identified by the RPA shown in Figure 2A and whose normalized mRNA levels were significantly associated with OS. We defined high and low MYC and BCL-2 expressors based on the median normalized MYC and BCL-2 mRNA levels. The high-risk groups

haematologica | 2021; 106(9)

A 3-gene signature in DLBCL

included the ABC and double expressor GCB/unclassified (GCB/U) DLBCL (hereafter defined as DEXP_mRNA); the low risk group was constituted by the non-DEXP_mRNA GCB/U subset. Since the expression levels of MYC and BCL-2 on one hand and NFKBIA on the other hand had opposing patterns being inversely associated with OS (with higher MYC/BCL-2 and lower NFKBIA levels associated with worse outcome), we combined the expression levels of the three genes in a synthetic predictor called MBNsignature (MBN-Sig) and defined as:

MBN-Sig= (MYC + BCL-2)/NFKBIA Detailed information on study cohorts (Online Supplementary Table S1), T-GEP procedures with list of genes and target sequences, fluorescence in situ hybridization (FISH), IHC methods and antibodies (Online Supplementary Table S2), and random forest (RF) classifier are described in the Online Supplementary Appendix.

Results Univariate analyses and a decision-tree classification model integrating the cell of origin and MYC/BCL-2 status Given their established clinical relevance, we first investigated the prognostic significance of T-GEP-based COO classification and MYC/BCL-2 status in the R-HDS0305 and DLCL04 trials34,35 (discovery cohort). Patient’s characteristics are summarized in Table 1. In line with previous findings11,12 COO classification by T-GEP clearly outperformed the immunohistochemical Hans algorhitm for survival prediction and retained its prognostic significance in the presence or absence of ASCT consolidation (Online Supplementary Figure S1A to D). In order to investigate the prognostic impact of concurrent overexpression of MYC and BCL-2, we defined high and low expressors based on

Figure 1. Study algorithm. On the left, the discovery cohort is represented; 224 diffuse large B-cell lymphoma (DLBCL) patients enrolled in the DLCL04 (n= 130) and R-HDS0305 (n= 94) trials with available formalin-fixed, paraffin embedded (FFPE) tissue were initially considered in this analysis. Targeted gene expression profiling (TGEP) success rate was 92.4% (n=207), with 17 cases not yelding enough highquality mRNA to undergo successful GEP assessment. Only cases originally diagnosed as DLBCL non-otherwise specified (NOS) were considered. Therefore 21 cases classified in different DLBCL categories were excluded; 99 NOS-DLBCL FFPE patient samples from the DLCL04 trial and 87 samples from the R-HDS0305 trial were finally included in this study. On the right, the three validation cohorts: a cohort of 928 patients from Sha and coworkers27 (469 treated with R-CHOP; 459 with RBCHOP), a public gene expression dataset (Affymetrix Human Genome U133 Plus 2.0 Array), GSE10846, (https://www.ncbi.nlm. nih.gov/geo/query/acc.cgi?acc=GSE1 0846), including 233 patients treated with R-CHOP regimen (Lenz et al. 2008)36; an additional validation cohort including 102 consecutive DLBCL NOS cases with available FFPE tissue, treated with R-CHOP/R-CHOP-like regimens. RB-CHOP: R-CHOP plus bortezomib.

haematologica | 2021; 106(9)

2407

E. Derenzini et al.

the median normalized MYC and BCL-2 mRNA levels, which correlated well with the respective protein levels assessed by IHC (Online Supplementary Figure S2A). MYC/BCL-2 mRNA double expressors (defined as DEXP_mRNA) patients showed a worse outcome compared to non-DEXP_mRNA cases (Online Supplementary Figure S2B). Although DEXP_mRNA cases were more prevalent in the ABC compared to the GCB/unclassified (GCB/U) subgroup9 (Online Supplementary Table S3), the

2408

prognostic relevance of the MYC/BCL-2 DEXP_mRNA status was particularly evident in the GCB/U subset (Online Supplementary Figure S2C to F). Focusing the analysis on the additional genes (STAT3, NFKBIA, PTEN, PIK3CA), which were selected based on their biologic relevance in potentially druggable pathways, only NFKBIA and STAT3 mRNA levels were significantly associated with patient’s outcome, with low STAT3 and low NFKBIA expression predicting worse prognosis (Online

Figure 2. Integrating cell of origin with MYC/BCL-2 DEXP_mRNA status for prognostication in diffuse large Bcell lymphoma. (A) Recursive partitioning analysis integrating cell of origin (COO) classification and DEXP_mRNA status, allowing segregation of patients in three main prognostic subgroups (a low risk non-DEXP-mRNA GCB/U subset, and two high risk groups: MYC/BCL-2-DEXP-mRNA GCB/U and ABC). (B) Box plot graphs indicating the expression levels of the additional targets included in the panel (MYC, BCL-2, NFKBIA, STAT3, PIK3CA, PTEN) in the three main patients subgroups identified by the recursive partitioning analysis (non-DEXPmRNA GCB/U, MYC/BCL-2 DEXP_mRNA GCB/U, ABCderived diffuse large B-cell lymphoma [DLBCL]. P-value was calculated with Student ttest by comparing non DEXP_mRNA GCB/U group, selected as a reference, versus other groups. ABC: activated B-cells; GCB: germinal center B cells; GCB/U: GCB unclassified; DEXP_mRNA: double expressor GCB/U DLBCL. COO Nano: COO as determined by T-GEP with NanoString profiling

haematologica | 2021; 106(9)

A 3-gene signature in DLBCL

Supplementary Figures S3A and B). In univariate analyses only the age adjusted International Prognostic Index (aaIPI) score (intermediate-high vs. high), the COO classification, MYC/BCL-2-DE status, NFKBIA and STAT3 levels determined by T-GEP, were significantly associated with OS (Table 2). As observed in the original studies,34,35 first-line ASCT consolidation was not associated with patient’s outcome. In line with the data presented above (Online Supplementary Figure S1 and S2), a recursive partitioning analysis integrating the COO with MYC/BCL-2 status identified three main patient subgroups: two high risk subsets with similar outcome (ABC [n=40) and MYC/BCL-2 DEXP_mRNA GCB/U [n=27]) and a low-risk subgroup including non-DEXP_mRNA GCB/U DLBCL, (n=119) (Figure 2A). Evaluating the relative expression of the additional genes included in the panel across the three groups identified by the recursive partitioning analysis (non-DEXP_mRNA GCB/U, DEXP_mRNA GCB/U and ABC DLBCL patients) (Figure 2B), we found that only MYC, BCL-2 and NFKBIA were significantly deregulated in both the high-risk ABC and MYC/BCL-2 DEXP_mRNA GCB/U subgroups, which were characterized by similarly increased MYC and BCL-2 and lower NFKBIA mRNA levels compared to the low risk non-DEXP_mRNA GCB/U subset. The NFKBIA gene, a frequent target of deletions and mutations in DLBCL,23 encodes for the IkB-α protein, which is a central node of the NF-kB pathway and inhibits nuclear translocation and activity of the NF-kB transcription factors.40 STAT3 levels were similar in the high risk ABC and low risk non-DE GCB/U cases being significantly downregulated only in the DEXP_mRNA GCB/U subset. PIK3CA and PTEN levels did not vary significantly across different groups (Figure 2B).

Development of a three-gene prognostic signature combining MYC, BCL-2 and NFKBIA In an effort to build a GEP signature aimed at refining current prognostication algorithms and suitable for clinical

practice, we considered only those genes whose expression was significantly associated with OS and differentially represented in both high risk (ABC and DEXP_mRNA GCB/U) versus low risk (non-DEXP mRNA GCB/U) patient subsets. Using these criteria, we constructed a prognostic signature considering three genes (MYC, BCL2 and NFKBIA), which combines the MYC/BCL-2 DEXP_mRNA status with NFKBIA expression (hereafter called MBN signature, see methods). Besides MYC and BCL-2, (defining the DEXP_mRNA status), NFKBIA emerged as the best survival predictor by gene ranking according to the predictive power (univariate z score) (Online Supplementary Figure S4). With this strategy, patients were divided in two risk categories characterized by different outcome: low risk patients (MBN-Sig low) had a very favorable prognosis (91% 5-year OS; 84% 5year PFS), whereas high-risk patients (MBN-Sig high) had a significantly worse prognosis (64% 5-year OS; 59% 5year PFS) (Figure 3A; Online Supplementary Figure S5A). Importantly the MBN signature retained its significance and outperformed the MYC/BCL-2 DEXP_mRNA status in multivariate analysis (Figure 3B; Online Supplementary Table S4). In fact, only the COO, the aaIPI score and the MBN signature were significantly associated with outcome in multivariate analyses (Figure 3B). These findings were confirmed in silico in a large independent validation cohort of 469 patients27 treated with R-CHOP (88% 5-year OS and 78% PFS for MBN-Sig low vs. 72% OS and 57% PFS for MBN-Sig high patients) (Figure 3C and D; Online Supplementary Figure S5B; Online Supplementary Table S5). The prognostic value of the MBN signature was further tested in a publicly available data set including 233 patients (from Lenz at al. 2008)36 treated with R-CHOP/R-CHOP-like regimens and in a real-life cohort (n=102 patients) with similar results (Online Supplementary Figure S6A and B). The MBN signature was able to identify a significant fraction of ABC-derived cases and about a third of GCB/U cases (Figure 3A and C; Online Supplementary Figure S6C and D).

Table 1. Patients characteristics.

Trial name N° of patients Immuno-CHT alone Immuno-CHT + ASCT Median age, y (range) COO NanoString ABC GCB Unclassified COO Hans IHC Non-GCB GCB-like Stage (Ann Arbor) aaIPI score Low-Low Intermediate (0-1) Intermediate-high (2) High (3)

RHDS0305

DLCL04

RHDS0305 + DLCL04

87 49 (56%) 38 (44%) 53 (21-65)

99 56 (57%) 43 (43%) 52(18-65)

186 105 (56%) 81 (44%) 52 (18-65)

15 (17%) 53 (61%) 19 (22%)

25 (25%) 58 (59%) 16 (16%)

ns ns

40 (21%) 111 (60%) 35 (19%)

58 (67%) 29 (33%) II-IV

60 (61%) 39 (39%) III-IV

ns -

118 (63%) 68 (37%) II-IV

55 (63%) 32 (37%)

81 (82%) 18 (18%)

ns 0.005

136 (73%) 50 (27%)

*Two-sided Fisher’s exact test: N°: number; Immuno-CHT: immunochemotherapy; ASCT: autologous stem cell transplantation; y: years; COO: cell of origin; IHC: immunohistochemistry; aaIPI: age adjusted international prognostic index; ns: not significant; GCB: germinal center B cells; ABC: activated B-cells; IHC: immunohistochemistry.

haematologica | 2021; 106(9)

2409

E. Derenzini et al.

Figure 3. Legend on next page.

2410

haematologica | 2021; 106(9)

A 3-gene signature in DLBCL

Figure 3. Survival curves according to MBN signature and multivariate analyses for overall survival. (A) Overall survival (OS) of the discovery cohort (RHDS0305+DLCL04; n=186 patients) according to the MBN signature (MBN-Sig) showing significant differences in outcome between MBN-Sig low versus MBN-Sig high patient subsets. P-values were calculated with the log rank test. Frequencies of MBN-Sig high versus low cases in activated B-cells (ABC) and germinal center B cells/ unclassified (GCB/U) subsets in the discovery cohort are represented in the pie chart. (B) Forest plot depicting multivariate analyses for OS (discovery cohort). Only factors significantly asociated with OS in univariate analyses were considered. According to this analysis only the cell of origin (COO) as determined by NanoString-based targeted gene expression profiling (T-GEP) (COO_Nano), the MBN-Sig and the age adjusted international prognostic index (aaIPI) score retained statistical significance for OS, whereas MYC/BCL-2 DEXP_mRNA status, STAT3 and NFKBIA levels determined by T-GEP were not significantly associated with OS. HR: hazard ratio. (C) OS of the 469 patients treated with R-CHOP in the Sha’s cohort according to the MBN signature showing significant differences in outcome between MBN-Sig low versus MBN-Sig high patient subsets. P-values were calculated with the log rank test. Frequencies of MBN-Sig high versus low cases in ABC and GCB/U subsets in the Sha’s cohort are represented in the pie chart. (C) Forest plot depicting multivariate analyses for OS (Sha’s dataset), confirming the significant independent association with OS of the MBN-Sig in this large validation cohort.

Real life applicability of the MBN signature In order to provide a risk stratification tool applicable to routine clinical practice in a prospective manner, we constructed an RF model with the expression of genes characterizing the MBN signature. First, the classifier was trained on the discovery cohort splitting it into training (80%) and test (20%) dataset; in this case, the accuracy of the threegene model was 93% in the training and 94% in test set. In order to confirm the reliability of this three-gene model, we further tested it in an independent dataset (validation set) consisting of the real-life cohort (n=102 cases). Of note, these cases were profiled with the same T-GEP panel and methods used in the discovery cohort, mitigating batch effects phenomena. As result, the three-gene model accurately classified 85% (87 of 102) cases as either MBN-Sig high or MBN-Sig low subgroups (Figure 4A). As reported in Figure 4B, the model effectively identified MBN-high and low categories with sensitivity (SE) and specificity (SP) of 94% and 76% respectively. Receiver operating characteristic (ROC) curve analysis revealed that the area under the curve (AUC) was 0.94 in the validation set (Figure 4C). Furthermore, this strategy produced a very efficient survival prediction, which as expected showed a worse outcome for the MBN-high subset (Figure 4D) and mirrored the OS curve based on the median MBN value depicted in the Online Supplementary Figure S6B.

Correlation of the MBN signature with fluorescence in situ hybridization status and clinical variables Focusing the analyses on our discovery cohort of 186 patients (DLCL04 and R-HDS0305 trials),34,35 we observed that the MBN signature significantly stratified the prognosis GCB/U patients (Figure 5A). Since the MBN signature effectively stratified GCB/U DLBCL patients, we investigated correlations between the MBN-signature, FISH status and clinical variables in our discovery cohort. As shown in Figure 5B, we observed a significantly higher frequency of MYC and BCL-2 re-arrangements in the MBN-Sig high subgroup compared to the GCB/U MBNSig low subset. According to these observations, there was a significant enrichement of HG-BCL w/DH in the MBNSig high subgroup compared to the MBN-Sig low subset (Figure 5B; Online Supplementary Figure S7A). No differences in the number of cases with missing FISH analyses were observed between groups (data not shown). In line with the literature,23,26,27 all these cases, except one, were GCB-derived (data not shown). As previously shown in Figure 3, ABC-derived DLBCL were significantly more represented in the MBN-Sig high subgroup (Figure 5B; Online Supplementary Figure S7A). Finally, no significant differences in the aaIPI score (intermediate high vs. high) were observed between groups (Figure 5B; Online Supplementary Figure S7A). These findings were validated in silico in the larger cohort from Sha et al.27 (Figure 5C and D). As observed in the discovery cohort, the MBN signahaematologica | 2021; 106(9)

Table 2. Univariate analysis for overall survival. aaIPI Intermediate-High High COO Nano GCB ABC Unclassified ASCT No Yes MYC-BCL-2 DEXPmRNA No Yes STAT3 Low High NFKBIA Low High

Hazard Ratio

95% CI

Ref 2.04

1.10-3.78

0.023

1.76-6.52 0.47-2.96

<0.001 0.736

Ref 0.88

0.47-1.63

0.68

Ref 2.32

1.26-4.29

0.007

Ref 0.37

0.19-0.7

0.004

Ref 0.34

0.17-0.68

0.002

Ref 3.39 1.17

Signif: significance; ref: reference; IPI: international prognostic index; COO: cell of origin; COO_Nano: COO defined by NanoString; ASCT: autologous stem cell transplant; MYC/BCL-2 DEXP_mRNA: double expressor status defined by NanoString: CI: Confidence Interval. *P<0.05; **P<0.01.

ture stratified the prognosis of GCB/U patients (Figure 5C) and identified the vast majority of DH cases. Again ABCderived DLBCL were more highly represented in the MBN-Sig high subgroup (Figure 5D; Online Supplementary Figure S7B). In this study, the application of a gene expression classifier identified a molecular high grade (MHG) subgroup strongly enriched in DH lymphomas and comprising 9% of the total patient population.27 In order to evaluate how our MBN signature performed in the same patient population, we compared the MBN signature with the MHG signature and with the FISH status (Figure 5C). Notably the MBN-high subgroup was significantly enriched in MHG cases, identifying 76% of MHG DLBCL and the vast majority of DH (Figure 5D; Online Supplementary Figure S7B). Also in this cohort there were no differences in IPI score between groups (Figure 5D; Online Supplementary Figure S7B).

Rationale for a precision therapy approach in MBN-Sig high-risk diffuse large B-cell lymphoma patients Since the MBN-Sig high subgroup is characterized by relatively higher MYC and BCL-2 expression and lower NFKBIA levels indicative of constitutive NF-kB activity, we next investigated the effect of differential therapeutic strategies in this high-risk patient subset. We first analyzed the impact of ASCT versus standard chemoimmunotherapy in the discovery cohort. ASCT consolida2411

E. Derenzini et al.

tion did not provide any significant PFS or OS advantage compared to standard chemoimmunotherapy in the MBN-Sig high subgroup (Online Supplementary Figure S8A and B). The aberrant activation of NF-kB observed in lymphoma is associated with decreased abundance of IkB-α (which is encoded by the NFKBIA gene).41,42 Since bortezomib is known to increase IkB-α levels by blocking its

ubiquitination and therefore inhibiting NF-kB activity,43-45 we next examined the Sha dataset18,27 performing an exploratory ad hoc analysis to investigate the impact of the addition of bortezomib to standard R-CHOP (RB-CHOP) in the MBN-Sig high subset (characterized by decreased NFKBIA levels).18,27 Interestingly, RB-CHOP determined a significant PFS advantage in the MBN-Sig high population

Figure 4. Real-life applicability of the MBN signature. (A) Heatmap representing the three informative genes of the MBN signature (MBN-Sig) shown as rows and diffuse large B-cell lymphoma (DLBCL) tissue samples shown as columns in the real-life cohort of 102 patients, with the actual MBN-Sig and the predicted MBN-Sig class based on the application of a random forest (RF) model built on the discovery cohort on the top of the heatmap. (B) Violin plot showing the fractions of false predictions (false positive [FP], and false negative [FN]) as well as true predictions (true positive [TP], and true negative [TN]) in the real-life cohort by applying a three-gene RF model. (C) ROC curve of the real-life cohort using RF classifier. (D) Overall survival (OS) curve of the real-life cohort (n=102) based on the predicted MBN-Sig class. P-value was calculated with the log rank test.

2412

haematologica | 2021; 106(9)

A 3-gene signature in DLBCL

Figure 5. Legend on following page.

haematologica | 2021; 106(9)

2413

E. Derenzini et al.

Figure 5. The MBN signature identifies prognostically distinct subgroups including activated B cells and a fraction of germinal center B cells/ unclassified diffuse large B-cell lymphoma (DLBCL) enriched in double hit DLBCL cases, providing opportunities for precision therapies. (A) Overall survival (OS) of the germinal center B cells/ unclassified (GCB/U) subset in the discovery cohort (n=146 patients) according to the integration of the targeted gene expression profiling (T-GEP) panel (Lymph2Cx) with the MBN signature (MBN-Sig), distinguishing two risk categories according to the MBN-Sig. (B) Heatmap representing the three informative genes of the MBN-Sig shown as rows and diffuse large B-cell lymphoma (DLBCL) tissue samples shown as columns, in the discovery cohort (n=186 patients). (C) OS of the GCB/U subset in the Sha’s validation cohort (n=340 patients) according to the MBN-Sig, showing superimposable results compared to Figure 4C. (D) Heatmap representing the three informative genes of the MBN-Sig shown as rows and DLBCL tissue samples shown as columns, in the Sha’s cohort (n=469 patients treated with R-CHOP). (E) Progression-free survival (PFS) of patients treated with R-CHOP versus RB-CHOP in the MBN-Sig high subgroup (Full Sha’s cohort, n=928 patients; MBNSig high n=464 patients). (F) OS of patients treated with R-CHOP versus RB-CHOP in the MBN-Sig high subgroup (Full Sha’s cohort, n=928 patients; MBN-Sig high n=464 patients). In all panels the P-value was calculated with the log rank test. NA: not available; DH: double hit; DE: double expressor (based on DEXP_mRNA status); aaIPI: age adjusted international prognostic index; COO: cell of origin; IPI: international prognostic index); MHG: molecular high grade. RB-CHOP: R-CHOP plus Bortezomib.

(P=0.012) (Figure 5E), which translated in an increased OS rate (P=0.052) (Figure 5F).

Discussion In this study we applied a customized T-GEP panel (including the Lymph2Cx signature for COO classification and additional genes of potential prognostic and therapeutic interest) to two randomized trials34,35 (n=186 patients) performed in the Rituximab era. The aims of this study were the integration of the COO with additional GEPbased variables, and the identification of a gene signature applicable to routine clinical practice, able to refine current prognostication algorithms. The genes of the T-GEP panel were selected considering the relevance of the respective signaling pathways in B-cell lymphomagenesis, but more importantly based on their potential druggability. Our study confirmed the prognostic value of GEP-based COO determination, which clearly outperformed the IHC-based Hans algorithm (the ABC DLBCL subgroups having a significantly inferior OS in all case series evaluated here) (Online Supplementary Figure S1). The COO retained its prognostic value in patients undergoing ASCT consolidation, suggesting that therapy intensification is not able to overcome the negative prognostic value of the COO. A recursive partitioning analysis integrating COO with MYC/BCL-2 DEXP_mRNA status identified three main subgroups (a low risk non-DEXP_mRNA GCB/U subset and two high-risk groups including DEXP_mRNA GCB/U and ABC-DLBCLs) (Figure 2A). The observation of lower NFKBIA levels in the ABC and DEXP_mRNA GCB/U subgroups (overexpressing MYC and BCL-2 to a similar extent) (Figure 2B) suggests that, despite known biologic differences, these DLBCL subsets could share similar oncogenic dependencies on MYC, BCL-2 and the NF-kB pathway (being NFKBIA a negative regulator of NF-kB signaling). This observations prompted us to design a three-gene prognostic signature integrating MYC, BCL-2 and NFKBIA, which we called the MBN signature. The signature was first tested in our discovery cohort of 186 patients, identifying two subgroups characterized by different outcome (Figure 3), and was then applied to three independent datasets (469 patients treated with R-CHOP in the Sha cohort,27 233 patients from the Lenz cohort,36 and 102 patients treated in real-life clinical practice with R-CHOP/R-CHOP-like regimens) confirming its high prognostic significance (total number of tested cases 990). Since the discovery cohort had some unique characteristics (such as lack of low aa-IPI cases, a relatively low fraction of ABC cases and no uniform first-line treatment), the extensive validation performed in three additional cohorts treated with R-CHOP/R-CHOP-like regimens confirms 2414

that the key findings of the present study are indeed applicable to an unselected DLBCL population. Importantly, the MBN signature defined a high-risk group including a significant fraction of ABC cases (in line with data shown in the Online Supplementary Table S3 demonstrating a higher incidence of MYC/BCL-2 DEXP_mRNA and low NFKBIA expressors in the ABC subgroup), and about 30% of GCB/U cases (Figure 3). Therefore the MBN signature could potentially identify an increased proportion of patients at high risk of treatment failure, compared to standard risk stratifications (COO or DE status). The MBN signature was an independent prognostic predictor, outperforming the MYC/BCL-2-DEXP_mRNA status in multivariate analyses (Figure 3), thus confirming the added value of the third gene (NFKBIA) for prognostic stratification. The possible clinical applicability of the MBN signature was tested in the real-life cohort using an RF prediction model built on the discovery cohort, providing a reliable tool for prospective risk stratification (Figure 4). Importantly, the integration of the MBN signature with the COO allowed the identification of two risk categories whithin the GCB/U subset. These findings, which were validated in independent cohorts, could have immediate implications (Figure 5A and C; Online Supplementary Figure S6A to D). Two recently published studies confirmed the heterogeneity of the GCB subgroup and identified gene signatures allowing better risk stratification of this patient subset.26,27 These signatures were able to identify a proportion of HG-BCL with DH/TH and a further group lacking MYC/BCL-2 re-arrangements but characterized by similar clinical features. However, the fact that these signatures are composed by several genes encompassing multiple pathways, could make their successful translation to clinical practice and precision therapy approaches quite challenging. Our data are in line with these findings confirming that the GCB/U DLBCL subset represents indeed a rather heterogeneous disease category. The MBN signature could identify the majority of tumors with high-grade molecular features (HG-BCL with DH/TH) in the discovery cohort and Sha’s cohort (Figure 5B and D; Online Supplementary Figure S7A and B). Moreover, by applying the MBN signature to the Sha validation cohort we observed that the MBN-Sig high subroup was significantly enriched in MHG DLBCL cases (Figure 5D; Online Supplementary Figure S7B). Taken together, these data indicate that a simple there-gene signature could efficiently identify high risk GCB/U DLBCL cases. Furthermore, the MBN-signature is based on potentially druggable targets or pathways. For example, NFKBIA (encoding for IkB-α) could be targeted by proteasome inhibitors43-45 and by bromodomain and extraterminal protein (BET) inhibitors, which are able to downregulate MYC while increasing IkB-α levels.46-48 Our haematologica | 2021; 106(9)

A 3-gene signature in DLBCL

analysis on the impact of bortezomib in the MBN-high subgroup of the Sha cohort27 (from the REMoDL-B trial) seems to confirm a potential druggability of the MBN signature: in fact treatment with RB-CHOP (R-CHOP plus bortezomib) was associated with a significantly prolonged PFS which translated in increased OS rates in the MBN-high subgroup, as compared to standard R-CHOP (Figure 5E and F). Proteasome inhibitors, BET inhibitors and selective BCL-2 inhibitors could be the basis for rationally-designed combinations for the MBN-Sig high DLBCL subgroup. Alternative strategies to target NF-kB include lenalidomide and B-cell receptor signaling inhibitors, all of which are under clinical investigation in DLBCL. Three COO-based phase III trials testing RCHOP + Ibrutinib (Phoenix trial16) or Lenalidomide (ROBUST trial17) or bortezomib (REMoDL-B trial18) did not meet their primary endpoints. Although several factors concurred to these negative results, the development of alternative and druggable molecular signatures represents an unmet need and could be of primary importance for the design of future precision medicine clinical trials. The results of our study indicate that a simple and costeffective three-gene assay (MBN signature) could refine current prognostic stratification algorithms providing the rationale for the implementation of precision medicine trials in the MBN-Sig high subset. Disclosures ACe and NC are employees and shareholders of NanoString technology; ED has received research funding from TGTherapeutics, ADC-Therapeutics, Takeda and sits on the Advisory Board for Gilead; ES has received support from Novartis and Eusapharma for educational events; ACh sit on the Advisory Boards with Celgene, Gilead-Kite, Janssen, Iqone, Takeda and has received honoraria for lectures from Celgene, Gilead-Kite, Janssen, Roche, Servier; UV has a consulting or advisory role for Celgene, Gilead and Janssen and is part of the speakers’ bureau with Roche, Celgene, Janssen, Gilead

References 1. 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. 2. Shipp MA, Ross KN, Tamayo P, et al. Diffuse large B-cell lymphoma outcome prediction by gene-expression profiling and supervised machine learning. Nat Med. 2002;8(1):68-74. 3. Davis RE, Brown KD, Siebenlist U, Staudt LM. Constitutive nuclear factor kappaB activity is required for survival of activated B cell-like diffuse large B cell lymphoma cells. J Exp Med. 2001;194(12):1861-1874. 4. 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. 5. Meyer PN, Fu K, Greiner TC, et al. Immunohistochemical methods for predicting cell of origin and survival in patients with diffuse large B-cell lymphoma treated with rituximab. J Clin Oncol. 2011; 29(2):200-207. 6 Lawrie CH, Ballabio E, Soilleux E, et al. Interand intra-observational variability in immunohistochemistry: a multicentre analysis of diffuse large B-cell lymphoma staining. Histopathology. 2012;61(1):18-25. 7. Castillo JJ, Beltran BE, Song MK, et al. The

haematologica | 2021; 106(9)

Sciences, Abbvie, Sandoz; AR sits on the National or International Advisory Boards for Gilead, Amgen, Novartis, Pfizer, Celgene, Italfarmaco, Sanofi-Aventis, Astellas, Roche, Omeros and has sponsored symposia for Amgen, Novartis, Celgene, Roche; PC has received honoraria for Advisory Board participation or as a lecturer from AbbVie, Amgen, Celgene, Daiichi Sankyo, Gilead, Incyte, Janssen, Kite, KiowaKirin, Novartis, Roche, Sanofi, Servier, Takeda; PLZ has received honoraria for speakers' bureau or Advisory Boards for Verastem, Celltrion, Gilead, Janssen-Cilag, BMS, Servier, Sandoz, MSD, Immune Design, Celgene, Portola, Roche, Eusapharma, Kyowa Kirin, Sanofi; CT sit on the Advisory Board for ADCTherapeutics; SP sits on the Advisory Boards for Celgene, NanoString, Roche; SM, FM, GM, MF, RB, CA, CC, SR, ACa, SF, VT, ACab, GP, AMG have no conflicts of interest to disclose. Contributions ED and SP designed the study, interpreted the data and wrote the manuscrip; SM and MF performed bioinformatics and statistical analyes, and SM helped with manuscript writing; FM and GM performed T-GEP experiments; VT, SF, CA and ACa performed immunohistochemistry; SP, ES, VT, SF and CA evaluated immunohistochemistry data; CC and SR performed FISH analyses; ACe and NC helped designing T-GEP experiments and helped with data interpretation; RB helped with data collection; ACab, GP, CT, AMG, PLZ, AR, PC, UV and ACh helped with data collection and interpretation. All authors critically reviewed the draft and approved the manuscript. Aknowledgments The authors wish to thank Pier Luigi Antoniotti, Sebastiano Spagnolo, Marco Giuffrida and Virginia Maltoni for technical assistance. Funding This study was funded by the AIRC 5x1000 grant to SP (n. 21198) and Italian Ministry of Health with Ricerca Corrente.

Hans algorithm is not prognostic in patients with diffuse large B-cell lymphoma treated with R-CHOP. Leuk Res. 2012;36(4):413-417. 8. Coutinho R, Clear AJ, Owen A, et al. Poor concordance among nine immunohistochemistry classifiers of cell-of-origin for diffuse large B-cell lymphoma: implications for therapeutic strategies. Clin Cancer Res. 2013;19(24):6686-6695. 9. Hu S, Xu-Monette ZY, Tzankov A, et al. MYC/BCL2 protein coexpression contributes to the inferior survival of activated Bcell subtype of diffuse large B-cell lymphoma and demonstrates high-risk gene expression signatures: a report from The International DLBCL Rituximab-CHOP Consortium Program. Blood. 2013; 121(20): 4021-4031. 10. Reinke S, Richter J, Fend F, et al. Round-robin test for the cell-of-origin classification of diffuse large B-cell lymphoma-a feasibility study using full slide staining. Virchows Arch. 2018;473(3):341-349. 11. Scott DW, Wright GW, Williams PM, et al. Determining cell-of-origin subtypes of diffuse large B-cell lymphoma using gene expression in formalin-fixed paraffin-embedded tissue. Blood. 2014;123(8):1214-1217. 12. Scott DW, Mottok A, Ennishi D, et al. Prognostic significance of diffuse large B-cell lymphoma cell of origin determined by digital gene expression in formalin-Fixed paraf-

fin-embedded tissue biopsies. J Clin Oncol. 2015;33(26):2848-2856. 13. Painter D, Barrans S, Lacy S, et al. Cell-of-origin in diffuse large B-cell lymphomafindings from the UK’s population-based Haematological Malignancy Research Network. Br J Haematol. 2019; 185(4):752– 806. 14. Veldman-Jones MH, Lai Z, Wappett M, et al. Reproducible, quantitative, and flexible molecular subtyping of clinical DLBCL samples using the NanoString nCounter System. Clin Cancer Res. 2015; 21(10):2367-2378. 15. Rimsza LM, Wright G, Schwartz M, et al. Accurate classification of diffuse large B-cell lymphoma into germinal center and activated B-cell subtypes using a nuclease protection assay on formalin-fixed, paraffinembedded tissues. Clin Cancer Res. 2011;17(11):3727-3732. 16. Younes A, Sehn LH, Johnson P, et al. Randomized phase III trial of ibrutinib and rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone in non-germinal center B-cell diffuse large B-cell lymphoma. J Clin Oncol. 2019;37(15):1285-1295. 17. Vitolo U, Witzig T, Gascoyne R, et al. ROBUST: first report of phase III randomized study of lenalidomide/R CHOP (R2 CHOP) vs placebo/R CHOP in previously untreated ABC type diffuse large B cell

2415

E. Derenzini et al. lymphoma. Hematol Oncol. 2019; 37(S2):3637. 18. Davies A, Cummin TE, Barrans S, et al. Gene-expression profiling of bortezomib added to standard chemoimmunotherapy for diffuse large B-cell lymphoma (REMoDLB): an open-label, randomised, phase 3 trial. Lancet Oncol. 2019;20(5):649-662. 19. Green TM, Young KH, Visco C, et al. Immunohistochemical double-hit score is a strong predictor of outcome in patients with diffuse large B-cell lymphoma treated with rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone. J Clin Oncol. 2012;30(28):3460-3467. 20. Johnson NA, Slack GW, Savage KJ, et al. Concurrent expression of MYC and BCL2 in diffuse large B-cell lymphoma treated with rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone. J Clin Oncol. 2012;30(28):3452-3459. 21. taiger AM, Ziepert M, Horn H, et al. Clinical impact of the Cell-of-Origin Classification and the MYC/ BCL2 dual expresser status in diffuse large B-cell lymphoma treated within prospective clinical trials of the German High-Grade Non-Hodgkin's Lymphoma Study Group. J Clin Oncol. 2017;35(22):2515-2526. 22. Swerdlow SH, Campo E, Harris NL, et al. WHO Classification of Tumour of Haematopoietic and Lymphoid Tissues, Revised 4th edition. 2017. IARC Press, Lyon. 23. Reddy A, Zhang J, Davis NS, et al. Genetic and functional drivers of diffuse large B cell lymphoma. Cell. 2017;171(2):481-494. 24. 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. 25. 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(8):679-690. 26. Ennishi D, Jiang A, Boyle M, et al. Double-hit gene expression signature defines a distinct subgroup of germinal center B-cell-like diffuse large B-cell lymphoma. J Clin Oncol. 2019;37(3):190-201.

2416

27. Sha C, Barrans S, Cucco F, et al. Molecular high-grade B-cell lymphoma: defining a poor-risk group that requires different approaches to therapy. J Clin Oncol. 2019;37(3):202-212. 28. Pasqualucci L, Dalla-Favera R. Genetics of diffuse large B-cell lymphoma. Blood. 2018; 131(21):2307-2319. 29. Roschewski M, Staudt LM, Wilson WH. Diffuse large B-cell lymphoma-treatment approaches in the molecular era. Nat Rev Clin Oncol. 2014;11(1):12-23. 30. Paul J, Soujon M, Wengner AM, et al. Simultaneous inhibition of PI3Kδ and PI3Kα induces ABC-DLBCL regression by blocking BCR-dependent and -independent activation of NF-κB and AKT. Cancer Cell. 2017;31(1):64-78. 31. Pfeifer M, Grau M, Lenze D, et al. PTEN loss defines a PI3K/AKT pathway-dependent germinal center subtype of diffuse large B-cell lymphoma. Proc Natl Acad Sci U S A. 2013;110(30):12420-12425. 32. Compagno M, Lim WK, Grunn A, et al. Mutations of multiple genes cause deregulation of NF-kappaB in diffuse large B-cell lymphoma. Nature. 2009;459(7247):717-721. 33. Pan YR, Chen CC, Chan YT, et al. STAT3coordinated migration facilitates the dissemination of diffuse large B-cell lymphomas. Nat Commun. 2018;9(1):3696. 34. Cortelazzo S, Tarella C, Gianni AM, et al. Randomized trial comparing R-CHOP versus high-dose sequential chemotherapy in highrisk patients with diffuse large B-cell lymphomas. J Clin Oncol. 2016; 34(33):40154022. 35. Chiappella A, Martelli M, Angelucci E, et al. Rituximab-dose-dense chemotherapy with or without high-dose chemotherapy plus autologous stem-cell transplantation in highrisk diffuse large B-cell lymphoma (DLCL04): final results of a multicentre, open-label, randomised, controlled, phase 3 study. Lancet Oncol. 2017;18(8):1076-1088. 36. Lenz G, Wright G, Dave SS, et al. Stromal gene signatures in large-B-cell lymphomas. N Engl J Med 2008;359(22):2313-2323. 37. Kaplan EL, Meier P. Nonparametric estimations from incomplete observations. J Am

Stat Assoc.1958;53(282):457-481. 38. Curran WJ Jr, Scott CB, Horton J, et al. Recursive partitioning analysis of prognostic factors in three Radiation Therapy Oncology Group malignant glioma trials. J Natl Cancer Inst. 1993;85(9):704-710. 39. R Core Team. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2014. URL http://www.R-project.org/. 40. Oeckinghaus A, Ghosh S. The NF-κB family of transcription factors and its regulation. Cold Spring Harbor Perspectives in Biology. 2009;1(4):a000034. 41. Jost PJ, Ruland J. Aberrant NF-κB signaling in lymphoma: mechanisms, consequences, and therapeutic implications. Blood. 2007; 109(7):2700-2707. 42. Packham G. The role of NF-kappaB in lymphoid malignancies. Br J Haematol. 2008;143(1):3-15. 43. McConkey DJ, Zhu K. Mechanisms of proteasome inhibitor action and resistance in cancer. Drug Resist Updat. 2008;11(4-5):164179. 44. Mujtaba T, Dou QP. Advances in the understanding of mechanisms and therapeutic use of bortezomib. Discov Med. 2011;12(67): 471-480. 45. Bu R, Hussain AR, Al-Obaisi KA, Ahmed M, Uddin S, Al-Kuraya KS. Bortezomib inhibits proteasomal degradation of IκBα and induces mitochondrial dependent apoptosis in activated B-cell diffuse large B-cell lymphoma. Leuk Lymphoma. 2014; 55(2):415424. 46. Delmore JE, Issa GC, Lemieux ME, et al. BET bromodomain inhibition as a therapeutic strategy to target c-Myc. Cell. 2011;146(6): 904-917. 47. Derenzini E, Mondello P, Erazo T, et al. BET inhibition-induced GSK3β feedback enhances lymphoma vulnerability to PI3K inhibitors. Cell Rep. 2018;24(8):2155-2166. 48. Ceribelli M, Kelly PN, Shaffer AL, et al. Blockade of oncogenic IκB kinase activity in diffuse large B-cell lymphoma by bromodomain and extraterminal domain protein inhibitors. Proc Natl Acad Sci U S A. 2014;111(31):11365-11370.

haematologica | 2021; 106(9)

ARTICLE

Non-Hodgkin Lymphoma

Long-term outcomes from the phase II L-MIND study of tafasitamab (MOR208) plus lenalidomide in patients with relapsed or refractory diffuse large B-cell lymphoma Johannes Duell,1 Kami J. Maddocks,2 Eva González-Barca,3 Wojciech Jurczak,4 Anna Marina Liberati,5 Sven de Vos,6 Zsolt Nagy,7 Aleš Obr,8 Gianluca Gaidano,9 Pau Abrisqueta,10 Nagesh Kalakonda,11 Marc André,12 Martin Dreyling,13 Tobias Menne,14 Olivier Tournilhac,15 Marinela Augustin,16 Andreas Rosenwald,17 Maren Dirnberger-Hertweck,18 Johannes Weirather,18 Sumeet Ambarkhane18 and Gilles Salles19°

Ferrata Storti Foundation

Haematologica 2021 Volume 106(9):2417-2426

Medizinische Klinik und Poliklinik II, Universitätsklinik Würzburg, Würzburg, Germany; Department of Internal Medicine, Arthur G James Comprehensive Cancer Center, Ohio State University Wexner Medical Center, Columbus, OH, USA; 3Department of Hematology, Institut Catalá d’Oncologia (ICO), Hospital Duran i Reynals, Universitat de Barcelona, Barcelona, Spain; 4Maria Sklodowska–Curie National Research Institute of Oncology, Kraków, Poland; 5Università degli Studi di Perugia, Azienda Ospedaliera Santa Maria di Terni, Terni, Italy; 6Department of Medicine, Ronald Reagan UCLA Medical Center, Santa Monica, CA, USA; 71st Department of Internal Medicine, Semmelweis University, Budapest, Hungary; 8Department of Hemato-Oncology, Palacký University and University Hospital, Olomouc, Czech Republic; 9Division of Hematology, Department of Translational Medicine, University of Eastern Piedmont, Novara, Italy; 10 Department of Hematology, Vall d’Hebron Institute of Oncology (VHIO), Vall d’Hebron University Hospital, Barcelona, Spain; 11Molecular and Clinical Cancer Medicine, University of Liverpool and The Clatterbridge Cancer Centre, Liverpool, UK; 12 Department of Haematology, Université Catholique de Louvain, CHU UCL Namur, Yvoir, Belgium; 13Department of Medicine III, LMU University Hospital, Munich, Germany; 14Department of Haematology, Freeman Hospital, Newcastle Upon Tyne Hospitals NHS Foundation Trust, Newcastle Upon Tyne, UK; 15Service d’Hématologie Clinique et de Thérapie Cellulaire, CHU Estaing, Clermont-Ferrand, France; 16 Department of Hematology and Oncology, Paracelcus Medical University, Klinikum Nürnberg, Nürnberg, Germany; 17Institute of Pathology, University of Würzburg, Würzburg, Germany; 18MorphoSys AG, Planegg, Germany and 19Hématologie, Hospices Civils de Lyon and Université de Lyon, Lyon, France. 1 2

°Current address: Lymphoma Service, Memorial Sloan Kettering Cancer Center, New York, NY, USA.

ABSTRACT

afasitamab (MOR208), an Fc-modified, humanized, anti-CD19 monoclonal antibody, combined with the immunomodulatory drug lenalidomide was clinically active with a good tolerability profile in the open-label, single-arm, phase II L-MIND study of patients with relapsed/refractory diffuse large B-cell lymphoma (DLBCL) ineligible for autologous stem-cell transplantation. To assess long-term outcomes, we report an updated analysis with ≥35 months’ follow-up. Patients were aged >18 years, had received one to three prior systemic therapies (including ≥1 CD20-targeting regimen) and Eastern Cooperative Oncology Group performance status 0-2. Patients received 28-day cycles of tafasitamab (12 mg/kg intravenously), once weekly during cycles 1-3, then every 2 weeks during cycles 4-12. Lenalidomide (25 mg orally) was administered on days 1-21 of cycles 1-12. After cycle 12, progression-free patients received tafasitamab every 2 weeks until disease progression. The primary endpoint was best objective response rate. After ≥35 months’ follow-up (data cut-off: October 30, 2020), the objective response rate was 57.5% (n=46/80), including a complete response in 40.0% of patients (n=32/80) and a partial response in 17.5% of patients (n=14/80). The median duration of response was 43.9 months (95% confidence interval [95% CI]: 26.1-not reached), the median overall survival was 33.5 months (95% CI: 18.3-not reached) and the median progression-free survival was 11.6 months (95% CI: 6.3-45.7). There haematologica | 2021; 106(9)

Correspondence: GILLES SALLES sallesg@mskcc.org Received: November 13, 2020. Accepted: May 11, 2021. Pre-published: July 1, 2021. https://doi.org/10.3324/haematol.2020.275958

2417

J. Duell et al.

were no unexpected toxicities. Subgroup analyses revealed consistent long-term efficacy results across most subgroups of patients. This extended follow-up of L-MIND confirms the long duration of response, meaningful overall survival, and well-defined safety profile of tafasitamab plus lenalidomide followed by tafasitamab monotherapy in patients with relapsed/refractory diffuse large B-cell lymphoma ineligible for autologous stem cell transplantation. ClinicalTrials.gov identifier: NCT02399085.

Introduction Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of non-Hodgkin lymphoma, accounting for 25-45% of new cases of lymphoma each year.1 The introduction of rituximab treatment, an anti-CD20 antibody, alongside cyclophosphamide, doxorubicin, prednisone, and vincristine (R-CHOP) as an initial standard-of-care immunotherapy has improved patients’ outcomes; however, 30–40% of patients continue to experience relapse or are refractory to this first-line therapy.2 For these relapsed or refractory (R/R) patients, alternative effective and tolerable treatment options are limited and, thus, their prognosis is poor.2 Current treatment options for R/R DLBCL include salvage chemotherapy followed by high-dose chemotherapy and autologous stem-cell transplantation (ASCT).3,4 However, the majority of patients with R/R DLBCL who undergo ASCT subsequently relapse.2 More recentlydeveloped therapies, such as chimeric antigen receptor (CAR) T-cell therapy and the antibody-drug conjugate polatuzumab vedotin in combination with bendamustine and rituximab, have shown improved patients’ outcomes.5-7 However, CAR T-cell therapies have been associated with severe adverse events, including grade ≥3 cytokine release syndrome and neurotoxicity, and some can be difficult to administer safely and successfully.5,6 Thus, there remains an urgent need for novel, tolerable, and easy-to-administer treatment options for patients with R/R DLBCL, particularly those ineligible for ASCT. The combination of tafasitamab (MOR208, previously XmAb5574), an Fc-modified, humanized anti-CD19 monoclonal antibody, with lenalidomide has been shown to be effective and well-tolerated in patients with R/R DLBCL who are ineligible for ASCT.8 The phase II study, L-MIND, demonstrated an objective response rate of 60%, with 43% of patients achieving a complete response (CR).8 Moreover, the responses were durable, with a median duration of response (DoR) of 21.7 months.8 To further determine the long-term clinical efficacy and safety of tafasitamab plus lenalidomide treatment in patients with R/R DLBCL, we provide updated data based on a minimum follow-up of 35 months. Additionally, to understand the effectiveness of this novel treatment regimen in clinically relevant subgroups of patients, we present long-term efficacy analyses stratified according to important baseline covariates of prognostic significance.

Clinical Practice guidelines and the Declaration of Helsinki; all patients provided written informed consent. We present data after 35 months of follow-up from the last patient enrolled.

Study design and patients Details of the L-MIND study have been published elsewhere; eligibility criteria are further described in the Online Supplementary Methods.8 Patients with primary refractory disease were excluded, although until a protocol amendment in June 2016, primary refractoriness was defined as no response or progressive disease (PD) within <3 months of frontline therapy, rather than 6 months. Therefore, prior to this amendment patients with relapse or PD 3-6 months from frontline therapy were included, and form a subgroup of ‘primary refractory patients’ as per B-cell lymphoma National Comprehensive Cancer Network guidelines.3 Patients with rituximab-refractory disease had no response to or PD following a rituximab-containing regimen within <6 months of completion of therapy. Patients received up to 12 cycles (28 days each) of tafasitamab and lenalidomide, followed by tafasitamab monotherapy in patients with stable disease or better, until PD. Tafasitamab (12 mg/kg intravenously) was administered on days 1, 8, 15, and 22 during cycles 1-3, with a loading dose on day 4 of cycle 1, and on days 1 and 15 from cycle 4 onwards. Lenalidomide (25 mg orally) was self-administered on days 1-21 of each 28-day cycle. For further details see the Online Supplementary Methods.

Study outcomes The primary endpoint was the objective response rate (CR plus partial response [PR]), assessed by an independent review committee (IRC), according to the 2007 International Working Group response criteria for malignant lymphoma.9 Secondary endpoints included DoR (time from initial CR or PR to first observation of PD), progression-free survival (PFS; time from first dosing to lymphoma progression or death), overall survival (OS; time from first dosing to date of death), and incidence and severity of adverse events. Exploratory subgroup analyses were performed to evaluate DoR, PFS, and OS by refractoriness to prior treatment, as well as age, gender, International Prognostic Index (IPI) score, prior ASCT, and number of prior treatment lines. Rituximab refractoriness was defined as a response less than PR to any rituximab-containing regimen during the course of treatment or PD within ≤6 months of treatment completion. Refractoriness to last prior treatment and primary refractoriness were defined as a best response less than PR to the most recent therapy or to first-line treatment, respectively, or PD before or ≤6 months after completion of that treatment.

Statistical analyses Methods Study conduct L-MIND was an open-label, single-arm, multicenter, phase II study (NCT02399085).8 The study was approved by the institutional review boards at each study site, and conducted in accordance with International Council for Harmonization Good

2418

The previously published primary analysis for the L-MIND study (data cut-off: November 30, 2018)8 was carried out when all patients had completed a minimum of 12 months’ follow-up. The data cut-off date for the present analyses was October 30, 2020. The full analysis set comprised patients who received both tafasitamab and lenalidomide and was used to analyze efficacy outcomes. The safety analysis set comprised patients who received any study medication.

haematologica | 2021; 106(9)

Long-term outcomes with tafasitamab in R/R DLBCL

Results Patients Overall, 81 patients received at least one dose of either drug and were evaluated for safety. Of those, 80 patients received ≥1 dose of both tafasitamab and lenalidomide and were evaluated for efficacy (Figure 1). A total of 34 patients received tafasitamab monotherapy after discontinuing lenalidomide (30/34 patients had completed 12 cycles of tafasitamab plus lenalidomide and 4/34 had discontinued lenalidomide prior to cycle 12 and continued tafasitamab). Fifteen of these 34 patients had discontinued tafasitamab treatment at the data cut-off for this analysis; thus, 19 patients were still receiving tafasitamab monotherapy. Of the 62/81 patients who had discontinued study treatment, 42 had died, 13 were alive and included in the survival follow-up and 7 had been lost to follow-up at the data cut-off for this report. The full baseline characteristics of the patients in the LMIND study have already been published.8 Briefly, the patients had a median age of 72 years (range, 41-86) at enrollment and had received a median of two (range, 1-4) prior lines of therapy. All patients had received R-CHOP or equivalent chemoimmunotherapy prior to study entry. With the availability of additional data from a central pathology review of two patients, the baseline patients’ characteristics for cell of origin by immunohistochemistry and gene expression profiling have been updated since the primary analysis (Table 1). There was one patient each with double- and triple-hit DLBCL. Patient subgroups of clinical interest included 15 patients (18.5%) with primary refractory disease, 33 patients (41.3%) with rituximab-refractory disease, and 35 patients (43.8%) who were refractory to their last ther-

apy. Most patients who were refractory to their last line of therapy had received two prior lines of treatment (71.4%), and the last prior line included chemotherapy in 94.4% and rituximab in 80.0% of cases. The baseline characteristics of patients in the refractory subgroups were generally comparable with those of the overall population (Table 1), although patients in refractory subgroups were more likely to have increased lactate dehydrogenase and germinal center B cell of origin by immunohistochemistry. Prior treatment regimens for patients refractory to their last treatment are shown in Online Supplementary Table S1.

Efficacy outcomes After the primary analysis, the best responses for three patients were revised based on an IRC re-adjudication due to a disagreement between the two primary radiologists. At this long-term data cut-off after at least 35 months’ follow-up, the IRC-assessed objective response rate was 57.5% (46/80; 95% confidence interval [95% CI]: 45.9-68.5), the CR rate was 40.0% (32/80) and the PR rate was 17.5% (14/80) (Table 2). Additionally, 16.3% of patients (13/80) had stable disease. The median time to response was 2.1 months (range, 1.7-34.7) and the median time to CR was 6.8 months (range, 1.7-46.3). Thirty patients had completed the combination treatment phase of 12 cycles on both study drugs and achieved a best response of CR (n=24), PR (n=3), or stable disease (n=3) as per IRC. Time-to-event endpoints are shown in Table 2 with Kaplan-Meier plots in Figure 2. The median IRC-assessed DoR was 43.9 months (95% CI: 26.1-not reached [NR]), and was not reached in patients who achieved a CR (95% CI: 43.9-NR). The median IRC-assessed PFS was 11.6

Figure 1. CONSORT (Consolidated Standards of Reporting Trials) diagram of the L-MIND study at the October 30, 2020 data cut-off.

haematologica | 2021; 106(9)

2419

J. Duell et al.

months (95% CI: 6.3-45.7) with a median follow-up for PFS of 33.9 months. A total of 38 patients were censored at data cut-off; 21/38 patients (55.3%) were ongoing on PFS follow-up. The Kaplan-Meier plot of PFS suggests a plateau at around 18 months (Figure 2B). The median OS had not been reached at the primary analysis and was 33.5 months (95% CI: 18.3-NR) in this analysis, with a median survival follow-up of 42.7 months. Figure 2C shows the impact of response quality on OS; among the patients with a CR, the median OS was not reached, and OS estimates were 96.9% (95% CI: 79.8-99.6) at 18 months, 90.6% (95% CI: 73.7-96.9) at 24 months, and 81.3% (95% CI: 62.9-91.1) at 36 months. Among patients with a PR, the median OS was 22.5 months (95% CI: 8.6NR), and OS estimates were 59.8% (95% CI: 28.5-81.0) at 18 months, 42.7% (95% CI: 15.9-67.5) at 24 months and 34.2% (95% CI: 10.7-59.8) at 36 months. In patients who received tafasitamab plus lenalidomide as second-line treatment (n=40), the median PFS was 23.5 months (95% CI: 7.4-NR), the median DoR was 43.9 months (95% CI: 9.1-NR) and the median OS was 45.7 (95% CI: 24.6-NR). In patients receiving tafasitamab plus lenalidomide as third- or later-line treatment (n=40), the median PFS was 7.6 months (95% CI: 2.7-NR), the median DoR was not reached (95% CI: 15.0-NR) and the median OS was 15.5 months (95% CI: 8.6-NR). Following the discontinuation of treatment in L-MIND, 33 patients received subsequent salvage therapies, which included stem cell transplant in two patients and CAR Tcell therapy in two other patients, following further chemotherapy (see Online Supplementary Results). Additionally, five patients who achieved a CR in L-MIND but discontinued the treatment for reasons other than disease progression were alive at the data cut-off date for this analysis, without further therapeutic intervention.

Subgroup analyses Overall response and CR rates were consistent regardless of refractoriness in patient subgroups of clinical interest although, as expected, the median PFS and OS were short in patients with primary refractory disease (5.3 months and 13.8 months, respectively), rituximab-refractory and last-line refractory disease (both 7.6 months and 15.5 months, respectively) (Table 2). Forest plots for Kaplan-Meier estimates of 30-month time-to-event endpoints are shown in Figure 3. Across DoR, PFS and OS, the only patient subgroup that consistently had a significantly poorer prognosis than the overall group was that of patients with an intermediate-high and high-risk IPI score. Patients in the rituximab-refractory (n=33 evaluable) and last-line-refractory (n=35 evaluable) subgroups had similar 30-month DoR and PFS rates to the rest of the population (DoR: 66.2% vs. 65.5% and 57.7% vs. 69.5%; PFS: 40.0% vs. 42.6% and 37.2% vs. 44.2%, respectively), whereas 30-month DoR and PFS rates were lower in patients with primary refractory disease (n=15; DoR: 50.0% vs. 66.7%; PFS: 33.9% vs. 42.3%) (Figure 3A, B). In all refractory subgroups, the 30-month OS rate was lower compared with that of the rest of the population (Figure 3C). Kaplan-Meier plots for PFS in the refractory subgroups are shown in Online Supplementary Figure S1. Based on medical history and central pathology diagnosis, eight patients had DLBCL arising from transformation of low-grade lymphoma, and there was one patient each with double- and triple-hit lymphoma. Of the eight 2420

Table 1. Updated baseline characteristics and patient subgroups of clinical interest.

All patients Number 81 Median age, years (range) 72 (41-86) Age >70 years, n (%) 45 (56) Median prior lines 2 (1-4) of treatment (range) Stage III/IV, n (%) 61 (75) Increased LDH, n (%) 45 (56) IPI 3-5, n (%) 41 (51) Prior ASCT, n (%) 9 (11) Cell of origin (by IHC), n (%) GCB 39 (48) Non-GCB 22 (27) Unknown 20 (25) Cell of origin (by GEP), n (%) GCB 8 (10) ABC 20 (25) Unclassified 6 (7) Not evaluable 5 (6) Missing 42 (52) Patients with transformed lymphoma,* n (%) B-cell lymphoma 4 (5) Marginal zone lymphoma 2 (3) NHL unspecified histology 1 (1) Case reported by central 1 (1) pathology review

Primary refractory disease

Rituximab Last therapy refractory refractory disease disease

15 34 36 73 (48-82) 72.5 (41-82) 72.5 (41-82) 9 (60) 19 (56) 20 (56) 2 (1-4) 2 (1-4) 2 (1-4) 10 (67) 10 (67) 8 (53) 0

24 (71) 22 (65) 19 (56) 3 (9)

27 (75) 25 (69) 21 (58) 4 (11)

12 (80) 1 (7) 2 (13)

21 (62) 6 (18) 7 (21)

21 (58) 8 (22) 7 (19)

2 (13) 5 (33) 1 (7) 2 (13) 5 (33)

5 (15) 8 (24) 1 (3) 3 (9) 17 (50)

5 (14) 8 (22) 4 (11) 3 (8) 16 (44)

1 (7) 1 (7) 0 0

2 (6) 0 0 0

2 (3) 1 (3) 0 1 (3)

*Defined from records in the medical history for seven patients with transformed lymphoma and as a current medical condition (ongoing at cycle 1, day 1) for one B-cell lymphoma patient. Refractory subgroups may overlap. Primary refractory disease was defined as progression during first-line treatment and/or progressive disease or stable disease as response to first-line treatment or progressive disease within 6 months after completion of first-line treatment. Rituximabrefractory disease was defined as progressive disease or stable disease in response to any rituximab-containing regimen or progressive disease during or within 6 months of completion of any rituximab-containing therapy line. Last therapy-refractory disease was defined as progressive disease or stable disease in response to the most recently administered therapy before study entry. LDH: lactate dehydrogenase; IPI: International Prognostic Index; ASCT: autologous stem-cell transplant; IHC: immunohistochemistry; GCB: germinal center B cell; GEP:gene expression profiling; ABC: activated B-cell; NHL: non-Hodgkin lymphoma.

patients with transformed lymphoma, four experienced PR, three experienced CR and one had stable disease as best response. The patient with double-hit lymphoma (MYC and BCL2 translocations) was refractory to his last line of therapy before L-MIND (rituximab-dexamethasone-cytarabine-cisplatin) and achieved a PR to tafasitamab and lenalidomide, and was progression-free for >6 months. The patient with triple-hit lymphoma (MYC, BCL2 and BCL6 translocations) had previously experienced a CR for 4.5 months in response to R-CHOP and started tafasitamab plus lenalidomide 1 month after relapse. This patient experienced a CR in L-MIND with sustained remission for >30 months. Swimmer plots for all of these patients are shown in Online Supplementary Figure S2.

Safety outcomes As of October 30, 2020, the median duration of exposure to study treatment (either lenalidomide or tafasitamab) was 9.2 months (range, 0.2-54.7). The median duration of exposure to tafasitamab monotherapy (following haematologica | 2021; 106(9)

Long-term outcomes with tafasitamab in R/R DLBCL

Table 2. Efficacy outcomes in the primary and follow-up analyses.

Tafasitamab plus lenalidomide (N=80)‡ Primary analysis Follow-up analysis (data cut-off: (data cut-off: Nov 30, 2018)8 Oct 30, 2020) Best objective response, n (%) Complete response Partial response Stable disease Progressive disease Not evaluable* ORR (CR + PR), n (%) [95% CI]† Median DoR (IRC), months (95% CI) Median PFS (IRC), months (95% CI) Median OS, months (95% CI)

34 (42.5) 14 (17.5) 11 (13.8) 13 (16.3) 8 (10.0) 48 (60.0) [48.4-70.9] 21.7 (21.7-NR) 12.1 (5.7-NR) NR (18.3-NR)

32 (40.0) 14 (17.5) 13 (16.3) 13 (16.3) 8 (10.0) 46 (57.5) [45.9-68.5] 43.9 (26.1-NR) 11.6 (6.3-45.7) 33.5 (18.3-NR)

Clinically relevant subgroups (follow-up analysis) Primary refractory Rituximab-refractory Last-therapydisease disease refractory (n=15) (n=33) (n=35) 5 (33.3) 3 (20.0) 2 (13.3) 3 (20.0) 2 (13.3) 8 (53.3) [26.6-78.7] NR (1.8-NR) 5.3 (0.9-NR) 13.8 (1.3-NR)

13 (39.4) 5 (15.2) 4 (12.1) 7 (21.2) 4 (12.1) 18 (54.5) [36.4-71.9] NR (5.8-NR) 7.6 (2.7-NR) 15.5 (8.6-NR)

14 (40.0) 7 (20.0) 3 (8.6) 7 (20.0) 4 (11.4) 21 (60.0) [42.1-76.1] NR (5.8-NR) 7.6 (2.7-NR) 15.5 (8.6-NR)

*Non-evaluable patients had no valid post-baseline response assessments. †Using the two-sided 95% Clopper-Pearson exact method based on a binomial distribution. ‡One patient received tafasitamab only. ORR: objective response rate; CR: complete response; PR: partial response; 95% CI: 95% confidence interval; DoR: duration of response; IRC: independent review committee; PFS: progression-free survival; OS: overall survival; NR: not reached.

discontinuation of lenalidomide at any time [n=52]) was 13.9 months (range, 0.2-43.4), compared with a median of 4.1 months’ exposure to tafasitamab monotherapy in the primary analysis (range, 0.1-20.8 months; data cut-off November 30, 2018).8 However, with the exception of one patient with recurrence of a previously diagnosed marginal zone lymphoma that was documented as an adverse event (Figure 1), no patients discontinued the study due to adverse events during the tafasitamab extended monotherapy phase. Overall, 64 (79.0%) patients required a temporary interruption of tafasitamab, of which 73.4% cases were due to adverse events. During combination therapy, 43 (53.1%) patients required no dose reduction of lenalidomide from the starting dose of 25 mg. Lenalidomide interruptions were required by 28 (34.6%) patients, being due to adverse events in 89.3% of cases, and 37 patients (45.7%) required a lenalidomide dose reduction. The most frequent treatment-emergent adverse event (TEAE) leading to treatment interruption for tafasitamab (± lenalidomide) and lenalidomide (± tafasitamab) was neutropenia (28 [34.6%] patients and 24 [29.6%] patients, respectively). During the extended tafasitamab monotherapy phase, 21 (52.5%) patients had an interruption of tafasitamab treatment due to at least one TEAE, the most common reasons being neutropenia or leukopenia (9 patients) and respiratory tract infections (6 patients). At the current analysis, 42 patients (51.9%) had died. There were eight deaths (9.9%) on treatment (5 related to PD, plus 1 stroke, 1 sudden death and 1 respiratory failure), and 34 deaths (42.0%) after treatment (26 related to PD, plus 1 intracerebral hemorrhage, 1 pulmonary edema due to heart failure, 1 pneumonia, 1 end-stage marrow failure, 1 progressive multifocal leukoencephalopathy, 1 congestive heart failure and 1 acute myeloid leukemia considered by the investigator to be secondary to past chemotherapy, and 1 unknown cause). At a median follow-up for OS of 42.7 months, compared with 19.6 months at the primary analysis (an additional follow-up duration of 23.1 months), TEAE were consistent in incidence and severity with the those of the primary analysis (Table 3), with the most common TEAE (all grades) at extended follow-up remaining neutropenia (51%) and anemia (37%). The adverse event burden, haematologica | 2021; 106(9)

expressed in terms of number of adverse events per patient-year of exposure to study medication, decreased greatly during the tafasitamab monotherapy phase compared with that during the combination therapy phase (Table 4). Consistent with the safety profile of tafasitamab monotherapy in other studies,10,11 the most common adverse events during the monotherapy phase were neutropenia, cough, diarrhea, anemia, nasopharyngitis, and pyrexia, and the majority of adverse events were of grade 1 or 2. Similar to the primary analysis, the most common grade ≥3 TEAE were neutropenia (49%), thrombocytopenia (17%) and febrile neutropenia (12%). Treatment-emergent serious adverse events (SAE) were reported in 43 patients (53.1%). The most common SAE were pneumonia (7 patients [8.6%]), febrile neutropenia (5 patients [6.2%]), pulmonary embolism (3 patients [3.7%]), bronchitis, lower respiratory tract infection, atrial fibrillation and congestive cardiac failure (all 2 patients [2.5%]). Of these, pneumonia and lower respiratory tract infection had been reported in an additional two and one patients, respectively, compared with the primary analysis, while the rest remained unchanged. Overall, ten patients (12.3%) experienced febrile neutropenia (grade 3 or 4). Five of these patients also developed infections whose timing was associated with febrile neutropenia (urinary tract infection [grade 3 adverse event]; sepsis and urinary tract infection [both grade 4 SAE]; Enterobacter bacteremia [grade 3 SAE]; staphylococcal skin infection [grade 2 adverse event]; rhinitis [grade 1 adverse event] and respiratory syncytial virus infection [grade 3 SAE]), and all recovered within 3-24 days; the other five patients developed no infections at all or their timing was not associated with febrile neutropenia. Between the primary analysis and this update, there were few new adverse events reported related to infection and rash (Online Supplementary Table S2). This observation is consistent with the low incidence of these events associated with tafasitamab monotherapy. Eleven patients (13.6%) experienced 13 TEAE of special interest, including tumor flare (3 events in 3 patients [3.7%]), allergic dermatitis (3 events in 3 patients [3.7%]), basal cell carcinoma (4 events in 2 patients [2.5%]), myelodysplastic conditions (2 events in 2 patients [2.5%]), and Bowen disease (1 event in 1 patient [1.2%]). 2421

J. Duell et al.

A Figure 2. Proportion of patients in remission. (A-C) Kaplan-Meier plots of duration of response (A), progression-free survival. (B) and overall survival (C) after 35 months of follow-up. 95% CI. 95% confidence interval; CR: complete response; DoR: duration of response; NE: not evaluable; NR: not reached; OS: overall survival; PD: progressive disease; PFS: progression-free survival; PR: partial response; SD: stable disease.

2422

haematologica | 2021; 106(9)

Long-term outcomes with tafasitamab in R/R DLBCL

Figure 3. Kaplan-Meier estimates of 30-month time-toevent endpoints. (A) Duration of response,* (B) progression-free survival and (C) overall survival rates. *Based on patients who achieved an objective response (CR or PR) in the respective subgroups. 95% CI: 95% confidence interval; DoR: duration of response; IPI: International Prognostic Index; nC: number of patients censored; nE: number of patients with event; nR: number of patients at risk; n#: number of responders within each subgroup (A: DoR), or number of overall patients within each subcategory (B: PFS; C: OS); OS: overall survival; PFS: progression-free survival. The vertical line indicates the 30month DoR (A), PFS (B) and OS (C) rates across all responders/patients.

haematologica | 2021; 106(9)

2423

J. Duell et al.

Table 3. Treatment-emergent adverse events occurring in ≥10% of patients, or grade 3-5 treatment-emergent adverse events in >1 patient, reported at the updated L-MIND analysis.

All grades (≥10%) n (%) Hematologic events Neutropenia Anemia Thrombocytopenia Leukopenia Febrile neutropenia Lymphopenia Non-hematologic events Diarrhea Asthenia Cough Peripheral edema Pyrexia Decreased appetite Back pain Hypokalemia Fatigue Constipation Muscle spasms Nausea Bronchitis Vomiting All infective pneumonia* All urinary tract infection* Dyspnea C-reactive protein increased Respiratory tract infection Upper respiratory tract infection Hypertension

Grade ≥3 (>1 patient) n (%)

41 (50.6) 30 (37.0) 25 (30.9) 12 (14.8) 10 (12.3) 6 (7.4)

40 (49.4) 6 (7.4) 14 (17.3) 9 (11.1) 10 (12.3) 3 (3.7)

29 (35.8) 20 (24.7) 22 (27.2) 19 (23.5) 19 (23.5) 18 (22.2) 16 (19.8) 15 (18.5) 14 (17.3) 14 (17.3) 12 (14.8) 12 (14.8) 13 (16.0) 12 (14.8) 10 (12.3) 10 (12.3) 10 (12.3) 9 (11.1) 9 (11.1) 8 (9.9) 7 (8.6)

1 (1.2) 2 (2.5) 1 (1.2) 0 1 (1.2) 0 2 (2.5) 5 (6.2) 2 (2.5) 0 0 0 1 (1.2) 0 8 (9.9) 2 (2.5) 1 (1.2) 0 0 2 (2.5) 3 (3.7)

*Defined by customized Medical Dictionary for Regulatory Activities (MedDRA) query.

There were no cases of grade ≥3 infusion-related reactions, tumor lysis syndrome (of any grade), or cytokine release syndrome (of any grade) during the study. In total, 14 patients (17.3%) received blood transfusions during the study, and 37 patients (45.7%) received granulocyte colony-stimulating factor. The median duration of common adverse events (all grades) was longest for opportunistic infections12 (20 days; 1 event each of progressive multifocal leukoencephalopathy, grade 5; hepatitis B reactivation, grade 2; Clostridium difficile colitis, grade 2; and skin candida, grade 1; 7 events of herpes viral infection, grade 1-4, including 1 event of grade 4 disseminated varicella zoster virus infection in blood, gut, lungs and liver), followed by pneumonia and fatigue or asthenia (18 and 15 days, respectively) and shortest for nausea and vomiting (2 days).

Discussion The primary analysis of the L-MIND study, at a median follow-up of 13.2 months, showed that combination therapy with tafasitamab and lenalidomide resulted in a promising response, including durable CR in a significant proportion of patients, and was well tolerated in transplant-ineligible patients with R/R DLBCL.8 With followup of at least 35 months, these long-term data confirm and extend the results of the primary analysis and provide more information on the consolidation tafasitamab 2424

Table 4. Summary of hematologic and non-hematologic treatmentemergent adverse events (any grade) by patient-years of exposure to tafasitamab.

N=81 Tafasitamab plus lenalidomide§

Extended tafasitamab monotherapy¶

13.95

25.77

6.64

1.73 0.58 0.55 0.44 0.13 0.06

3.79 1.16 1.39 0.91 0.30 0.16

0.48 0.22 0.06 0.14 0.04 0

0.51 0.31 0.30 0.29 0.24 0.23 0.19 0.18 0.18 0.17 0.17 0.16 0.15 0.15

0.89 0.48 0.52 0.64 0.39 0.52 0.39 0.43 0.27 0.36 0.27 0.32 0.14 0.09

0.28 0.18 0.17 0.08 0.17 0.04 0.08 0.03 0.10 0.06 0.11 0.06 0.15 0.19

Overall‡ Any TEAE, events/PYE Hematologic, events/PYE* Neutropenia Anemia Thrombocytopenia Leukopenia Lymphopenia Febrile neutropenia Non-hematologic, events/PYE† Diarrhea Pyrexia Asthenia Peripheral edema Cough Hypokalemia Fatigue Nausea Hypomagnesemia Constipation Bronchitis Decreased appetite Respiratory tract infection Hyperglycemia

Treatment-emergent adverse events (TEAE) were defined as any adverse event reported in the following time interval (including the lower and upper limits): date of first administration of study treatment; date of last administration of study treatment + 30 days, or if they were considered to be related to the study drug. The Medical Dictionary for Regulatory Activities (MedDRA) version 21.0 coding dictionary was used. *Threshold for hematologic TEAE: ≥0.05 events per patient-years of exposure (PYE). †Threshold for non-hematologic TEAE: ≥0.15 events per PYE. ‡PYE was defined as the sum of duration of exposure for all patients, where duration of exposure was calculated as [(date of last dose of tafasitamab) – (date of first dose of tafasitamab) + 1]/365.25. §PYE was defined as the sum of duration of exposure for all patients, where duration of exposure was calculated as [(earliest date either study drug was discontinued) – (earliest date of administration of both study drugs) + 1]/365.25. Adverse event counts were for the combination treatment (tafasitamab + lenalidomide) period only. ¶PYE was defined as the sum of duration of exposure for all patients, where duration of exposure was calculated as [(discontinuation date of tafasitamab) – (earliest date of tafasitamab infusion after lenalidomide discontinuation) + 1]/365.25. Adverse event counts were for the tafasitamab monotherapy period only.

monotherapy phase of the study, with an objective response rate of 57.5%. This regimen was granted accelerated approval by the US Food and Drug Administration for patients with R/R DLBCL not eligible for ASCT, based on a high response rate to therapy and prolonged DoR.13 This long-term follow-up analysis shows clinically significant durable responses for combination therapy followed by tafasitamab monotherapy. The median DoR was nearly 44 months with a median OS of 33.5 months; neither the median DoR nor the median OS was reached in patients with a CR, with 80.1% and 81.3% of patients with a CR in response or alive at 36 months, respectively (Figure 2A, C). The median PFS was notable in patients who received tafasitamab plus lenalidomide as secondline therapy compared with those who received the combination third-line or later (23.5 months vs. 7.6 months [n=40, both groups]). The corresponding median OS were 45.7 months vs. 15.5 months, suggesting that patients derive more benefit from this regimen when it is given in an earlier treatment setting. Good response rates were also achieved with combinahaematologica | 2021; 106(9)

Long-term outcomes with tafasitamab in R/R DLBCL

tion therapy in the subgroups of patients with primary refractory, rituximab-refractory and last-therapy-refractory disease, especially given that these patients are considered difficult to treat, and those responses were durable. The median PFS and OS were, however, shorter than those for the overall population, especially in primary refractory patients, so there is still room for improvement in outcomes for difficult-to-treat patients. Notably, two patients with double- and triple-hit lymphoma and seven out of eight patients with transformed lymphoma responded to therapy. In the exploratory subgroup analysis, the only disease characteristic that appeared to have a negative effect on prognosis was IPI score ≥3 (i.e., intermediate-high- or high-risk disease); a high IPI score has long been recognized as a risk factor for poor outcomes in DLBCL.14 In regard to safety, there was little change in the adverse event profile since the primary analysis, which indicates a good tolerability profile for tafasitamab monotherapy. There was a reduction in the burden of common hematologic and non-hematologic adverse events as patients transitioned from combination therapy to tafasitamab monotherapy, with a residual tolerability profile similar to that in previous studies of tafasitamab monotherapy.10,11 This observation is of considerable importance for frail or elderly patients, who may prefer treatment with limited effects on their quality of life. In particular, the low incidence of infusion-related reactions (all of which were grade 1) and absence of cytokine release syndrome with tafasitamab plus lenalidomide is an important consideration for therapy in frail patients, given the occurrence of these events with CAR T-cell and other antibody therapies. Subsequent treatment, including ASCT and CAR T cells, was not precluded by previous administration of tafasitamab and lenalidomide in patients who experienced disease progression during this combination regimen. In this trial, lenalidomide was given for a limited time of up to 12 months, which is in line with the median DoR of lenalidomide monotherapy in R/R non-Hodgkin lymphoma of 10.5 months,15,16 and the observation that the best responses with tafasitamab plus lenalidomide typically occur within this time window. Treatment until progression with tafasitamab is a novel concept, and although the exact contribution of the monotherapy phase cannot be delineated in this trial, it deserves further investigation. The excellent durability of CR achieved raises the question of whether cure is possible with tafasitamab plus lenalidomide; longer follow-up data will be needed to assess this. Patients with R/R DLBCL who are not eligible for ASCT have few options. In patients who had previously received rituximab, cytotoxic chemotherapy with six to eight cycles of rituximab plus gemcitabine and oxaliplatin was associated with a CR/unconfirmed CR rate of 42% with a median PFS of 4 months and median OS of 8 months, and an overall high incidence of grade ≥3 neutropenia (73%) and thrombocytopenia (44%), requiring transfusions of blood (33%) and platelets (23%).17 In patients with third- or later-line disease, the median PFS of 7.6 months and median OS of 15.5 months with tafasitamab plus lenalidomide are comparable with those achieved with other options such as polatuzumab plus bendamustine and rituxiamb (approved for R/R DLBCL haematologica | 2021; 106(9)

in the European Union)7 and CAR T-cell therapy.6,18 The median DoR has not been reached, with more than 80% of patients with a best response of CR still in remission after 3.5 years. The L-MIND regimen is readily available to administer in an outpatient setting, with oral lenalidomide self-administered by the patient and weekly tafasitamab infusions (fortnightly after the first 3 months of therapy). In conclusion, combination therapy with tafasitamab plus lenalidomide followed by tafasitamab monotherapy provided clinically significant durable responses in patients with R/R DLBCL who were not eligible for ASCT, including those with refractory disease, with manageable toxicity during combination treatment and a reduced adverse event burden during tafasitamab monotherapy. These long-term data further validate tafasitamab plus lenalidomide followed by extended tafasitamab monotherapy as a valuable option for patients with R/R DLBCL who are not eligible for ASCT. Disclosures EGB reports receiving personal fees for consultancy from Janssen, Gilead, Sandoz, Celltrion, and Celgene, and honoraria from Roche, Janssen, AbbVie, and Takeda. GG reports receiving personal fees for advisory board participation from AbbVie, Janssen, AstraZeneca, and Sunesys, and for participation in a speaker bureau from AbbVie and Janssen. GS reports receiving personal fees for consultancy from Epizyme and Ipsen; advisory board participation or symposia from MorphoSys, AbbVie, BeiGene, Genmab, Velosbio, Celgene/BMS, Incyte, Janssen, Novartis, Gilead/Kite, and Genentech/Roche outside the submitted work; and has a patent issued (WO2012010561A1: characterization of another anti-CD19 monoclonal antibody with antibody-dependent cell-mediated cytotoxicity, developed in collaboration with IDD-biotech); this antibody and the company has no relationship with the anti-CD19 antibody described in the current paper (tafasitamab) and has not been licensed to any third party. KM reports receiving personal fees for advisory board participation from MorphoSys during the conduct of the study, and from Pharmacyclics, BMS, Celgene, Kite, and Seattle Genetics outside the submitted work. MD reports receiving an institutional research grant from AbbVie, Bayer, Celgene, Janssen, and Roche; personal fees for advisory board participation from AstraZeneca, Bayer, BeiGene, Celgene, Genmab, Gilead, Incyte, Janssen, Novartis, and Roche; and speaker fees from Amgen, AstraZeneca, Bayer, Celgene, Gilead, Janssen, and Roche, all outside the submitted work. NK reports receiving research funding from Celgene, outside the submitted work. OT reports receiving personal fees from Roche, Gilead, AbbVie, Celgene, Janssen, Sandoz and iQuone, and travel grants from Roche, Gilead, AbbVie, Celgene and Janssen outside the submitted work. WJ reports receiving research funding from MorphoSys during the conduct of the study, and Roche, Sandoz and Celltrion outside the submitted work. AO reports honoraria from Roche and personal fees for advisory board participation from Janssen. JW, MDH, and SA are employees of MorphoSys AG, Planegg, Germany. All other authors declare no competing interests. Contributions JD, GS, JW, MD-H, and SA analyzed and interpreted the data. All authors contributed to data acquisition, manuscript development, and approval. All authors interpreted the results and agree on accountability for all study aspects, including accuracy, integrity, and protocol adherence. All authors contributed to study design or conduct, data analyses, or manuscript writing. 2425

J. Duell et al.

Acknowledgments The authors would like to thank the patients and their families, clinical researchers, and their teams and hospitals that participated in this study. The authors thank Günter Fingerle-Rowson and

References 1. World Health Organization. World Cancer Report 2020: Cancer Research for Cancer Prevention. Cancer Control. 2020;199. 2. Sarkozy C, Sehn LH. New drugs for the management of relapsed or refractory diffuse large B-cell lymphoma. Ann Lymphoma. 2019;3:10. 3. National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines: B-Cell Lymphomas V3.2021. Published 2021. https://www.nccn.org/professionals/physician_gls/pdf/b-cell.pdf. Accessed February 7, 2021. 4. Tilly H, Gomes da Silva M, Vitolo U, et al. Diffuse large B-cell lymphoma (DLBCL): ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2015;26(Suppl 5):v116-v125. 5. Neelapu SS, Locke FL, Bartlett NL, et al. Axicabtagene ciloleucel CAR T-cell therapy in refractory large B-cell lymphoma. N Engl J Med. 2017;377(26):2531-2544. 6. Schuster SJ, Bishop MR, Tam CS, et al. Tisagenlecleucel in adult relapsed or refractory diffuse large B-cell lymphoma. N Engl J Med. 2019;380(1):45-56. 7. Sehn LH, Herrera AF, Flowers CR, et al. Polatuzumab vedotin in relapsed or refractory diffuse large B-cell lymphoma. J Clin Oncol. 2020;38(2):155-165. 8. Salles G, Duell J, González Barca E, et al. Tafasitamab plus lenalidomide in relapsed

2426

Emanuel Lohrmann of MorphoSys AG for their input to this manuscript. This study was sponsored by MorphoSys AG. Medical writing assistance was provided by Rebecca Hurst, PhD of Syneos Health and funded by MorphoSys AG.

or refractory diffuse large B-cell lymphoma (L-MIND): a multicentre, prospective, single-arm, phase 2 study. Lancet Oncol. 2020;21(7):978-988. 9. Cheson BD, Pfistner B, Juweid ME, et al. Revised response criteria for malignant lymphoma. J Clin Oncol. 2007;25(5):579586. 10. Woyach JA, Awan F, Flinn IW, et al. A phase 1 trial of the Fc-engineered CD19 antibody XmAb5574 (MOR00208) demonstrates safety and preliminary efficacy in relapsed CLL. Blood. 2014;124(24):3553-3560. 11. Jurczak W, Zinzani PL, Hess G, et al. A phase IIa, open-label, multicenter study of single-agent tafasitamab (MOR208), an Fcoptimized anti-CD19 antibody, in patients with relapsed or refractory B-cell nonHodgkin’s lymphoma: long-term followup, final analysis. Blood. 2019;134 (Suppl_1):4078. 12. Panel on Opportunistic Infections in Adults and Adolescents with HIV. Guidelines for the Prevention and Treatment of Opportunistic Infections in Adults and Adolescents with HIV. Published 2020. https://clinicalinfo.hiv.gov/sites/default/file s/guidelines/documents/Adult_OI.pdf. Accessed July 21, 2020. 13. US Food & Drug Administration. FDA grants accelerated approval to tafasitamabcxix for diffuse large B-cell lymphoma. Published online 2020:1-2. h t t p s : / / w w w. f d a . g o v / d r u g s / d r u g -

approvals-and-databases/fda-grants-accelerated-approval-tafasitamab-cxix-diffuselarge-b-cell-lymphoma. Accessed February 7, 2021. 14. The International Non-Hodgkin’s Lymphoma Prognostic Factors Project. A predictive model for aggressive nonHodgkin’s lymphoma. N Engl J Med. 1993;329(14):987-994. 15. Witzig TE, Vose JM, Zinzani PL, et al. An international phase II trial of single-agent lenalidomide for relapsed or refractory aggressive B-cell non-Hodgkin’s lymphoma. Ann Oncol. 2011;22(7):1622-1627. 16. Zinzani PL, Rigacci L, Cox MC, et al. Lenalidomide monotherapy in heavily pretreated patients with non-Hodgkin lymphoma: an Italian observational multicenter retrospective study in daily clinical practice. Leuk Lymphoma. 2015;56(6):1671-1676. 17. Mounier N, El Gnaoui T, Tilly H, et al. Rituximab plus gemcitabine and oxaliplatin in patients with refractory/relapsed diffuse large B-cell lymphoma who are not candidates for high-dose therapy. A phase II Lymphoma Study Association trial. Haematologica. 2013;98(11):1726-1731. 18. Locke FL, Ghobadi A, Jacobson CA, et al. Long-term safety and activity of axicabtagene ciloleucel in refractory large B-cell lymphoma (ZUMA-1): a single-arm, multicentre, phase 1–2 trial. Lancet Oncol. 2019;20(1):31-42.

haematologica | 2021; 106(9)

ARTICLE

Non-Hodgkin Lymphoma

Genetic manipulation of primary human natural killer cells to investigate the functional and oncogenic roles of PRDM1

Ferrata Storti Foundation

Gehong Dong,1,2* Yuping Li,1* Logan Lee,1 Xuxiang Liu,1 Yunfei Shi,1,3 Xiaoqian Liu,1,4 Alyssa Bouska,5 Qiang Gong,1 Lingbo Kong,1 Jinhui Wang,6 Chih-Hong Lou,7 Timothy W. McKeithan,1 Javeed Iqbal5 and Wing C. Chan1 1

Department of Pathology, City of Hope National Medical Center, Duarte, CA, USA; Department of Pathology, Beijing Tiantan Hospital, Capital Medical University, Beijing, China; 3Department of Pathology, Peking University Cancer Hospital & Institute, Key Laboratory of Carcinogenesis and Translational Research (Ministry of Education), Beijing, China; 4Department of Hematology, Affiliated Yantai Yuhuangding Hospital, Qingdao University, Yantai, Shandong, China; 5Pathology and Microbiology, University of Nebraska Medical Center, Omaha, NE, USA; 6Department of Molecular and Cellular Biology, City of Hope, Duarte, CA, USA and 7The Gene Editing and Viral Vector Core, Department of Shared Resources, Beckman Research Institute of City of Hope, Duarte, CA, USA 2

Haematologica 2021 Volume 106(9):2427-2438

*GD and YL contributed equally as co-first authors.

ABSTRACT

xtra-nodal natural killer (NK)/T-cell lymphoma, nasal type (ENKTCL) is a highly aggressive lymphoma, in which the tumor suppressor gene PRDM1 is frequently lost or inactivated. We employed two different CRISPR/Cas9 approaches to generate PRDM1-/primary NK cells to study the role of this gene in NK-cell homeostasis. PRDM1-/- NK cells showed a marked increase in cloning efficiency, higher proliferation rate and less apoptosis compared with their wild-type counterparts. Gene expression profiling demonstrated a marked enrichment in pathways associated with proliferation, cell cycle, MYC, MYB and TCR/NK signaling in PRDM1-/- NK cells, but pathways associated with normal cellular functions including cytotoxic functions were downregulated, suggesting that the loss of PRDM1 shifted NK cells toward proliferation and survival rather than the performance of their normal functions. We were also able to further modify a PRDM1-deleted clone to introduce heterozygous deletions of common tumor suppressor genes in ENKTCL such as TP53, DDX3X, and PTPN6. We established an in vitro model to elucidate the major pathways through which PRDM1 mediates its homeostatic control of NK cells. This approach can be applied to the study of other relevant genetic lesions and oncogenic collaborations in lymphoma pathogenesis.

Introduction Extra-nodal natural killer (NK)/T-cell lymphoma, nasal type (ENKTCL) is a highly aggressive lymphoma that is consistently associated with Epstein-Barr virus (EBV) infection and predominantly affects middle-aged men in Asia and Central and South America.1,2 It typically presents as tumors or destructive lesions in the nasal cavity, maxillary sinuses or palate. Despite a localized presentation in most patients, it tends to relapse locally or at other extra-nodal sites, such as the skin, and the 5-year overall survival of affected individuals is 40-50% with current therapeutic regimens.2,3 About 80-90% of ENKTCL originate from the NK-cell lineage with the rest of cases derived from T cells. Regardless of the cell of origin, the pathology, clinical behavior and treatment are similar. Aggressive NK-cell leukemia, also EBV-associated and derived from NK cells, is regarded as the leukemic form of ENKTCL.4 Our previous genomic analysis of ENKTCL,5,6 including identification of copy number abnormalities, mutation analysis, and DNA methylation studies, suggested that PRDM1, located in 6q21, is a tumor suppressor gene that is frequently inactivat-

Correspondence: WING C. (JOHN) CHAN jochan@coh.org Received: April 8, 2020. Accepted: July 30, 2020. Pre-published: July 30, 2020. https://doi.org/10.3324/haematol.2020.254276

2427

G. Dong et al.

ed by deletion, methylation and mutation. The minimal common region of the 6q21 deletion contains a number of genes (ATG1, AIM1, etc.) but apart from PRDM1, they have not been found to be mutated or methylated either by us or others. We therefore prioritized PRDM1 for further study, and functional analysis of the gene supports its role as a tumor suppressor gene.6-8 PRDM1 has also proven to be a tumor suppressor gene in diffuse large B-cell lymphomas4 and anaplastic large T-cell lymphoma.9 Accumulating evidence supports the concept that it is not only critical for terminal effector cell differentiation in B cells10 but that it is also important in the homeostasis of T cells and in T-cell4 effector differentiation. The level of PRDM1 increased progressively with NK-cell activation with a corresponding drop in MYC level, termination of proliferation and increased cellular apoptosis.6 The precise role of PRDM1 in this process is not clear, and the inability to maintain human NK cells in long-term culture in vitro with interleukin (IL)-2 or IL-15 is a major impediment to further analysis. We are now able to perform long-term in vitro cultures of primary, normal NK cells which allow sufficient time for us to perform specific genetic manipulations and functional studies with genome-edited cells and single-cell clones using the recently developed clustered regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated protein 9 (Cas9) system for targeted gene editing.11 Here, we report the functional consequences of PRDM1 gene knock-out (KO) in primary NK cells from healthy donors and the implication of these findings on NK-cell homeostasis and ENKTCL pathogenesis. While PRDM1 is a frequently mutated tumor suppressor gene in NK-cell lymphomagenesis,6-8 it is not sufficient by itself to generate a lymphoma in a murine model, and additional alterations are necessary. As many of the frequent genetic and epigenetic changes in ENKTCL are lossof-function alterations, the CRISPR/Cas 9 system enables highly efficient targeted gene editing11 to investigate these abnormalities.12-16 Here, we demonstrated other potential tumor suppressor genes that are readily modified and the feasibility of inducing pairs of deletions to study cooperative mutations in NK-cell lymphomagenesis.

Methods Primary NK-cell enrichment Primary NK cells were isolated from peripheral blood mononuclear cells of healthy donors (donor #1 and donor #2) using the EasySep™ Human NK Cell Enrichment Kit (Stemcell, USA; #19055) according to the manufacturer’s protocol. The purity of isolated NK cells was determined by flow cytometry analysis with FITC-labeled anti-human CD56 (Biolegend, USA; n. 362545) and PE-labeled anti-human CD3 (Biolegend, USA, n. 300407) double staining.

PRDM1 knockout mediated by CRISPR/Cas9 with plasmid PX458-sgRNA4 The PX458-sgRNA4 plasmid was delivered into stimulated primary NK cells by electroporation using the Amaxa® Nucleofector® II Device (Lonza, France) according to the manufacturer’s suggested U001 protocol (5 μg plasmid per 2x106 cells) (Figure 1A). Cloning of the modified cells is described in the Online Supplementary Methods. Sequential gene KO was processed similarly but on identified PRDM1-/- NK clone #3. The various guide RNA

2428

used in these experiments are listed in Online Supplementary Tables S1 and S2.

CRISPR/Cas9-mediated disruption of PRDM1 by introduction of a fluorescent protein through homologous recombination Cas9/sgRNA ribonucleoprotein (RNP) complexes targeting PRDM1 exon 5 (Figure 1B), together with a double-stranded DNA repairing template consisting of a fluorescent protein gene (GFP/DsRed) flanked by long homologous arms of the PRDM1 gene, were electroporated into cells, allowing the edited cells, of which both PRDM1 loci were disrupted, to be sorted by fluorescence activated cell sorting (FACS). Details of the experiment are shown in Online Supplementary Figure S1 and Figure 1B. sgRNA2 used in this experiment is shown in Online Supplementary Table S1.

Other experimental methods Cell lines used and cell culture methods, the CRISPR/Cas9 experiments, western blotting, cell proliferation and apoptosis assays, cell cycle analysis, quantitative real-time polymerase (qRTPCR), RNA-sequencing, next-generation sequencing, and statistical analysis, are described in the Online Supplementary Information.

Results Generation of PRDM1-/- primary NK cells by two different CRISPR/Cas9 methods Generation of PRDM1-/- clones #3 and #5 using PX458-sgRNA4 plasmid electroporation With our feeder cell culture system, we successfully cloned primary NK cells after single-cell seeding by FACS. The expanded NK-cell single clones (G-2 and G-3) were observed for approximately 3 weeks (Online Supplementary Figure S2) and were enumerated for evaluation of cloning efficiency. According to our grading system, as specified in the Methods section, the GFP+ cells (plasmid-transfected cells) had significantly higher cloning efficiency (47.7% vs. 11.7%; P<0.05) than parental primary NK cells (Online Supplementary Table S3). Sequencing analysis revealed that 66% (61 of 92) of the clones showed PRDM1 frame-shift deletions around the target site within exon 4 (Online Supplementary Figure S3), with the likelihood of their having been four founders based on the pattern of deletions. Two of these four founder clones with distinct homozygous deletions (Figure 2A, B) were the most prevalent clones isolated by single-cell cloning, indicating that homozygous deletion of PRDM1 confers a growth advantage among these clones. Clones #3 and #5 belonged to one of the homozygous deletions and were chosen as the biological duplicates in our subsequent studies. The loss of PRDM1 protein expression was confirmed by western blot analysis in these clones (Figure 2C, upper panel). We also measured the expression of the PRDM1 target gene MYC in the edited cells and demonstrated that the expression of MYC was upregulated (9-fold) upon PRDM1 KO (Figure 2C, lower panel).

Generation of bulk PRDM1-/- NK cells through fluorescent protein knock-in by homologous DNA repair using Cas9/sgRNA2 ribonucleoprotein electroporation To avoid the selection process inherent in cloning and potential spurious results due to off-target modification by sgRNA4, we modified a recently described technology haematologica | 2021; 106(9)

Functional and oncogenic roles of PRDM1 in NK cells

Figure 1. Legend on following page.

haematologica | 2021; 106(9)

2429

G. Dong et al.

Figure 1. Schematic diagram of knock-out of PRDM1 in NK cells by the CRISPR/Cas9 system. (A) PRDM1 was knocked out by using a Cas9-sgRNA plasmid. The plasmid with both Cas9-GFP and sgRNA is shown. This approach requires cloning of GFP+ NK cells on irradiated feeder cells. Genome editing was evaluated by a high resolution melting-polymerase chain reaction assay and the PRDM1 knock-out was confirmed by western blot and Sanger sequencing. (B) PRDM1 was knocked out by electroporation of the Cas9/sgRNA RNP complex plus a homology directed repair (HDR) template with fluorescence protein gene. Double-stranded HDR DNA templates with inserted GFP or DsRed open reading frame and about 300 bp homologous sequence of exon 5 at each side of the CRISPR/Cas9 cutting site of sgRNA2 were prepared. FACS for GFP and DsRed double positive cells identified PRDM1 knock-out cells, which were confirmed by western blot and Sanger sequencing. Cas9: CRISPR-associated protein 9; GFP: green fluorescent protein; NK: natural killer; PBMC: peripheral blood mononuclear cells; NHEJ: non-homologous end joining; CRISPR: clustered regularly interspaced short palindromic repeat; FACS: fluorescence activated cell sorting; PCR: polymerase chain reaction; KO: knock-out; HDRT: homology directed repair template; RNP: ribonucleoprotein.

developed for primary T-cell CRISPR editing17 for our primary NK-cell genome editing. We chose another guide RNA for exon 5, sgRNA2 (Online Supplementary Table S1, Figure 3A). A homologous DNA repair (HDR) template encompassing an in-frame fusion of either GFP or DsRed open reading frame followed by a strong stop signal from SV40 poly (A) tail was included (Figure 3B, Online Supplementary Figure S1). This approach utilized HDR that introduced GFP or DsRed as markers of a disrupted PRDM1 locus. Cells with two colors indicated bi-allelic insertion of the fluorescent protein and the termination of PRDM1 expression and could be FACS-sorted (Figure 1B). After expansion of the GFP+/DsRed+ double-positive NK cells, we performed genotyping, qRT-PCR and western blot to confirm the bi-allelic PRDM1 KO (Figures 3C and 4A-C, Online Supplementary Table S4). The low PRDM1 expression remaining in qRT-PCR (Figure 4C) and western blotting of PRDM1-/- NK cells (Figure 3C) was from the contaminating feeder cells (Online Supplementary Figure S4). When we used the PRDM1 primers located in exon 2 and exon 3, upstream of the sgRNA2 cutting site, to perform qRT-PCR, a high level of expression was observed. As PRDM1 is an autologous repressor, the upregulation of PRDM1 transcripts upstream of the truncation should not be a surprise and is consistent with earlier observations (Figure 4D). MYC transcription was upregulated in these cells (Figure 3C). These results indicate that we had developed a separate method of generating PRDM1-/- NK cells.

PRDM1-/- NK cells showed growth advantage compared with normal wild type NK cells -/-

PRDM1 NK cells had a growth advantage As shown in Figure 5A, the PRDM1-/- NK clones #3 and #5 had higher growth rates (>2 fold) compared with their normal wild-type (WT) counterparts as measured by an MTS assay. A similar effect was observed with GFP+/DsRed+ PRDM1-/- NK cells (Figure 5B). In fact, the GFP+/DsRed+ PRDM1-/- NK cells grew even better than clones #3 and #5 and expanded 4.2-fold in 6 days, while the WT NK cells expanded less than 2-fold. It must be noted that the GFP+/DsRed+ PRDM1-/- NK cells were relatively younger (~20 days after CRISPR), whereas clone #3 and clone #5 were examined ~90 days after CRISPR. A steady decline in growth rate was observed in PRDM1-/- cells after long-term in vitro culture.

PRDM1-/- NK cells had an increased fraction of cells in the S/G2M phase We performed cell cycle analysis on PRDM1-/- NK cells (clone #3 and clone #5) at around 90 days of culture, and cells undergoing DNA synthesis were measured by EdU incorporation into DNA for 5 consecutive days. Remarkably, 40% of the PRDM1-/- NK cells were in the S/G2M phase versus ~4% in WT NK cells (Figure 5C, D) on the 3rd day after adding fresh feeder cells. This proliferation was feeder-dependent, as the proliferating rates were 2430

reduced after feeder cells had all died (Figure 5D), but the PRDM1-/- clones demonstrated a more sustained proliferation. As observed on the 7th day, >20% of the total PRDM1-/- cells showed EdU incorporation (decreased about 1.5-fold), whereas 1% of the parental cells showed EdU incorporation. Similar results were obtained with GFP+/DsRed+ PRDM1-/- NK cells. The proportion of EdUpositive cells in PRDM1-/- NK cells was 44.2% on the 3rd day and 17.8% on the 7th day (~2.5-fold drop), whereas the proportions of EdU-positive WT NK cells were 23.7% on the 3rd day and 4.75% on the 7th day (~5-fold drop) (Figure 5E). These results indicated that, compared with WT NK cells, PRDM1-/- NK cells had a higher rate of proliferation and a more sustained response to feeders, in agreement with the findings from PRDM1-/- clones #3 and #5. These results indicated that PRDM1 regulates cell proliferation and longevity in the presence of feeder cell stimulation. We also performed cell cycle analysis and observed that a higher proportion of PRDM1-/- NK cells were in the S/G2M phase (22.9~28.5%) compared with WT control cells (18.7%) (Figure 6D). Similar results were obtained with GFP+/DsRed+ PRDM1-/- NK cells, of which 26.4% were in S/G2M phase versus the 17.0% of WT NK cells in the S/G2M phase (Figure 5F). Moreover, in both groups of NK cells, we found there was a slightly higher ratio of the S phase cell fractions (1.5~2.2-fold) in PRDM1-/- NK cells than in WT NK cells. These results unequivocally demonstrated that PRDM1 regulates cell proliferation and growth in NK cells.

PRDM1-/- primary NK cells had fewer apoptotic cells Consistent with the above analysis, there were fewer early apoptotic cells (Q3) in PRDM1-/- clones #3 and #5 than WT NK cells at different days of observation after adding fresh feeder cells (Figure 6A). Statistical analysis showed the general trend of increase of early apoptotic cells in both PRDM1-/- and WT NK cells after feeder cell stimulation (Figure 6B). This observation paralleled the decrease in growth and proliferation of the cells. It is possible that as the cells entered a more quiescent phase, apoptosis also decreased. Similar results of fewer early apoptotic cells in PRDM1-/- NK cells compared with their WT counterpart were obtained with GFP+/DsRed+ PRDM1-/- NK cells (Figure 6C). However, the percentages of early apoptotic cells (Q3) tended to get lower, not higher, with the days in culture with feeder cells. This may be due to the fact that the cells in the second batch were at a relative earlier phase of their lifespan than the single clones #3 and #5, so the cells responded better to the feeder cell stimulation.

Generation of dual-gene knockout NK cells and off-target evaluation Other than single-gene KO, we also introduced deletions in genes often deleted or mutated in NK-cell lymphomas (e.g. TP53, DDX3X and PTPN6). We obtained several different dual-gene KO NK cells using the pSpCas9(BB)-2Ahaematologica | 2021; 106(9)

Functional and oncogenic roles of PRDM1 in NK cells

Figure 2. Sanger sequencing demonstrating the PRDM1 deletion in clones #3 and #5. (A) Guide RNA sgRNA4 sequence and schematic gene structure of PRDM1. Light green boxes and numbers represent the exons of the PRDM1 gene. (B) TOPO cloning showed that both clones #3 and #5 harbored the same bi-allelic deletion, a 79-bp deletion in one allele (comprising 35 bp from exon 4 [411-445] and 44 bp from the adjacent intron 3 and a 11-bp deletion in the other allele (413-424) (Ref: NM_001198). (C) Western blot analysis of PRDM1 protein in clones #3 and #5 (upper panel). Parental wild-type natural killer (NK) cells were used as a positive control. NK lymphoma cell lines, NKYS and KAI3, were used as positive and negative controls, respectively. PRDM1 target gene MYC expression of PRDM1-/- clone #5 versus wild-type NK cells was measured by quantitative reverse transcription polymerase chain reaction (qRT-PCR) (lower panel). The target gene expression levels were normalized to RPL13A, and relative expression was calculated using the 2^-DDCt method. The expression of MYC in WT NK cells was set at 1.0.

GFP plasmid vector based approach18 on PRDM1-/- NK clone #3 (Online Supplementary Figures S5-S7). All the double modified cells we obtained harbored heterozygous deletions of the second targeted gene, TP53, DDX3X or PTPN6. There were no major changes in cellular characteristics or cloning efficiency despite the loss of additional tumor suppressor genes, suggesting that heterozygous deletion of these tumor suppressor genes did not provide additional growth or proliferation advantage to NK cells with PRDM1 deletion. To evaluate whether the observed changes could be largely due to off-target effects from CRISPR/Cas9 gene editing, we performed custom capture and sequencing of known driver mutations in lymphoma using a panel of 334 genes (Online Supplementary Table S5) but did not observe any mutations or copy number abnormalities in PRDM1-edited clones compared with the WT counterpart. Thus, the changes observed in PRDM1-/- cells were not due to alterations of the exomes of any of the genes tested in the panel (334 genes).

Gene expression analysis of PRDM1 knockout cells To elucidate the functional alterations resulting from PRDM1 deletion, we performed RNA-sequencing analysis on PRDM1-deficient NK cells or clones and cell-agematched WT NK cells from the same donors, including two biological replicates. We were able to detect the deletion of exon 4 sequences and insertion of the GFP sequence in the haematologica | 2021; 106(9)

RNA-sequencing analysis (Online Supplementary Figure S8AC), thus disrupting the PRDM1 open reading frame. The truncated mRNA in PRDM1-/- clones were transcribed at higher levels (>2-fold) compared with their WT counterparts, likely due to loss of negative autoregulation by PRDM1. As shown in Online Supplementary Figure S8A, these truncated mRNA did not result in any PRDM1 protein expression. Initial analysis using hierarchical clustering suggested that clusters were partly driven by distinct donor profiles and that PRDM1-/- clones tended to form tight clusters (Figure 7A). Approximately 30% (3,684 of 12,633) of transcripts were differentially expressed (1,419 downregulated, 1,498 upregulated; P<0.05 and false discovery rate <0.3) (Figure 7B). As anticipated, numerous genes and pathways associated with proliferation and cell cycle regulation and progression (e.g., cell cycle control, chromosomal replication, centrosome maturation, RNA splicing, DNA repair, S-phase and G2/M transition targets, E2F targets, MYC, IRF4, E2F3 induction) were upregulated/enriched in the PRDM1-deleted clones (Figure 7C). Genes regulated by the DREAM complex (Dimerization partner, RB-like, E2F4 and multi-vulval class B, master coordinator of cell cycle-dependent gene expression) were enriched upon PRDM1 loss, suggesting that the DREAM repressor complex converts into an activating complex in the absence of PRDM1. Metabolic changes, such as glycogen metabolism, also showed enrichment in PRDM1-/- NK cells. Various genes associated with NK-cell biology (TOX2, 2431

G. Dong et al.

TLR4,19 CCR4,20 VEGF,21 TP63 and IRF4) were upregulated >20-fold in PRDM1-/- cells (Online Supplementary Figure S8D). TOX2 is a critical transcription factor for NK-cell maturation/development upstream of T-bet/TBX21, suggesting that PRDM1 may regulate NK-cell development and function via TOX2.22 Similar to a previous study in mouse NK cells, there was a marked increase in the expression of IRF4 and its target genes upon PRDM1 KO.23 In contrast to the PRDM1 KO cells, the WT cells showed enrichment of genes associated with the FOXO3 pathway, TP53/63, and NK-cell-mediated cytotoxicity (Online Supplementary Figure S8E). Several gene signatures associated with endosomal sorting, lysosomes and secretion, and cell-to-cell communication (e.g., E-cadherin stabilization) were enriched in these cells. Other than these, gene sets associated with quiescence and IL-12, P38 MAPK and TNFR1 signaling were also enriched (Figure 7D). The genes that were downregulated included T-cell and NK-cell signaling and KIR3DL1-3 or KIR3DL members, which generally transduce inhibitory signals upon ligand binding (Online Supplementary Figure S8F). A recent report indicated that several immune checkpoint molecules in CD8+ T cells,24 including LAG-3, are upregulated by PRDM1 alone or in combination with MAF. We examined our data and found that the co-inhibitory receptors/molecules LAG3, LILRB1, LILRB3, and CD244 were downregulated in PRDM1-/- cells, which may thus impair immune checkpoints in NK cells similarly to cytotoxic T-cells. We performed qRT-PCR on selected transcripts based on their significant alteration on RNA-sequencing and their potentially important biological functions, comparing LAG3, GNLY, PRF1, TOX2 and CCR4 expression in bulk donor #2 NK PRDM1-/- cells versus donor #2 NK WT cells.

The qRT-PCR results were concordant with our RNAsequencing results with higher levels of TOX2 and CCR4 and lower levels of LAG3, GNLY and PRF1 in PRDM1-/- cells (Online Supplementary Figure S8G). Furthermore, flow analysis indicated a concordant decrease in LAG3 protein expression in NK PRDM1-/- cells (Online Supplementary Figure S8H).

Discussion We and other groups have identified several frequent mutations in ENKTCL, including a number of potential tumor suppressor genes such as PRDM1, TP53, DDX3X and BCOR.5-8,25-27 We also found that ENKTCL has a marked DNA hypermethylation phenotype with inactivation of multiple tumor suppressors through this mechanism.5 Some of the tumor suppressor genes are inactivated by a combination of genetic and epigenetic mechanisms, including PRDM1 and DDX3X. The most common activating mutations involve the JAK/STAT3 pathway, affecting ~20-30% of cases.15,16 This may be related to the critical dependence of NK cells on IL-2 or IL-15, which signal through STAT3, STAT5A and STAT5B.25 Interestingly, there is frequent DNA methylation of PTPN6 (SHP1), a negative regulator of STAT3 and NK-cell receptor activation.28-32 As PRDM1 is a commonly inactivated tumor suppressor gene in ENKTCL, and as there is good evidence that PRDM1 regulates normal NK-cell as well as T-cell homeostasis,6,26 we therefore focused on PRDM1 deletion to elucidate the functions of PRDM1 in normal NK cells and how these relate to its tumor suppressor function. The major challenge in deciphering the role of a genetic aberration in the pathogenesis of a lymphoma is the fre-

Figure 3. Schematic illustration of CRISPR/Cas9-mediated insertion of GFP or DsRed fusion tag to PRDM1 exon 5 through a homology directed repair template to induce early termination of PRDM1 expression. (A) Guide RNA sgRNA2 sequence and schematic gene structure of PRDM1. Light green boxes and numbers represent the exons of the PRDM1 gene. The sequence of guide RNA sgRNA2 used for the homology directed repair (HDR) template-mediated CRISPR/Cas9 is shown above. (B) HDR templates were assembled with fluorescent tag GFP or DsRed open reading frame and a SV40 virus transcriptional stop signal (red box) in exon 5. (C) Western blot analysis of PRDM1 protein expression in PRDM1-/- NK cells and wild-type control cells (left panel). PRDM1 target gene MYC expression was measured by quantitative reverse transcription polymerase chain reaction (right panel). Target gene normalization and relative expression levels were determined as described in Figure 2.

2432

haematologica | 2021; 106(9)

Functional and oncogenic roles of PRDM1 in NK cells

quent presence of a large number of genetic and epigenetic changes in an established tumor as well as tumor heterogeneity. In addition, gene expression studies of the bulk tumor contain signals from the stromal elements as well as the tumor cells. We decided to develop an approach to study selected lesions in the normal cellular counterpart of the tumor, thereby determining the precise functional alteration induced by a single lesion or a known combination of

lesions. A prerequisite of this approach in studying NK-cell lymphoma is the ability to grow human NK cells in vitro for a sufficiently long time to allow genetic manipulation and functional studies. Normal primary NK cells have a limited lifespan in vitro, but these cells can grow for >3 months with a feeder cell line (modified K562 cell line) and IL-2, thus allowing for genetic manipulations, selection and characterization of the mutants in vitro.

Figure 4. Sanger sequencing of the edited region of the PRDM1 locus. . (A) The diagram shows the expected homology directed repair (HDR) template inserted into the PRDM1 locus. (B) The DsRed reading frame, followed by the SV40 stop signal, was inserted in-frame into the PRDM1 gene locus. The enlarged boxes show the junctional sequence between each adjacent fragment. Blue and red arrows indicate the primer pair located outside of the HDR template area to amplify the genome DNA fragment for Sanger sequencing. (C, D) Relative PRDM1 transcript levels of exon 5 and exon 2 in PRDM1-/- NK cells versus wild-type NK cells were measured by quantitative reverse transcription polymerase chain reaction (RT-PCR) and normalized to RPL13A. The expression levels were calculated by the 2^-DDCt method, and PRDM1 expression level in wild-type NK cells was set at 1.0.

haematologica | 2021; 106(9)

2433

G. Dong et al.

We identified the tumor suppressor function of PRDM1 in NK cells by knocking down PRDM1 by shRNA transduction.6 The recent development of CRISPR/Cas9 technology provides a powerful and versatile approach to edit the genome, thereby facilitating our understanding of the functional alterations induced by specific genetic alterations. Gene inactivation by methylation or mutation can be simulated by introducing small out-of-frame indels or knocking-in mutations. We were able to successfully KO selected tumor suppressor genes identified in ENKTCL, including mono-allelic and bi-allelic deletions. KO experiments rely on non-homologous end joining mechanisms and are very

robust, but in the absence of a readily selectable marker, we have to rely on cloning to select for modified cells. This is a long and demanding process and may introduce a selection bias. Therefore, we adopted another approach to knock-in a fluorescent protein gene to disrupt the PRDM1 locus while introducing a marker for selection. We were able to knock-in a GFP or DsRed gene into the PRDM1 locus through HDR so that mono- or bi-allelic KO cells could be identified by single or double fluorescence, respectively. This approach shortened the experimental time and generated a bulk population of PRDM1-/- cells to complement and validate the observations from the cloning approach.

Figure 5. Figure continued on following page.

2434

haematologica | 2021; 106(9)

Functional and oncogenic roles of PRDM1 in NK cells

Figure 5. Proliferation and cell cycle distribution of PRDM1-/- NK cells versus wild-type NK cells. Cell growth analysis was measured by CellTiter 96® AQueous One Solution. The growth curve shows the relative fold-change of OD490 compared to the first day. (A) PRDM1 was knocked out in NK cells (donor #1) using a Cas9sgRNA plasmid. (B) PRDM1 was knocked out of NK cells (donor #2) by Cas9/sgRNA ribonucleoprotein (RNP) and homology directed repair template (HDRT) electroporation. Cell proliferation was analyzed with the Click-iT™ Alexa Fluor 647-EdU Flow Cytometry assay during 5 consecutive days starting from the third day after fresh irradiated feeder cells had been added. (C) Flow analysis of clones #3 and #5 on day 3 and day 7 and donor #2 cells. FITC anti-CD56 antibody was used to identify NK cells. (D) Each staining of Alexa Fluor 647-EdU was performed in duplicate, and the figure shows the average. (E) Flow analysis of EdU assay on NK cells (donor #2). GFP positivity (for PRDM1-/- cells). (F) Cell cycle distribution was analyzed by cell DNA content staining with propidium iodide/RNase or DAPI with flow cytometric assay at the third day after fresh feeder cells had been added. The percentage of cells in each phase is shown in each box. Upper panel: corresponding WT cells and PRDM1 knocked-out cells produced using the Cas9-sgRNA plasmid method. Lower panel: corresponding WT cells and PRDM1 knocked-out cells produced using the Cas9/sgRNA RNP and HDRT method.

We were able to generate multiple homozygous PRDM1 KO clones from different donors and demonstrated that PRDM1 KO cells had a much higher cloning efficiency, a faster growth rate and a higher percentage of cycling cells than WT cells. There was also a modest reduction in apoptosis. The KO cells were also able to maintain their growth and proliferation for a longer period in vitro. These observations were largely validated using the bulk population of PRDM1 KO cells, although there were some differences, which may have been related to the age of the cells in culture. When experiments were performed with younger cells, they exhibited higher growth and proliferation potentials. These observations suggested that PRDM1 is a key negative regulator of NK cells, and removing this control allows a striking increase in cloning efficiency, proliferation, growth and lifespan. This corroborated our previous observations6 that primary NK cells could proliferate better with shRNA knock-down of PRDM1, and that the reconstitution of PRDM1 through retroviral transduction into a PRDM1-deficient NK-cell lymphoma line (KHYG) was associated with G2/M arrest, increased apoptosis and a strong negative selection pressure. This again strongly supports our findings that knock-out of PRDM1 by CRISPR/Cas9 in primary NK promotes cell proliferation. In PRDM1 function rescue experiments, the rate of cell haematologica | 2021; 106(9)

growth of pMIG-PRDM1 electroporated NK PRDM1 KO cells was significantly repressed compared to that of cells electroporated with empty vector (Online Supplementary Figure S9). Thus, elimination of PRDM1 likely contributes to the malignant transformation of NK-cells. To further understand the basis of the functional changes induced by PRDM1 KO, we compared the gene expression profiles of PRDM1-/- and WT NK-cell clones. As expected, there was a higher level of MYC and activation of the MYC signature with the loss of PRDM1, since MYC is a direct target of PRDM1. Many proliferation- and cell cyclerelated pathways were highly enriched, including upregulation of many genes regulated by the DREAM complex. Signatures associated with IL-6, IL-15 and IL-2 stimulation were enriched, indicating that functional activities related to these cytokines are normally repressed by PRDM1. Similarly, TCR/NK-cell signaling was negatively regulated in WT cells compared with KO cells, indicating that removal of PRDM1 facilitates activation of these pathways. Additional mechanisms may contribute to driving the cell cycle, proliferation and survival, including the downregulation of the pro-apoptotic factor BIM and the upregulation of additional factors such as TLR4, TOX2, CCR4, VGEFA, MYB, BCAT1, FGFR1 and SIPR1. We repeated the experiments with another gene editing approach without subsequent cloning and obtained similar 2435

G. Dong et al.

Figure 6. Apoptosis of PRDM1-/- NK cells versus wild-type NK cells. Apoptotic cells were analyzed with annexin V/propidium iodide (or DAPI) staining and a flow cytometry assay. Numbers in Q3 (percentage of APC or Alexa Fluor 647 annexin V-positive and propidium iodide- or DAPI-negative cells) were considered the early apoptotic cells. FITC anti-CD56 antibody staining was used to identify NK cells. Each staining was performed in duplicate. (A) Flow analysis of clones #3 and #5 from NK cells (donor #1) on day 3 and day 7 after addition of feeder cells. (B) Statistical analysis of the average percentage. (C) Experiments were repeated on NK cells (donor #2) edited by Cas9/sgRNA RNP plus homology directed repair. GFP positivity for PRDM1-/- cells and FITC CD56 positivity to identify NK cells.

results, supporting the validity of the findings. Thus, the loss of PRDM1 altered the transcriptome with upregulation of MYC, MYB and many pathways associated with growth and proliferation, including those associated with cytokine stimulation and receptor signaling. On the other hand, pathways associated with normal cellular functions including cytotoxic functions were down-regulated, suggesting that the loss of PRDM1 shifted the cell toward proliferation and survival rather than the performance of its normal effector function. The loss of immune checkpoint molecules may be more relevant in the in vivo setting, in which the cells may be able to escape from extrinsic controls, and could be relevant to lymphomagenesis. CRISPR/Cas9 may generate off-target modifications that could potentially compromise our data and interpretation. We examined a number of clones for mutations using a custom capture platform (Online Supplementary Table S5) with an extensive set of genes known to harbor lymphoma-associated mutations and did not observe any mutations in these genes. The observation suggested that the CRISPR/Cas9 modifications did not induce off-target mutations of known oncogenic drivers and that our observations in the PRDM1 KO cells were likely PRDM1-specific. Our approach cannot exclude the editing of genetic loci not examined. Therefore, we used a different approach to generate KO NK cells from different donors, with different guide RNA, and without subsequent cloning. Since off-target editing is unlikely to be the same in different cells and with a different guide RNA, critical findings were con2436

firmed in our second approach with modification of a bulk population. One of our long-term goals is to study cooperative effects of two or more mutations. With the PRDM1-deleted background, we were able to generate double mutant (i.e. PRDM1-/-/TP53+/-, PRDM1-/-/DDX3X+/-, and PRDM1-/-/ PTPN6+/-) clones successfully using the plasmid/cloning approach. The growth of the NK cells started to slow down after prolonged in vitro culture. Therefore, to accelerate the experiments and to reduce secondary changes that may accumulate in prolonged culture, we tested the feasibility of simultaneously editing multiple genes in NK cells. Although it is feasible to modify two genes simultaneously, modifying three genes simultaneously was very inefficient using our current approach, probably partly because the cell cannot tolerate multiple double-stranded DNA breaks. Using the plasmid CRISPR/Cas9 KO strategy, we have only isolated heterozygous deleted clones, but with the current, more efficient Cas9/sgRNA RNP electroporation procedure, besides inserting a fluorescent protein gene in one allele, 70% of the other alleles were also modified by indels as analyzed by an online software ICE (Synthego, CA, USA) (Online Supplementary Figure S10). This suggested a more promising homozygous gene editing by the RNP electroporation method. The CRISPR/Cas9 technology is rapidly advancing, and innovative approaches can be incorporated in the future as they appear. While the loss of function of PRDM1 is frequent5,6 and likely to be one of the early alterations in the development haematologica | 2021; 106(9)

Functional and oncogenic roles of PRDM1 in NK cells

Figure 7. RNA-sequencing analysis of PRDM1-/- cells versus wild-type NK cells. (A) Unsupervised clustering of gene expression profiling data of PRDM1 knock-out clones and cell age-matched parental wild-type NK cells. (B) Volcano plot of the differentiated transcriptional profile between PRDM1-/- and PRDM1+/+ NK cells. (C) Heatmap of genes differentially expressed between PRDM1-/- and PRDM1+/+ NK cells. (D) Summary of gene set enrichment analysis.

of NK-cell malignancies, it probably occurs after EBV infection of the NK cells. The ideal NK-cell lymphoma model may need to be derived from EBV-infected NK cells. We have not been successful in obtaining a viable expanding population of EBV-infected WT or PRDM1 KO cells using EBV from Akata cells as reported.33,34 Further studies will need to be performed to find the appropriate conditions and developmental stage of NK cells for EBV infection and persistence. In summary, we have reported a disease modeling approach through the introduction of a tumor-driving mutation into normal NK cells through genetic editing. We examined the functional consequences of PRDM1 deletion and elucidated the major pathways through which PRDM1 mediates its homeostatic control of NK cells. The recent development of CRISPR/Cas9 and long-term culture technologies enables selected lesions to be introduced singly or in combination into normal human NK cells. Associated functional alterations can then be assessed in the absence of the noise arising from the many other abnormalities present in tumor samples. This provides a powerful approach to dissect oncogenic interactions, thereby facilitating our understanding of the mechanistic basis of their cooperativity in oncogenesis. Future development of the technology will improve the range, speed and specificity of genetic haematologica | 2021; 106(9)

editing, making this a feasible approach for studying functional changes resulting from a combination of oncogenic events and the essential changes necessary in the generation of a lymphoma. This approach can be applied to the study of T- and B-cell lymphomas and is particularly valuable in the study of tumors for which authentic cell lines or animal models are not available. Disclosures No conflicts of interest to disclose. Contributions GD and WCC conceived and designed the project; GD, YL and LL performed the NK-cell cloning, CRISPR experiments, and KO cell functional studies; XuL, YS and XiL performed the expression level assay and cell maintenance; CL helped with the CRISPR sgRNA design and HDRT construction; LK helped with the plasmid construction and identification of KO clones; JW performed RNA-sequencing experiments; AB, JI, QG and TWM analyzed the data; GD, YL, JI and WCC wrote and finalized the manuscript. Acknowledgments We thank Dr. Dean A. Lee, MD, PhD (Division of Hematology/Oncology/BMT, Nationwide Children’s Hospital, 2437

G. Dong et al.

Columbus, OH, USA) for providing us feeder cells for NK-cell stimulation and Ni Feng and Lucy Brown from the City of Hope Analytical Cytometry Core (Duarte, CA, USA) for flow cytometric sorting. Research reported in this publication also included work performed in the Integrative Genomics Core and Gene Editing and Viral Vector Core in City of Hope.

References 1. Yang Y, Cao JZ, Lan SM, et al. Association of improved locoregional control with prolonged survival in early-stage extranodal nasal-type natural killer/T-cell lymphoma. JAMA Oncol. 2017;3(1):83-91. 2. Li X, Cui Y, Sun Z, et al. DDGP versus SMILE in newly diagnosed advanced natural killer/T-cell lymphoma: a randomized controlled, multicenter, open-labelsStudy in China. Clin Cancer Res. 2016; 22(21):5223-5228. 3. Tse E, Kwong YL. Diagnosis and management of extranodal NK/T cell lymphoma nasal type. Expert Rev Hematol. 2016; 9(9):861-871. 4. Boi M, Zucca E, Inghirami G, Bertoni F. PRDM1/BLIMP1: a tumor suppressor gene in B and T cell lymphomas. Leuk Lymphoma. 2015;56(5):1223-1228. 5. Kucuk C, Hu X, Jiang B, et al. Global promoter methylation analysis reveals novel candidate tumor suppressor genes in natural killer cell lymphoma. Clin Cancer Res. 2015;21(7):1699-1711. 6. Kucuk C, Iqbal J, Hu X, et al. PRDM1 is a tumor suppressor gene in natural killer cell malignancies. Proc Natl Acad Sci U S A. 2011;108(50): 20119-20124. 7. Karube K, Nakagawa M, Tsuzuki S, et al. Identification of FOXO3 and PRDM1 as tumor-suppressor gene candidates in NKcell neoplasms by genomic and functional analyses. Blood. 2011;118(12):3195-3204. 8. Iqbal J, Kucuk C, Deleeuw R, et al. Genomic analyses reveal global functional alterations that promote tumor growth and novel tumor suppressor genes in natural killer-cell malignancies. Leukemia. 2009;23(6):1139. 9. Boi M, Rinaldi A, Kwee I, et al. PRDM1/BLIMP1 is commonly inactivated in anaplastic large T-cell lymphoma. Blood. 2013;122(15):2683-2693. 10. Savage HP, Yenson VM, Sawhney SS, et al. Blimp-1-dependent and -independent natural antibody production by B-1 and B-1derived plasma cells. J Exp Med. 2017;214(9):2777-2794. 11. Cong L, Ran FA, Cox D, et al. Multiplex

2438

Funding This project was supported in part by the National Cancer Institute of the National Institutes of Health under grant number P30CA033572. It was also partly supported by the Dr. Norman and Melinda Payson Professorship in Hematologic Cancers and Tony Stephenson Lymphoma Center of City of Hope.

genome engineering using CRISPR/Cas systems. Science. 2013;339(6121):819-823. 12. Dobashi A, Tsuyama N, Asaka R, et al. Frequent BCOR aberrations in extranodal NK/T-cell lymphoma, nasal type. Genes Chromosomes Cancer. 2016;55(5):460-471. 13. Jiang L, Gu ZH, Yan ZX, et al. Exome sequencing identifies somatic mutations of DDX3X in natural killer/T-cell lymphoma. Nat Genet. 2015;47(9):1061-1066. 14. Lee S, Park HY, Kang SY, et al. Genetic alterations of JAK/STAT cascade and histone modification in extranodal NK/T-cell lymphoma nasal type. Oncotarget. 2015; 6(19):17764-17776. 15. Koo GC, Tan SY, Tang T, et al. Janus kinase 3-activating mutations identified in natural killer/T-cell lymphoma. Cancer Discov. 2012;2(7):591-597. 16. Sim SH, Kim S, Kim TM, et al. Novel JAK3activating mutations in extranodal NK/Tcell lymphoma, nasal type. Am J Pathol. 2017;187(5):980-986. 17. Roth TL, Puig-Saus C, Yu R, et al. Reprogramming human T cell function and specificity with non-viral genome targeting. Nature. 2018;559(7714):405-409. 18. Ran FA, Hsu PD, Wright J, et al. Genome engineering using the CRISPR-Cas9 system. Nat Protoc. 2013;8(11):2281-2308. 19. Mian MF, Lauzon NM, Andrews DW, et al. FimH can directly activate human and murine natural killer cells via TLR4. Mol Ther. 2010;18(7):1379-1388. 20. Stolberg VR, Martin B, Mancuso P, et al. Role of CC chemokine receptor 4 in natural killer cell activation during acute cigarette smoke exposure. Am J Pathol 2014; 184(2):454-463. 21. Chen WS, Kitson RP, Goldfarb RH. Modulation of human NK cell lines by vascular endothelial growth factor and receptor VEGFR-1 (FLT-1). In Vivo. 2002; 16(6):439-445. 22. Vong QP, Leung WH, Houston J, et al. TOX2 regulates human natural killer cell development by controlling T-BET expression. Blood. 2014;124(26):3905-3913. 23. Kallies A, Carotta S, Huntington ND, et al. A role for Blimp1 in the transcriptional network controlling natural killer cell matura-

tion. Blood. 2011;117(6):1869-1879. 24. Chihara N, Madi A, Kondo T, et al. Induction and transcriptional regulation of the co-inhibitory gene module in T cells. Nature. 2018;558(7710):454-459. 25. Kucuk C, Jiang B, Hu X, et al. Activating mutations of STAT5B and STAT3 in lymphomas derived from gammadelta-T or NK cells. Nat Commun 2015;6:6025. 26. El-Tayeb A, Iqbal J, Behrenswerth A, et al. Nucleoside-5'-monophosphates as prodrugs of adenosine A2A receptor agonists activated by ecto-5'-nucleotidase. J Med Chem. 2009;52(23):7669-7677. 27. Kucuk C, Hu X, Iqbal J, et al. HACE1 is a tumor suppressor gene candidate in natural killer cell neoplasms. Am J Pathol. 2013; 182(1):49-55. 28. Demosthenous C, Han JJ, Hu G, et al. Loss of function mutations in PTPN6 promote STAT3 deregulation via JAK3 kinase in diffuse large B-cell lymphoma. Oncotarget. 2015;6(42):44703-44713. 29. Sharma Y, Ahmad A, Bashir S, et al. Implication of protein tyrosine phosphatase SHP-1 in cancer-related signaling pathways. Future Oncol. 2016;12(10):12871298. 30. Yin S, Wu H, Lv J, et al. SHP-1 arrests mouse early embryo development through downregulation of Nanog by dephosphorylation of STAT3. PloS One 2014; 9(1):e86330. 31. Nakamura MC, Niemi EC, Fisher MJ, et al. Mouse Ly-49A interrupts early signaling events in natural killer cell cytotoxicity and functionally associates with the SHP-1 tyrosine phosphatase. J Exp Med. 1997; 185(4):673-684. 32. Mahmood S, Kanwar N, Tran J, et al. SHP1 phosphatase is a critical regulator in preventing natural killer cell self-killing. PloS One. 2012;7(8):e44244. 33. Isobe Y, Sugimoto K, Yang L, et al. EpsteinBarr virus infection of human natural killer cell lines and peripheral blood natural killer cells. Cancer Res. 2004;64(6):2167-2174. 34. Isobe Y, Sugimoto K, Yang L, et al. EpsteinBarr virus infection of human natural killer cell lines and peripheral blood natural killer cells. Cancer Res. 2004;64(6):2167-2174.

haematologica | 2021; 106(9)

ARTICLE

Platelet Biology & its Disorders

CAMT-MPL: congenital amegakaryocytic thrombocytopenia caused by MPL mutations heterogeneity of a monogenic disorder a comprehensive analysis of 56 patients

Ferrata Storti Foundation

Manuela Germeshausen and Matthias Ballmaier Central Research Facility Cell Sorting, Hannover Medical School, Hannover, Germany

ABSTRACT

ongenital amegakaryocytic thrombocytopenia caused by deleterious homozygous or compound heterozygous mutations in MPL (CAMT-MPL) is a rare inherited bone marrow failure syndrome presenting as an isolated thrombocytopenia at birth progressing to pancytopenia due to exhaustion of hematopoietic progenitors. The analysis of samples and clinical data from a large cohort of 56 patients with CAMT-MPL resulted in a detailed description of the clinical picture and reliable genotype-phenotype correlations for this rare disease. We extended the spectrum of CAMT causing MPL mutations regarding number (17 novel mutations) and impact. Clinical courses showed great variability with respect to the severity of thrombocytopenia, the development of pancytopenia and the consequences from bleedings. The most severe clinical problems were (i) intracranial bleedings pre- and perinatally and the resulting long-term consequences, and (ii) the development of aplastic anemia in the later course of the disease. An important and new finding was that thrombocytopenia was not detected at birth in a quarter of the patients. The rate of non-hematological abnormalities in CAMT-MPL was higher than described so far. Most of the anomalies were related to the head region (brain anomalies, ocular and orbital anomalies) and consequences of intracranial bleedings. The present study demonstrates a higher variability of clinical courses than described so far and has important implications on diagnosis and therapy. The diagnosis CAMT-MPL has to be considered even for those patients who are inconspicuous in the first months of life or show somatic anomalies typical for other inherited bone marrow failure syndromes.

Introduction Congenital amegakaryocytic thrombocytopenia (CAMT, MIM #604498) is a rare inherited bone marrow failure syndrome (IBMFS) which usually presents as severe thrombocytopenia at birth without specific characteristics and progresses to aplastic anemia during the first years of life.1,2 Deleterious mutations in MPL coding for the thrombopoietin receptor have first been identified as single molecular cause of CAMT,3,4 but the disease is now regarded to be genetically heterogeneous.5 Indeed, mutations in the gene for thrombopoietin (THPO) have been recently described in some of these patients.6-8 Furthermore, newborns with other IBMFS like Dyskeratosis congenita, Fanconi anemia, MECOM associated syndrome or microdeletion syndromes can present phenotypically as CAMT since pathognomonic signs of these syndromes might be not yet apparent.9-12 In the following we use the term CAMT-MPL for the IBMFS caused by biallelic mutations in MPL. Previous descriptions of CAMT-MPL are based on single case reports or small case series, not allowing for a comprehensive evaluation of the phenotypic spectrum of the disease.1,2,13 Over the last 20 years we analyzed samples and clinical data from patients suspicious for inherited thrombocytopenia and could identify 56 patients with CAMT-MPL. The aims of our analysis of clinical, genetic and laboratory data are (i) a detailed description of the clinical picture of CAMT-MPL, (ii) the establishment of genotype-phenotype correlations allowing for the prediction

haematologica | 2021; 106(9)

Haematologica 2021 Volume 106(9):2439-2448

Correspondence: MANUELA GERMESHAUSEN germeshausen.manuela@mh-hannover.de MATTHIAS BALLMAIER ballmaier.matthias@mh-hannover.de Received: May s4, 2020. Accepted: July 17, 2020. Pre-published: July 23, 2020. https://doi.org/10.3324/haematol.2020.257972

2439

M. Germeshausen and M, Ballmaier

of development of aplastic anemia and malignancies, and (iii) a better understanding of the thrombopoietin-MPL system in vivo.

Methods Patients Patient material and clinical data were provided after informed consent. The study was approved by the local ethics committee. Patients suspected to have CAMT were analyzed for mutations in MPL. Twenty-three of the 56 CAMT-MPL patients included in this study were part of earlier publications of our group,4,14,15 two were the subject of single case studies.16,17 Six further patients had an already known heterozygous MPL mutation and a seemingly unaffected second allele.

patients, allele frequency 57%, 12 novel, Table 2D) Two hotspots (amino acids 102-104: 18 alleles, 12 patients; proline residues 135-136: six alleles, five patients) account for 21% of all mutated alleles. Fifteen of 24 missense mutations are predicted to be deleterious by all applied algorithms, 23 of 24 by at least one of the algorithms (Table 2D). Due to the small number of individual cases it is difficult to predict clinical courses from the individual missense mutations. Specifically severe courses were observed in patients affected from p.Arg102Pro, p.Trp154Arg, and p.Leu169His. The latter one was found in three unrelated patients from Chile suggesting a founder mutation with regional significance.

Sequencing Mutational analyses were performed by Sanger sequencing from leukocyte derived genomic DNA as described previously.4

In silico analysis of mutation data PROVEAN,18 SIFT,19 Polyphen2,20 and MutationTaster21 algorithms were used for prediction of the effect of MPL mutations on protein function. Putative splicing mutations were evaluated by BDGP splice site prediction,22 MaxEntScan algorithm,23 and Human Splicing Finder (HSF 3.1).24

Flow cytometric analyses Flow cytometric analyses of CD110 expression on early hematopoietic progenitors were performed as described earlier.25

Thrombopoietin levels Thrombopoietin serum or plasma levels were measured using a commercially available enzyme-linked immunosorbent assay (ELISA) kit (Quantikine, R&D systems).

Results MPL mutations* We identified 56 patients with homozygous (n=39) or compound heterozygous (n=17) mutations in MPL (Tables 1; Online Supplementary Table S1). We detected 38 different mutations (Figure 1, Table 2), 17 out of them are novel (Tables 1 and 2; Online Supplementary Table S2). Six different nonsense mutations (allele frequency 20%; including three novel mutations) and three different frame shift deletions (allele frequency 13%) affected 20 different patients (Table 2A and B; Online Supplementary Table S2). Five different splice site mutations (allele frequency 10%, two novel) affected 11 patients (ten families). With the exception of c.391+5G>C, all are predicted to lead to a complete loss of function (Table 2C). Prediction was confirmed for c.79+2T>A by measurement of missing CD110 surface expression on hematopoietic progenitors (Figure 2, see below) and for 213-1G>A and c.79+2T>A by the severe course of the disease in the affected patients. In contrast, patients with the mutation c.391+5G>C, allowing a residual natural splicing,26 had a less severe course and measurable CD110 expression on hematopoietic progenitors (Figure 2). The majority of mutations in our patient cohort were missense mutations (24 different mutations in 35 2440

Figure 1. MPL alleles in CAMT patients All MPL mutations found in our cohort of congenital amegakaryocytic thrombocytopenia (CAMT) patients are depicted on the left side beneath the exon structure of the MPL transcript with corresponding numbering of bases (coding sequence). Every symbol represents a mutated allele in one patient. The derived protein structure with the functional domains of the receptor protein and corresponding numbering of amino acids is shown on the right. Circles: missense mutations; diamonds: nonsense mutations; triangles: frame shift deletions; equal signs: splice site mutations. Mutations with less severe phenotypes are marked in green color (see main text). SP: signal peptide, TM: transmembrane domain

haematologica | 2021; 106(9)

Heterogeneity in CAMT-MPL

Less severe courses (thrombocytopenia not detected at birth or onset of pancytopenia not in early childhood) were observed in patients with mutations p.Met8Arg, p.Asp295Tyr, p.Pro394Ser and missense mutations in exons 11 and 12 affecting the intracytoplasmic part of the receptor molecule (p.Leu524Arg, p.Pro581Leu, p.Leu594Trp). p.Met8Arg is the most N-terminal mutation in MPL described so far. The mutation is located in the signal peptide region of the MPL precursor protein and might affect signaling of the molecule as well as the function of this codon as a possible alternative translation initiation site. It was homozygously found in a patient from consanguineous parents first diagnosed with thrombocytopenia at the age of 9 months. The mutation p.Arg454Pro which is predicted to be benign by all applied prediction algorithms was homozy-

gously found in a patient presenting at the age of 2 years with a profound isolated hypomegakaryocytic thrombocytopenia. p.Arg102His is the third mutation affecting Arg102 in CAMT: p.Arg102Cys and p.Arg102Pro cause a severe phenotype of CAMT in patients15 and disturb intracellular trafficking of the MPL protein27 although p.Arg102His as well as p.Arg102Pro are predicted to be benign by SIFT and PROVEAN algorithms (Table 2D). * The nomenclature of sequence variants follows the recommendations of the Human Genome Variation Society (HGVS). A discription at the DNA level is provided in the Online Supplementary Table S2. Missense mutations are described at protein level, other mutations on DNA level (coding sequence). Amino acid substitutions are deduced from DNA sequencing results, the recommended parentheses have been omitted for better readability.

Table 1. Congenital amegakaryocytic thrombocytopenia patients included in this study.

patient ID

sex

intron/exon

CDS

protein

patient ID

sex

intron/exon

CAMT001 CAMT006 CAMT007 CAMT009 CAMT011

f f m f f

CAMT113

CAMT122 CAMT123 CAMT125 CAMT130

f f f f

CAMT133 CAMT136 CAMT137

f m f

CAMT034 CAMT036 CAMT039

f m f

CAMT138 CAMT140 CAMT144

f m f

CAMT043 CAMT050 CAMT052 CAMT055 CAMT058

f f f m m

CAMT157 CAMT159 CAMT160 CAMT163 CAMT167 CAMT168 CAMT169

m f m f f m m

CAMT059 CAMT067

m m

CAMT178

CAMT075 CAMT082

m m

CAMT179

CAMT083 CAMT087 CAMT092

m m f

p.Arg43Ter p.Arg90Ter p.Arg102Pro p.Phe126LeufsTer5 p.Pro275Thr p.Arg102Pro p.Arg102Pro p.Leu79GlufsTer84 p.Arg102Pro p.Arg43Ter p.Arg43Ter p.Arg43Ter p.Asp27fs p.Phe126LeufsTer5 p.Phe126LeufsTer5 splicing defect p.Phe126LeufsTer5 p.Arg43Ter p.Phe126LeufsTer5 p.Arg123Ter p.Arg102Cys p.Trp154Arg p.Trp435Cys p.Arg257Leu p.Pro136Leu p.Leu594Trp p.Trp410Ter p.Arg102Pro p.Phe104Ser p.Arg102Pro p.Arg102Pro splicing defect p.Trp435Cys p.Arg43Ter splicing defect p.Gln460Ter splicing defect p.Pro135Arg

m f f

m f f f f f m f f

c.127C>T c.268C>T c.305G>C c.378delT c.823C>A c.305G>C c.305G>C c.235_236delCT c.305G>C c.127C>T c.127C>T c.127C>T c.79+2T>A c.378delT c.378delT c.213-1G>A c.378delT c.127C>T c.378delT c.367C>T c.304C>T c.460T>C c.1305G>C c.770G>T c.407C>T c.1781T>G c.1230G>A c.305G>C c.311T>C c.305G>C c.305G>C c.391+5G>C c.1305G>C c.127C>T c.391+5G>C c.1378C>T c.79+2T>A c.404C>G

CAMT101 CAMT102 CAMT108

CAMT012 CAMT013 CAMT015 CAMT017 CAMT018 CAMT019 CAMT030 CAMT031 CAMT033

E2 E3 E3 E3 E5 E3 E3 E3 E3 E2 E2 E2 I1 E3 E3 I2 E3 E2 E3 E3 E3 E4 E8 E5 E4 E12 E8 E3 E3 E3 E3 I3 E8 E2 I3 E9 I1 E4

CAMT180 CAMT181

m f

CAMT183

E6 E6 E3 I11 I2 I3 E12 E4 E4 I3 E5 E8 E9 I3 E5 E9 E8 E3 E4 E4 E1 E3 E9 E5 E11 E3 E3 E3 E4 E4 E6 E2 E3 I9 E4

CAMT098

haematologica | 2021; 106(9)

CDS

protein

c.883G>C p.Asp295Tyr c.883G>C p.Asp295Tyr c.378delT p.Phe126LeufsTer5 c.1653delG p.Lys553ArgfsX75 c.212+1G>A splicing defect c.391+5G>C splicing defect c.1742C>T p.Pro581Leu c.460T>C p.Trp154Arg c.407C>A p.Pro136His c.391+5G>C splicing defect c.769C>T p.Arg257Cys c.1180C>T p.Pro394Ser c.1361G>C p.Arg454Pro c.391+5G>C splicing defect c.769C>T p.Arg257Cys c.1390A>G p.Arg464Gly c.1180C>T p.Pro394Ser c.304C>T p.Arg102Cys c.407C>T p.Pro136Leu c.506T>A p.Leu169His c.23T>G p.Met8Arg c.235_236delCT p.Leu79GlufsX84 c.1431G>A p.Trp477Ter c.805T>C p.Trp269Arg c.1571T>G p.Leu524Arg c.305G>A p.Arg102His c.235_236delCT p.Leu79GlufsX84 c.305G>C p.Arg102Pro c.506T>A p.Leu169His c.407C>G p.Pro136Arg c.944T>G p.Phe315Cys c.127C>T p.Arg43Ter c.305G>C p.Arg102Pro c.1469-2A>T splicing defect c.506T>A p.Leu169His

Congenital amegakaryocytic thrombocytopenia (CAMT) patients included in this study. Patients are listed with MPL mutations (bold: novel mutations) and predicted effect on the MPL protein; f: female, m: male; ID: identifier; I: intron; E: exon; CDS: coding DNA sequence.

2441

M. Germeshausen and M, Ballmaier

CD110 expression on hematopoietic progenitors We analyzed the expression of the MPL encoded protein CD110 on CD34+CD38lo hematopoietic progenitors25 from 30 CAMT-MPL patients and eight healthy donors (Figure 2). There was a clear correlation between real CD110 expression and the predicted effects from the mutation analysis on the one hand and between CD110 expression and clinical course on the other hand: CD110 expression was not measurable on cells from patients with nonsense or frame shift mutations and mutations predicted to lead to a complete loss of a splice site (Figure 2, group A). In the group of patients with missense mutations we observed more variation in CD110 surface expression (Figure 2, group B) which was correlated with clinical courses: the higher CD110 expression observed in two patients homozygously affected by p.Asp295Tyr (Figure 2, violet squares) was correlated with a less severe course (CAMT101 and CAMT102). Cells from the patient with the p.Arg454Pro mutation predicted to be benign showed a nearly normal surface expression of CD110 (Figure 2, green square). In contrast, in patients with the mutation p.Arg102Pro (homozygous or compound heterozygous with a null mutation) and a relative

severe course we measured a very low CD110 signal (Figure 2, blue squares).

Thrombopoietin plasma levels Plasma levels of thrombopoietin are inversely proportional to the total mass of functional MPL in the body due to a direct negative feedback loop. Healthy donors usually have thrombopoietin plasma levels below 30 pg/mL (range <30-196 pg/mL). In contrast, thrombopoietin plasma levels were markedly elevated in all samples from 40 patients in this study and ranged from 400 to >4,000 pg/mL (median 1,493 pg/mL; Online Supplementary Table S1). Within the group of CAMT-MPL patients we did not find a significant correlation between THPO levels and either MPL expression levels on early hematopoietic progenitors or severity of the disease, but patients predicted to have a total receptor deficiency had a higher median thrombopoietin level (median 1,685 pg/mL, n=13) compared to patients with mutations allowing for a residual activity of the receptor (median 1,472 pg/mL, n=27). In our study, the measurement of MPL expression on hematopoietic precursors was a better predictor of the clinical course than THPO levels. Unexpectedly low

Table 2. MPL mutations in congenital amegakaryocytic thrombocytopenia patients. All mutations found in our group of congenital amegakaryocytic thrombocytopenia (CAMT) patients are listed regarding their type in Tables 2A to D together with their predicted impact on the MPL protein and their incidence in our patient group (bold: novel mutations). 2A (nonsense mutations) and 2B (frame shift mutations): prediction according to MutationTaster21 with probability; 2C (splice mutations): prediction according to BDGP splice site prediction,22 MaxEntScan algorithm (MaxEnt),23 and Human Splicing Finder (HSF).24 MDD: maximal dependency decomposition (only for donor sites), MM: Markov model (1st order), WMM: weighted matrix method. 2D (missense mutations): prediction according to MutationTaster,21 PROVEAN,18 and SIFT19 algorithms with the respective score values.*The mutation previously referred to as c.1653+1delG (now c.1653delG) should be also regarded as a frame shift mutation since the predicted effect on the splice donor site is marginal (Table 2C) and the effect on the protein is caused mainly by the frame shift.56 Table 2A. Nonsense mutations in congenital amegakaryocytic thrombocytopenia patients.

CDS c.127C>T c.268C>T c.367C>T c.1230G>A c.1378C>T c.1431G>A

Exon

protein

MutationTaster

incidence

E2 E3 E3 E8 E9 E9

p.Arg43Ter p.Arg90Ter p.Arg123Ter p.Trp410Ter p.Gln460Ter p.Trp477Ter

disease causing / 1 disease causing / 1 disease causing / 1 disease causing / 1 disease causing / 1 disease causing / 1

ho: n=7; het: n=0 ho: n=1; het: n=0 ho: n=0; het: n=1 ho: n=1; het: n=0 ho: n=0; het: n=1 ho: n=1; het: n=0

CDS: coding DNA sequence; ho: homozygous individuals: het: heterozygous individuals.

Table 2B. Frame shift mutations in congenital amegakaryocytic thrombocytopenia patients.

CDS

Exon

protein

MutationTaster

incidence

c.235_236delCT c.378delT c.1653delG*

E3 E3 E11

p.Leu79Glufs*84 p.Phe126LeufsX5 p.Lys553ArgfsX75

disease causing / 1 disease causing / 1 disease causing / 1

ho: n=2; het: n=1 ho: n=3; het: n=3 ho: n=0; het: n=1

CDS: coding DNA sequence; E: exon; ho: homozygous individuals: het: heterozygous individuals.

Table 2C. Splice site mutations in congenital amegakaryocytic thrombocytopenia patients.

CDS

Intron

HSF prediction

MaxEnt (wt/mut)

MDD (wt/mut)

MM (wt/mut)

WMM (wt/mut)

incidence

c.79+2T>A c.212+1G>A

I1 I2

most prob. broken donor s. most prob. broken donor s.

9.16/0.97 9.00/0.82

13.98/5.79 11.78/3.60

9.62/1.43 9.27/1.08

10.39/2.21 7.15/-1.03

ho: n=1; het: n=1 ho: n=0; het: n=1

c.213-1G>A c.391+5G>C

I2 I3

most prob. broken acc. s. most prob. broken donor s.

7.90/-0.85 9.14/6.25

12.18/9.98

7.69/-1.06 8.32/4.76

6.71/-2.04 9.69/5.86

ho: n=0; het: n=1 ho: n=0; het: n=5

c.1469-2A>T (c.1653+1delG*

I9 I11

most prob. broken acc. s. new donor site 1 base 5'

8.25/-0.11 10.90/8.40

15.68/12.58

9.41/1.05 10.63/7.39

11.03/2.67 9.35/6.49

ho: n=0; het: n=1 ho: n=0; het: n=1)

CDS: coding DNA sequence; I: Intron; ho: homozygous; het: heterozygous. HSF: Human Splicing Finder; Max/En: MaxEntScan algorithm; wt: wild-type; mut: mutant; MM: Markov model; WMM: weighted matrix method.

2442

haematologica | 2021; 106(9)

Heterogeneity in CAMT-MPL

Table 2D. Missense mutations in congenital amegakaryocytic thrombocytopenia patients

Exon

protein

MutationTaster

PROVEAN

SIFT

incidence

c.23T>G c.304C>T

CDS

E1 E3

p.Met8Arg p.Arg102Cys

polymorphism / 0.963 disease causing / 1.000

damag. /0.004 toler. / 0.064

ho: n=1; het: n=0 ho: n=1; het: n=1

c.305G>A c.305G>C

E3 E3

p.Arg102His p.Arg102Pro

disease causing /0.983 disease causing / 0.995

toler. / 0.291 toler. / 0.082

ho: n=0; het: n=1 ho: n=4; het: n=5

c.311T>C c.404C>G

E3 E4

p.Phe104Ser p.Pro135Arg

disease causing / 0.975 disease causing / 0.995

damag. / 0.001 damag. / 0.000

ho: n=0; het: n=1 ho: n=0; het: n=1

c.407C>A c.407C>T

E4 E4

p.Pro136His p.Pro136Leu

disease causing / 0.991 disease causing / 1.000

damag. / 0.000 damag. / 0.000

ho: n=1; het: n=0 ho: n=0; het: n=2

c.407C>G c.460T>C

E4 E4

p.Pro136Arg p.Trp154Arg

disease causing / 0.995 disease causing / 0.984

damag. / 0.000 damag. / 0.000

ho: n=0; het: n=1 ho: n=2; het: n=0

c.506T>A c.769C>T

E4 E5

p.Leu169His p.Arg257Cys

disease causing / 0.903 disease causing / 1.000

damag. / 0.001 damag. / 0.002

ho: n=1; het: n=1 ho: n=0; het: n=2

c.770G>T c.805T>C

E5 E5

p.Arg257Leu p.Trp269Arg

disease causing /1.000 disease causing / 0.994

damag. / 0.005 damag. / 0.000

ho: n=1; het: n=0 ho: n=1; het: n=0

c.823C>A c.883G>C

E5 E6

p.Pro275Thr p.Asp295Tyr

disease causing / 0.985 disease causing / 0.998

damag. / 0.008 damag. / 0.002

ho: n=0; het: n=1 ho: n=2; het: n=0

c.944T>G c.1180C>T

E6 E8

p.Phe315Cys p.Pro394Ser

disease causing / 0.999 disease causing / 0.981

damag. / 0.003 damag. / 0.004

ho: n=0; het: n=1 ho: n=2; het: n=0

c.1305G>C c.1361G>C

E8 E9

p.Trp435Cys p.Arg454Pro

disease causing / 1.000 polymorphism / 1.000

damag. / 0.000 toler. / 0.391

ho: n=2; het: n=0 ho: n=1; het: n=0

c.1390A>G c.1571T>G

E9 E11

p.Arg464Gly p.Leu524Arg

polymorphism / 0.993 disease causing / 0.807

damag. / 0.006 damag. / 0.002

ho: n=1; het: n=0 ho: n=1; het: n=0

c.1742C>T c.1781T>G

E12 E12

p.Pro581Leu p.Leu594Trp

disease causing / 0.981 polymorphism / 0.986

damag. / 0.004 damag. / 0.000

ho: n=1; het: n=0 ho: n=0; het: n=1

CDS: coding DNA sequence; ho: homozygous individuals: het: heterozygous individuals.

THPO values, despite complete MPL deficiency, could be due to recent platelet transfusions or duration and condition of sample shipment.

Clinical phenotype Inheritance Twenty-nine (52%) of the patients in our cohort had consanguineous parents and were homozygous for the particular MPL mutation. Homozygous mutations in patients with no evidence for parental consanguinity (n=9) were mainly affected from the most prevalent mutation c.305G>C (n=4) or from mutations with a higher prevalence in a specific region (c.506T>A, see above) or ethnic group (c.79+2T>A).28 Some families had more than one affected patient: CAMT009 + CAMT031, CAMT018 + CAMT019 + CAMT036 + CAMT180, and CAMT133 + CAMT140 each belongs to large kindreds with a high degree of consanguinity. Other cases of CAMT, aplastic anemia or not otherwise specified “bleeding disease” are reported in these kindreds. CAMT101 and CAMT102 as well as CAMT130 and CAMT137 are siblings from non-consanguineous families. CAMT083 is the fetus of a second pregnancy of the mother of CAMT052. Bone marrow analysis during autopsy revealed normal cellularity with absent megakaryocytes.

Pregnancies, deliveries, symptoms at birth Pregnancies and deliveries were unremarkable in the majority of the cases. Median gestational age was 40 haematologica | 2021; 106(9)

weeks (n=34, range, 31-42 weeks of gestation [wGA]), mean birth weight 3,080 g (n=25, range, 1,545-4,280 g). Intracranial bleeding in utero was detected in some patients (seven of 46),† retrospectively in four of seven. Two children were delivered by cesarean section due to diagnosis of cerebral hemorrhage: CAMT013 with a hydrops fetalis due to Rhesus incompatibility (wGA 31) and CAMT123 after an intracranial bleeding in wGA 28 (wGA 38). Pregnancy of CAMT083 was terminated in wGA 22 because of very poor prognosis after intracranial bleeding. There were three other cesarean sections for reasons only related to the mother. We found a significant female predominance in our cohort (62.5%, P<0.05 according to χ²-test).2 This is in contrast to most of the other IBMFS in which boys are affected more often.29 We have no information about the number and sex ratio of miscarriages in the patients’ families as a possible hint for the female predominance.

Thrombocytopenia, bleeding Although thrombocytopenia at birth has been classified as one of the diagnostic hallmarks of CAMT so far, 13 of 52† patients in this study with available information showed no signs of thrombocytopenia at birth and no blood counts were taken. Twelve of 13 patients had mutations allowing for a residual MPL activity. In the remaining patients (39 of 52)† thrombocytopenia was detected at birth (n=38) or at termination of pregnancy (n=1). Available data for platelet counts at birth ranged from 1-36 G/L (median 15 G/L, n=30). Petechiae or pur2443

M. Germeshausen and M, Ballmaier

Table 3. Non-hematological findings in congenital amegakaryocytic thrombocytopenia patients.

Figure 2. MPL expression on CD34+CD38lo hematopoietic progenitors from congenital amegakaryocytic thrombocytopenia patients The figure shows CD110 expression levels, calculated from cumulative subtraction on CD34+CD38lo hematopoietic progenitors of patients with nonsense, frame shift mutations or splice site mutations predicted to lead to a complete loss of MPL function (A), missense mutations (B), or patients who are compound heterozygous for different types of mutations (mix), in comparison to normal donors (ND). Horizontal lines represent the mean and standard error of the mean. Data from samples with special genotypes are labeled with the same colors, respectively. A+B: red: p.Arg43Ter, orange: p.Phe126LeufsTer5, grey: c.79+2T>A; green: p.Arg454Pro, pink: p.Leu169His, violet: p.Asp295Tyr, blue: p.Arg102Pro (half: compound heterozygous with p.Phe104Ser); mix: dark green/blue: c.391+5G>C/p.Arg102Pro, dark green/black: c.391+5>C/p.Arg257Cys.

pura at birth or in the first week of life were the presenting symptoms of thrombocytopenia in the majority of cases (25 of 38)†. Only few patients (five of 38)† presented with severe bleedings at birth (n=3) or shortly thereafter (n=2). Intracranial bleedings was reported only during pregnancy (n=7), at birth (n=2) or within the first 4 weeks of life (n=4). Hematemesis as an indication of gastrointestinal bleeding was observed in one patient (at birth). In contrast to intracranial and gastrointestinal bleeding, severe episodes of epistaxis were reported mainly during the later stages of the disease (n=3). Data regarding platelet courses confirmed our concept of CAMT I and CAMT II distinguishing between patients with severely low platelet counts over the whole course of the disease due to loss-of-function mutations in MPL (CAMT I) and those patients showing a spontaneous increase of platelet counts in the first months of life due to a residual function of the receptor (CAMT II):4,14,15 platelet counts over 50 G/L (not transfused) within the first year of life have been documented for 14 of 33† patients. For nine more patients with a late diagnosis of thrombocytopenia we can also assume higher platelet counts in the first months of life. None of these 23 patients bore a mutation predicted to lead to a complete loss of function. Nearly all CAMT I and CAMT II patients demonstrated a further decline of platelet counts during the development of aplastic anemia. Platelet counts of heterozygously affected parents and siblings of patients were in a normal range with the exception of one parent (c.305G>C) with mild thrombocytopenia (130-150 G/L).

Development of pancytopenia Development of additional anemia or neutropenia and reduced bone marrow cellularity are signs of developing bone marrow exhaustion. Bone marrow analyses from the first 6 months of life usually showed normal cellularity with reduced or absent megakaryocytes (13 of 15).† Accordingly, most of the patients presented with isolated 2444

Data available No non-hematological findings Abnormality of fetal development hydrops fetalis Anomalies of the nervous system optic nerve hypoplasia cerebellar hypoplasia agenesis of corpus callosum Dandy-Walker anomaly arachnoid cyst ventriculomegaly colpocephaly hydrocephalus Mental/psychomotor retardation Abnormality of the skin eczema hypopigmentation atopic dermatitis Abnormalities of the eye impaired vision nystagmus strabismus Abnormalities of the face hypertelorism high palate small uvula Skeletal abnormalities Cardiovascular abnormalities Abnormalities of the genitourinary system Other abnormalities obstructive sleep apnea diabetes short stature

# of patients (with ICH)

HPO-designation

50 25 1 (1) 1 (1) 7 (6) 1 (1) 1 (1) 1 (1) 2 (2) 3 (2) 1 (1) 1 (1) 1 (1) 7 (5) 4 2 (0) 1 (0) 1 (0) 10 1 (1) 4 (3) 9 (5) 4 (1) 2 (1) 2 (0) 1 (0) 0 0 0

HP:0001197 HP:0001789 HP:0000707 HP:0000609 HP:0001321 HP:0001274 HP:0001305 HP:0100702 HP:0002119 HP:0030048 HP:0000238 HP:0001263 HP:0000951 HP:0000964 HP:0001010 HP:0001047 HP:0000478 HP:0000505 HP:0000639 HP:0000486 HP:0000271 HP:0000316 HP:0000218 HP:0010812 HP:0000924 HP:0001626 HP:0000119

2 (0) 1 (0) 1 (0) 1 (0)

HP:0002870 HP:0000819 HP:0004322

Non-hematologic abnormalities found in congenital amegakaryocytic thrombocytopenia (CAMT) patients with homozygous or compound heterozygous mutations in MPL with the respective designations according to the human phenotype ontology (HPO)57 Number of patients with documented intracranial hemorrhages (ICH) in parentheses.

thrombocytopenia at birth or in the first weeks thereafter (36 of 51).† Only six patients showed signs of multi-lineage cytopenia within the first 6 months of live, four of them were already anemic immediately after birth (hemoglobin 56-76 g/L). For two patients a hypocellular bone marrow is documented in the first month after birth. Provided the data available at time of analysis only seven of 49 patients showed no signs of developing pancytopenia. Five of these patients were younger than 2 years at last examination. From the remaining 42 patients with documented aplasia only 24% were older than 4 years (n=10). Seventy-six percent (n=32) were younger than 4 years, half of them even younger than 2 years (Online Supplementary Figure S1). CAMT011 is the only patient without any signs of pancytopenia till adulthood. † Here and in all subsequent ratios, the denominator is the number of patients for whom information is available for a specific parameter. E.g., for “intracranial bleeding” 46 is the number of questionnaires with a yes-no-information from the attending physicians.

haematologica | 2021; 106(9)

Heterogeneity in CAMT-MPL

Table 4. Hematopoietic stem cell transplantation in congenital amegakaryocytic thrombocytopenia.

# of pts with available info Age at 1st HSCT [y]

Median: 3.55 y, range: 0.6 – 11 y <1 year 1-5 years

>5 years

HSC donor

HSC source

HLA matched related donor haploident related donor matched unrelated donor mismatched unrelated donor BM PBSC CB

% Pos outcome (%)

5 23

13 61

4/4 (100) 15/17 (88)

7/9 (78)

19 3 11 1 10 7 5

56 9 32 3 48 33 19

12/12 (100) 4/4 (100) 5/8 (63) 1*/1 (100) 8/10 (80) 6/7 (86) 3/4 (75)

information about neg. outcome CAMT039: TRD (no details, BM, MUD) CAMT015: death after GvHD-induced bronchiolitis obliterans (BM, MUD) CAMT007: death after graft rejection and sepsis (PBSC, MUD) CAMT157: death after GvHD grade 4 (lung, skin) and sepsis (CB)

The table summarizes the available information regarding age of hematopoietic stem cell transplantation (HSCT), donor and source of hematopoietic stem cell (HSC) and outcome. Information about outcome was not available for all transplantations. BM: bone marrow; CB: cord blood, PBSC: peripheral blood stem cells; pts: patients; TRD: transplantation related death;*: CB, 1 mismatch. GvHD: graft-versus-host disease; MUD: matched unrelated donor.

Chromosomal anomalies, leukemic development

Treatment

Cytogenetic data were inconspicuous for most of the patients with available data (n= 23 of 27, 85%). An abnormal karyotype has been detected in 4 patients: t2;11 (5%) in CAMT009,14 a not further specified additional marker chromosome (94%) in CAMT01314 and monosomy 7 in CAMT04330 (50%) and CAMT067 (13-30%). The latter has been diagnosed with MDS. All underwent hematopoietic stem cell transplantation (HSCT) because of aplastic anemia. In none of the patients a development of overt leukemia has been reported in the period of record.

Thirty-seven of 45 documented cases of our patient group received platelet transfusions, most of them transiently in a period immediately after diagnosis of severe thrombocytopenia and/or during the aplastic stage of the disease. During the advanced stage of the pancytopenia the patients often received erythrocyte transfusions (16 of 45). Neutropenia and associated infections were treated with antibiotics; two of the patients were treated with recombinant granulocyte colony-stimulating factor. Half of the patients (25 of 50) have been initially treated with immunoglobulins (23 of 50) or corticosteroids (15 of 50). Interestingly, three patients responded with a transient increase in platelet counts, which initially misled the diagnosis but none showed a persistent response. The only available curative treatment for CAMT-MPL is HSCT. Thirty-eight of 51 patients in our group were treated with HSCT, for another ten HSCT was planned for the near future. For 26 of 30 patients with information about the post-transplant course a positive outcome was documented (87%). The available information about age of transplantation, donor, stem cell source and outcome is summarized in Table 4. Three patients were unsuccessfully treated with recombinant IL-11 (oprelvekin). Two of them showed a slight and transient increase in platelet counts, followed by a prolonged phase of severe thrombocytopenia, which could be explained by an exhaustion of residual megakaryopoiesis by stimulation of cytoplasmic maturation.

Non-hematological abnormalities The rate of non-hematological abnormalities in our CAMT-MPL patients was markedly higher than reported: 50% of the patients with available data (25 of 50) had non-hematological abnormalities appearing as structural abnormalities or other abnormal clinical findings (Table 3). Most of the reported anomalies were related to the head region: brain anomalies (n=7), ocular and orbital anomalies (n=10), especially strabismus (n= 9), nystagmus (n=4) and facial abnormalities (n=4). Mental or psychomotor retardation was observed in seven patients, mostly correlated with brain anomalies. Intracranial bleedings are documented for five of seven patients with mental or psychomotor retardation, for six of seven with brain anomalies, and for six of ten with ocular anomalies (Table 3). Interestingly, we found some anomalies which are typical for other IBMFS and which misled the first diagnosis: eczema (n=2), hypopigmentation (n=1), high palate and/or small uvula (n=2). No skeletal, cardiac or urogenital abnormalities were observed. There was no correlation between type or localization of MPL mutations and non-hematological abnormalities. haematologica | 2021; 106(9)

Congenital amegakaryocytic thrombocytopenia with only one affected MPL allele In six patients with clinical diagnosis of CAMT we found only a single mutated allele (Online Supplementary Table S3), as judged by reproducible balanced distribution 2445

M. Germeshausen and M, Ballmaier

of both alleles from independently isolated genomic DNAs in five patients and an unbalanced distribution of both alleles in one patient.Two of these six patients were from families with other members affected by CAMTMPL: CAMT139 was heterozygously affected by the missense mutation c.1390A>G, homozygously detected in her sister CAMT138. Both sisters had similar clinical and laboratory findings. Both parents were heterozygous carrier of the mutation without any hematological problems. Thrombopoietin plasma levels were high in both sisters but not in the parents. However, in contrast to both parents who demonstrated a balanced distribution of both alleles the wild-type allele in CAMT139 was reproducibly markedly underrepresented (approximatly 20%), arguing for somatic mosaicism. Patient CAMT065 heterozygously harbored the c.127C>T nonsense mutation, which was homozygously found in his cousins CAMT036 and CAMT018. Besides these familial cases we identified four other patients with heterozygous MPL mutations. In one of these patients (CAMT129) we found CD110 expression on early hematopoietic progenitors comparable to that from patients with a predicted complete loss of the receptor (Figure 2). In patient CAMT73 we found a novel nonsense mutation in exon 7 together with a synonymous substitution c.585T>C (p.Pro195=). Although synonymous mutations can significantly influence protein levels via changes in translation efficiency,31 both codons are nearly equally used in human genes, and the mutation has no predicted effect on splicing.

Discussion This report summarizes the results of a long term study on the largest cohort of patients with CAMT-MPL caused by biallelic mutations in MPL. We limited our cohort to this group of patients (i) to provide a reliable definition of the clinical picture of CAMT-MPL, (ii) to define the effects of the MPL/THPO system in humans, and (ii) to allow for evidence based treatment recommendations. CAMT has been used in the past to describe an IBMFS with no characteristic malformations presenting as isolated thrombocytopenia at birth progressing to a general bone marrow failure.13,32 However, large differences in the reported percentages for MPL mutations, for the development of aplastic anemia and leukemia, and for somatic malformations reveal differences in the definition of this disease.29,33-35 This together with misleading combinations of findings from the pre-molecular era involves the risk of mistreatment e.g., HSCT of patients with CAMT due to THPO mutations. The most severe clinical problems for patients with CAMT-MPL are (i) - so far underestimated - pre- and perinatal bleedings and the resulting long-term consequences thereof, and (ii) the development of aplastic anemia in the later course of the disease. Severe bleedings, especially intracranial bleedings, occur mainly pre- or perinatally but much less frequently after the first weeks of life despite partly very low platelet counts. Specific functional deficits in neonatal platelets like a decreased P-Selectin expression and reduced platelet activation and secretion36-38 could be a possible explanation for the high bleeding tendency pre- or perinatally in combination with the thrombocytopenia. Furthermore, both life-span 2446

and thrombin dependent activation of platelet GPIIb/IIIa are markedly reduced in neonatal Mpl-/- mice compared to adult Mpl-/- mice.37 Our results indicate a possible functional impairment of platelets also in human fetuses and newborns with MPL defect which is in contrast to the assumption of a normal function of Mpl-/- platelets.39 Development of aplastic anemia due to exhaustion of three lineage hematopoiesis is a characteristic feature of CAMT-MPL and reveals the essential role of MPL for the maintenance of hematopoietic stem cells:40 almost all patients inevitably develop a fatal bone marrow failure. In our study we observed only one patient with an isolated thrombocytopenia until adulthood. In the literature one further patient is described with stable thrombocytopenia in the period of record.41 Half of the patients in our cohort exhibit non-hematopoietic abnormalities. This is in contrast to the characterization of CAMT as an IBMF with no physical anomalies (OMIM). Most of the non-hematopoietic abnormalities seen in our cohort are related to the brain and the eye. For neurological abnormalities, which have been reported for other CAMTMPL patients it has been argued, that these could be a direct consequence of the roles of thrombopoietin and MPL in the brain.42-45 However, the high correlation between structural abnormalities in the brain and intracranial bleedings argues for a secondary effect of thrombocytopenia. Indeed, most of these structural abnormalities observed in our cohort have also been reported as a consequence of intracranial bleedings,46-48 even strabismus and nystagmus.49 This is further supported by the observation that higher incidences of ocular anomalies have also been described for other BMFS going along with thrombocytopenia (Fanconi anemia, dendritic cells) but not for those with normal platelet counts (Diamond Blackfan anemia, Shwachman Diamond syndrome).50 Previous reports of other non-hematological abnormalities refer to CAMT patients with unreported or wild-type MPL genotype.14,51 Our data suggests that the primary effects of MPL deficiency are restricted to the hematopoietic system - most of the non-hematopoietic symptoms seem to be secondary to the thrombocytopenia or bone marrow failure. For other symptoms, especially those observed only in single cases or in highly consanguineous families52 we suppose that they emerged coincidentally. Although CAMT is regarded to be a preleukemic syndrome in most of the recent reviews, only weak evidence for this assumption exists. One single patient with CAMT and confirmed MPL mutation has been reported to develop a pre-B acute lymphoblastic leukemia.41 Increased accumulation of chromosomal aberrations, however, has been observed in our and previous studies.53 The exhaustion of hematopoietic stem cells due to MPL deficiency may be the reason for both, the acquisition of pre-leukemic cellular alterations due to increased hematopoietic stress, but also for early development of aplastic anemia leading to death or replacement of the hematopoietic system by means of HSCT, thereby preventing the development of overt leukemia. The debate about CAMT-MPL as a preleukemic syndrome therefore might be of less relevance. Genotype-phenotype correlations in CAMT-MPL have led us to our concept of CAMT I and II groups: a complete loss of MPL function results in persistently low platelet counts and a fast progression into pancytopenia in CAMT I patients whereas a residual function of the receptor leads to a milder course with a transient increase of platelet counts haematologica | 2021; 106(9)

Heterogeneity in CAMT-MPL

in the first year of life in CAMT II patients.4,14,15 The data from the present study allow for additional conclusions of clinical relevance: - the course of the disease is mainly determined by the type of mutation. The same MPL mutations lead to high similarities in the hematological courses of patients, even if they are from different families or different ethnical background (e.g., mild course in patients with c.391+5G>C). - the time course of pancytopenia development for patients with same MPL mutations is more variable than the course of thrombocytopenia. This could be caused by accelerated exhaustion of hematopoietic progenitors due to frequent bleedings or infections in some patients.54 - all patients with mutations leading to a complete loss of function (CAMT I) had a similar course with constantly severe thrombocytopenia. Platelet counts at birth and in the further course never exceed 50. and all of them showed a transition to pancytopenia. A complete MPL deficiency is probable in patients showing signs of aplastic anemia in the first months of life. - missense mutations predicted to allow a residual function of the MPL receptor lead to a more variable course of CAMT. The most severe phenotypes, comparable to CAMT I (severe thrombocytopenia, early development of aplasia), were observed in patients with mutations p.Leu169His and p.Trp154Arg. Milder phenotypes (late detection of thrombocytopenia and delayed development of aplasia) were observed in patients with mutations p.Asp295Tyr and p.Pro394Ser and missense mutations affecting the intracytoplasmic domain. - milder phenotypes with late development of aplastic anemia (respectively none during the period of record) have also been observed in patients with splice site mutations allowing for a residual normal splicing,26,41 namely c.391+5G>C and c.212+5G>A. - patients with germ line MPL mutations and a late onset form of amegakaryocytic thrombocytopenia or aplastic anemia (e.g., patients CAMT058, CAMT101, CAMT102 with moderate thrombocytopenia detected at the age of >2 years) should be also regarded as CAMT-MPL. This includes the patients previously described as familial aplastic anemia.55 - there may exist a small subgroup of CAMT II patients without development of pancytopenia. For patients with new mutations predicted to have minor impact on function or with mutations previously detected in patients with a mild course (namely c.391+5G>C, c.212+5G>A, or p.Pro275Thr) it might be appropriate to wait for first signs of bone marrow failure before proceeding to HSCT, especially if no appropriate family donor is available. - type and localization of MPL mutations are not predictive for pre- and perinatal intracranial hemorrhages. There are no differences in the frequency and severity of these

References 1. Ballmaier M, Germeshausen M. Advances in the understanding of congenital amegakaryocytic thrombocytopenia. Br J Haematol. 2009;146(1):3-16. 2. Ballmaier M, Germeshausen M. Congenital amegakaryocytic thrombocytopenia: clinical presentation, diagnosis, and treatment. Semin Thromb Hemost. 2011;37(6):673-681. 3. Ihara K, Ishii E, Eguchi M, et al. Identification of mutations in the c-mpl gene in congenital amegakaryocytic thrombocytopenia. Proc

haematologica | 2021; 106(9)

bleedings between patient groups CAMT I and CAMT II. The existence of a deleterious MPL mutation is a major risk factor for the occurrence of intracranial bleedings. - structural and clinical non hematologic abnormalities in CAMT-MPL are not correlated with specific mutations. - there is a small group of patients who present clinically as CAMT, but in whom a deleterious MPL mutation can only be detected in one allele. Possible explanations for the seeming inconsistency between genotype and phenotype include somatic mosaicism, deletions or changes in regulatory sequences that prevent the translation of a functional protein, or - rather unlikely especially in family cases - accidental coincidence. - a further consideration for clinical presentation of CAMT-MPL is whether, in addition to existing MPL mutations, mutations or functional single nucleotide polymorphisms in other genes or epigenetic differences are involved in the observed phenotypic heterogeneity of CAMT-MPL Our analysis of a large cohort of CAMT-MPL patients demonstrates a higher variability of clinical courses than described so far. The diagnosis CAMT-MPL has to be considered even for those patients who are inconspicuous in the first months of life or show somatic anomalies typical for other BMFS. Since almost all CAMT-MPL patients inevitably develop a fatal bone marrow failure that requires treatment with HSCT, all children with unclear forms of hypomegakaryocytic thrombocytopenia should be tested for MPL mutations. If molecular confirmation of CAMT is not possible, at least those IBMFS should be excluded for which HSCT is not an option (e.g., thrombopoietin production defect) or which need another transplantation regimen (e.g., Fanconi anemia, Diamond Blackfan anemia). Disclosures No conflicts of interest to disclose Contributions MG and MB designed and performed research, analyzed data, and wrote the manuscript. Acknowledgments The authors would like to thank all patients and their families who participated in this study. We are also grateful to the physicians who provided us with material and data from their patients. We would like to acknowledge the excellent technical assistance of Yvonne Peter and Christina Struckmann. Funding This work was supported in part by grants from the Federal Ministry of Education and Research (German Network on Congenital Bone Marrow Failure Syndromes) and by the transnational ERA-NET funding European Platelet Network (EUPLANE).

Natl Acad Sci U S A. 1999;96(6):3132-3136. 4. Ballmaier M, Germeshausen M, Schulze H, et al. c-mpl mutations are the cause of congenital amegakaryocytic thrombocytopenia. Blood. 2001;97(1):139-146. 5. Dokal I, Vulliamy T. Inherited bone marrow failure syndromes. Haematologica. 2010;95 (8):1236-1240. 6. Dasouki MJ, Rafi SK, Olm-Shipman AJ, et al. Exome sequencing reveals a thrombopoietin ligand mutation in a Micronesian family with autosomal recessive aplastic anemia. Blood. 2013;122(20):3440-3449.

7. Seo A, Ben-Harosh M, Sirin M, et al. Bone marrow failure unresponsive to bone marrow transplant is caused by mutations in thrombopoietin. Blood. 2017;130(7):875-880. 8. Pecci A, Ragab I, Bozzi V, et al. Thrombopoietin mutation in congenital amegakaryocytic thrombocytopenia treatable with romiplostim. EMBO Mol Med. 2018;10(1):63-75. 9. Basel-Vanagaite L, Dokal I, Tamary H, et al. Expanding the clinical phenotype of autosomal dominant dyskeratosis congenita caused by TERT mutations. Haematologica. 2008;

2447

M. Germeshausen and M, Ballmaier 93(6):943-944. 10. Giampietro PF, Verlander PC, Davis JG, Auerbach AD. Diagnosis of Fanconi anemia in patients without congenital malformations: an international Fanconi Anemia Registry Study. Am J Med Genet. 1997; 68(1):58-61. 11. Germeshausen M, Ancliff P, Estrada J, et al. MECOM-associated syndrome: a heterogeneous inherited bone marrow failure syndrome with amegakaryocytic thrombocytopenia. Blood Adv. 2018;2(6):586-596. 12. Christensen RD, Wiedmeier SE, Yaish HM. A neonate with congenital amegakaryocytic thrombocytopenia associated with a chromosomal microdeletion at 21q22.11 including the gene RUNX1. J Perinatol. 2013;33 (3):242-244. 13. Geddis AE. Congenital amegakaryocytic thrombocytopenia. Pediatr Blood Cancer. 2011;57(2):199-203. 14. King S, Germeshausen M, Strauss G, Welte K, Ballmaier M. Congenital amegakaryocytic thrombocytopenia: a retrospective clinical analysis of 20 patients. Br J Haematol. 2005;131(5):636-644. 15. Germeshausen M, Ballmaier M, Welte K. MPL mutations in 23 patients suffering from congenital amegakaryocytic thrombocytopenia: the type of mutation predicts the course of the disease. Hum Mutat. 2006;27(3):296. 16. Stoddart MT, Connor P, Germeshausen M, Ballmaier M, Steward CG. Congenital amegakaryocytic thrombocytopenia (CAMT) presenting as severe pancytopenia in the first month of life. Pediatr Blood Cancer. 2013;60(9):E94-96. 17. Ok Bozkaya I, Yarali N, Isik P, Unsal Sac R, Tavil B, Tunc B. Severe clinical course in a patient with congenital amegakaryocytic thrombocytopenia due to a missense mutation of the c-MPL gene. Turk J Haematol. 2015;32(2):172-174. 18. Choi Y, Chan AP. PROVEAN web server: a tool to predict the functional effect of amino acid substitutions and indels. Bioinformatics. 2015;31(16):2745-2747. 19. Kumar P, Henikoff S, Ng PC. Predicting the effects of coding non-synonymous variants on protein function using the SIFT algorithm. Nat Protoc. 2009;4(7):1073-1081. 20. Adzhubei IA, Schmidt S, Peshkin L, et al. A method and server for predicting damaging missense mutations. Nat Methods. 2010;7 (4):248-249. 21. Schwarz JM, Cooper DN, Schuelke M, Seelow D. MutationTaster2: mutation prediction for the deep-sequencing age. Nat Methods. 2014;11(4):361-362. 22. Reese MG, Eeckman FH, Kulp D, Haussler D. Improved splice site detection in Genie. J Comput Biol. 1997;4(3):311-323. 23. Yeo G, Burge CB. Maximum entropy modeling of short sequence motifs with applications to RNA splicing signals. J Comput Biol. 2004;11(2-3):377-394. 24. Desmet FO, Hamroun D, Lalande M, Collod-Beroud G, Claustres M, Beroud C. Human Splicing Finder: an online bioinformatics tool to predict splicing signals. Nucleic Acids Res. 2009;37(9):e67. 25. Ballmaier M, Holter W, Germeshausen M. Flow cytometric detection of MPL (CD110) as a diagnostic tool for differentiation of congenital thrombocytopenias. Haematologica. 2015;100(9):e341-344.

2448

26. Gandhi MJ, Pendergrass TW, Cummings CC, Ihara K, Blau CA, Drachman JG. Congenital amegakaryocytic thrombocytopenia in three siblings: molecular analysis of atypical clinical presentation. Exp Hematol. 2005;33(10):1215-1221. 27. Varghese LN, Zhang JG, Young SN, et al. Functional characterization of c-Mpl ectodomain mutations that underlie congenital amegakaryocytic thrombocytopenia. Growth Factors. 2014;32(1):18-26. 28. Jalas C, Anderson SL, Laufer T, et al. A founder mutation in the MPL gene causes congenital amegakaryocytic thrombocytopenia (CAMT) in the Ashkenazi Jewish population. Blood Cells Mol Dis. 2011;47(1):79-83. 29. Alter BP. Diagnosis, genetics, and management of inherited bone marrow failure syndromes. Hematology Am Soc Hematol Educ Program. 2007;2007:29-39. 30. Steele M, Hitzler J, Doyle JJ, et al. Reduced intensity hematopoietic stem-cell transplantation across human leukocyte antigen barriers in a patient with congenital amegakaryocytic thrombocytopenia and monosomy 7. Pediatr Blood Cancer. 2005;45(2):212-216. 31. Tuller T, Waldman YY, Kupiec M, Ruppin E. Translation efficiency is determined by both codon bias and folding energy. Proc Natl Acad Sci U S A. 2010;107(8):3645-3650. 32. Dokal I, Vulliamy T. Inherited aplastic anaemias/bone marrow failure syndromes. Blood Rev. 2008;22(3):141-153. 33. Weinzierl EP, Arber DA. The differential diagnosis and bone marrow evaluation of new-onset pancytopenia. Am J Clin Pathol. 2013;139(1):9-29. 34. Rivers A, Slayton WB. Congenital cytopenias and bone marrow failure syndromes. Semin Perinatol. 2009;33(1):20-28. 35. Geddis AE. Inherited thrombocytopenia: Congenital amegakaryocytic thrombocytopenia and thrombocytopenia with absent radii. Semin Hematol. 2006;43(3):196-203. 36. Sola-Visner M. Platelets in the neonatal period: developmental differences in platelet production, function, and hemostasis and the potential impact of therapies. Hematology Am Soc Hematol Educ Program. 2012;2012:506-511. 37. Lorenz V, Ramsey H, Liu ZJ, et al. Developmental stage-specific manifestations of absent TPO/c-MPL signalling in newborn mice. Thromb Haemost. 2017;117(12):23222333. 38. Baker-Groberg SM, Lattimore S, Recht M, McCarty OJ, Haley KM. Assessment of neonatal platelet adhesion, activation, and aggregation. J Thromb Haemost. 2016;14(4): 815-827. 39. Bunting S, Widmer R, Lipari T, et al. Normal platelets and megakayocytes are produced in vivo in the absence of thrombopoietin. Blood. 1997;90(9):3423-3429. 40. Ballmaier M, Germeshausen M, Krukemeier S, Welte K. Thrombopoietin is essential for the maintenance of normal hematopoiesis in humans: development of aplastic anemia in patients with congenital amegakaryocytic thrombocytopenia. Ann N Y Acad Sci. 2003;996:17-25. 41. Steinberg O, Gilad G, Dgany O, et al. Congenital amegakaryocytic thrombocytopenia-3 novel c-MPL mutations and their phenotypic correlations. J Pediatr Hematol Oncol. 2007;29(12):822-825.

42. Dame C, Wolber EM, Freitag P, Hofmann D, Bartmann P, Fandrey J. Thrombopoietin gene expression in the developing human central nervous system. Brain Res Dev Brain Res. 2003;143(2):217-223. 43. Ehrenreich H, Hasselblatt M, Knerlich F, et al. A hematopoietic growth factor, thrombopoietin, has a proapoptotic role in the brain. Proc Natl Acad Sci U S A. 2005;102(3):862-867. 44. Ivanova A, Wuerfel J, Zhang J, Hoffmann O, Ballmaier M, Dame C. Expression pattern of the thrombopoietin receptor (Mpl) in the murine central nervous system. BMC Dev Biol. 2010;10:77. 45. Hoffmann O, Rung O, Im AR, et al. Thrombopoietin Contributes to Neuronal Damage in Experimental Bacterial Meningitis. Infect Immun. 2011;79(2):928936 46. Castro Conde JR, Martinez ED, Rodriguez RC, Rodriguez De Hoyos AL. CNS siderosis and dandy-walker variant after neonatal alloimmune thrombocytopenia. Pediatr Neurol. 2005;32(5):346-349. 47. Goto T, Kakita H, Takasu M, et al. A rare case of fetal extensive intracranial hemorrhage and whole-cerebral hypoplasia due to latent maternal vitamin K deficiency. J Neonatal Perinatal Med. 2018;11(2):191194. 48. Marszal E, Jamroz E, Pilch J, Kluczewska E, Jablecka-Deja H, Krawczyk R. Agenesis of corpus callosum: clinical description and etiology. J Child Neurol. 2000;15(6):401-405. 49. O'Keefe M, Kafil-Hussain N, Flitcroft I, Lanigan B. Ocular significance of intraventricular haemorrhage in premature infants. Br J Ophthalmol. 2001;85(3):357-359. 50. Tsilou ET, Giri N, Weinstein S, Mueller C, Savage SA, Alter BP. Ocular and orbital manifestations of the inherited bone marrow failure syndromes: Fanconi anemia and dyskeratosis congenita. Ophthalmology. 2010;117 (3):615-622. 51. Yıldırım AT, Günes BT, Oymak Y, et al. Congenital amegakaryocytic thrombocytopenia: three case reports from patients with different clinical diagnoses and somatic abnormalities. Blood Coagul Fibrinolysis. 2015;26(3):337-341. 52. Zlotogora J. What is the birth defect risk associated with consanguineous marriages? Am J Med Genet. 2002;109(1):70-71. 53. Maserati E, Panarello C, Morerio C, et al. Clonal chromosome anomalies and propensity to myeloid malignancies in congenital amegakaryocytic thrombocytopenia (OMIM 604498). Haematologica. 2008;93(8):12711273. 54. Matatall KA, Jeong M, Chen S, et al. Chronic infection depletes hematopoietic stem cells through stress-induced terminal differentiation. Cell Rep. 2016;17(10):2584-2595. 55. Walne AJ, Dokal A, Plagnol V, et al. Exome sequencing identifies MPL as a causative gene in familial aplastic anemia. Haematologica. 2012;97(4):524-528. 56. Savoia A, Dufour C, Locatelli F, et al. Congenital amegakaryocytic thrombocytopenia: clinical and biological consequences of five novel mutations. Haematologica. 2007;92(9):1186-1193. 57. Kohler S, Carmody L, Vasilevsky N, et al. Expansion of the human phenotype ontology (HPO) knowledge base and resources. Nucleic Acids Res. 2019;47(D1):D1018D1027.

haematologica | 2021; 106(9)

ARTICLE

Platelet Biology & its Disorders

Efficacy, safety and immunological profile of combining rituximab with belimumab for adults with persistent or chronic immune thrombocytopenia: results from a prospective phase IIb trial Matthieu Mahévas,1,2,3 Imane Azzaoui,1,3* Etienne Crickx,1,2,3* Florence Canoui-Poitrine,4 Delphine Gobert,5 Laetitia Languille,1 Nicolas Limal,1 Constance Guillaud,1 Laure Croisille,6 Mohamed Jeljeli,7 Fréderic Batteux,7 Samia Baloul,4 Olivier Fain,5 France Pirenne,3 Jean-Claude Weill,2 Claude-Agnès Reynaud,2 Bertrand Godeau1,3 and Marc Michel1,3

Ferrata Storti Foundation

Haematologica 2021 Volume 106(9):2449-2457

*IA and EC contributed equally as co-second authors 1

Service de Médecine Interne, Centre National de Référence des Cytopénies AutoImmunes de l’Adulte, Centre Hospitalier Universitaire Henri-Mondor, Assistance Publique-Hôpitaux de Paris, Université Paris Est Créteil (UPEC), Créteil; 2Institut Necker Enfants Malades, INSERM U1151 CNRS UMS 8253, Université Paris Descartes, Sorbonne Paris Cité, Paris; 3Equipe n°2 "Transfusion et Maladies du Globule Rouge", EFS Île-de-France, IMRB U955 INSERM, Hôpital Henri-Mondor, AP-HP, Créteil; 4CEpiA (Clinical Epidemiology and Ageing), EA 7376-IMRB, Université Paris Est Créteil (UPEC), Hôpital Henri-Mondor, AP-HP Department of Public Health, Clinical Research Unit (URCMondor), Créteil; 5Sorbonne Université, Service de Médecine Interne, Hôpital SaintAntoine, Assistance Publique-Hôpitaux de Paris, Paris; 6Etablissement Français du Sang, Service d'Immunologie Plaquettaire, Hôpital Henri-Mondor, Créteil and 7Service d'Immunologie Biologique, Hôpital Cochin, AP-HP, INSERM U1016, Institut Cochin, Paris, France

ABSTRACT

-cell activating factor may be involved in the failure of B-cell depleting therapy with rituximab in immune thrombocytopenia (ITP) by promoting the emergence of splenic long-lived plasma cells. From results obtained in mouse models, we hypothesized that combining rituximab with sequential injections of belimumab could increase the rate of response at 1 year in patients with persistent or chronic ITP by preventing the emergence of these long-lived plasma cells. The study was a single-center, single-arm, prospective phase IIb trial investigating the safety and efficacy of rituximab given at a fixed dose of 1,000 mg, 2 weeks apart, combined with five infusions of belimumab, 10 mg/kg at week 0 (W0)+2 days, W2+2 days, W4, W8 and W12 for adults with primary persistent or chronic ITP. The primary endpoint was the total number of patients achieving an overall response (complete response + response) at W52 according to a standard definition. In total, 15 non-splenectomized adults, nine (60%) with persistent IPT and six (40%) with chronic ITP, were included. No severe adverse event, infection, or severe hypogammaglobulinemia was observed. Thirteen patients achieved an initial overall response. At W52, 12 (80%) patients achieved an overall response, including ten (66.7%) with complete response. When compared with a cohort of patients receiving rituximab alone, the kinetics of B-cell repopulation appeared similar, but the number of circulating T-follicular helper cells was significantly decreased with belimumab combination therapy. Combining rituximab and belimumab seems a promising strategy in ITP, with high efficacy and acceptable safety (clinicaltrials gov. Identifier: NCT03154385).

haematologica | 2021; 106(9)

Correspondence: MATTHIEU MAHÉVAS matthieu.mahevas@aphp.fr Received: May 18, 2020. Accepted: August 6, 2020. Pre-published: August 13, 2020. https://doi.org/10.3324/haematol.2020.259481

2449

M. Mahevas et al.

Introduction Primary immune thrombocytopenia (ITP) is a bleeding disorder mainly mediated by autoreactive B cells and plasma cells (PC) secreting pathogenic anti-platelet autoantibodies, eventually leading to accelerated platelet destruction and impaired megakaryopoiesis.1,2 First-line treatments include steroids and intravenous immunoglobulins (IVIg). Because less than 40% of newly diagnosed ITP adults will achieve a spontaneous remission within 12 months after disease onset, second-line treatments are frequently needed.3 Over the past 20 years, the anti-CD20 monoclonal antibody rituximab (RTX) has been considered an off-label second-line option in many countries and most guidelines. RTX leads to an overall response rate of 40% at 1 year.4,5 Whereas an almost complete B-cell depletion is achieved in peripheral blood and in secondary lymphoid organs after RTX in ITP,6 approximately half of the patients do not respond to RTX, which raises many questions and has led to some investigations in the past years. In ITP, pathogenic antibody-secreting PC are constantly generated in the spleen, mainly through the germinal center pathway.6,7 Because most of these splenic PC are shortlived and have lost CD20 expression, the clinical improvement observed after RTX is thought to result mainly from germinal center depletion, thus limiting PC generation.8,9 However, analysis of spleen samples from ITP patients with failure of RTX revealed that despite complete peripheral B-cell depletion, residual splenic PC secreting antiplatelet antibody persisted.6 More surprisingly, transcriptomic analysis showed that these splenic PC had acquired a long-lived program, similar to bona fide bone-marrow long-lived PC. Quantitatively, the data suggested that B-cell depletion had induced the differentiation of shortlived PC into long-lived ones, rather than the selection of pre-existing long-lived PC, thus providing clues for explaining RTX failure in the context of ITP.6 By using a fate mapping mouse model, we recently demonstrated that B-cell activating factor (BAFF) played a major role in the emergence of these splenic long-lived PC.10 BAFF is a pro-survival key cytokine for the B-cell lineage,11 and elevated levels of unconsumed BAFF are observed in serum and spleen after RTX therapy in ITP patients.6 Combining anti-CD20 with four infusions of anti-BAFF antibodies in this mouse model significantly reduced the number of splenic PC, with little impact on bone marrow PC.10 Hence, we hypothesized that combining two fixed doses of 1,000 mg of RTX with five sequential injections of belimumab (Benlysta®, 10 mg/kg dose) could increase the rate of response at 1 year in patients with persistent or chronic ITP by preventing the emergence of autoreactive splenic long-lived PC. Here, we report the efficacy and safety of this new strategy in ITP during a prospective phase IIb pilot trial.

Methods

fixed dose of 1,000 mg, 2 weeks apart, combined with five intravenous infusions of belimumab (Benlysta®, 10 mg/kg) at week 0 (W0) + 2 days, W2 + 2 days, W4, W8 and W12 (see Online Supplementary Figure S1). The rationale for the choice of this schedule was based on previous results obtained in a mouse model showing that BAFF inhibition should start early.10 The belimumab dosing was similar to the one approved in systemic lupus erythematosus. Research was conducted in accordance with the Declaration of Helsinki and was approved by the Comité de Protection des Personnes Ile-de-France VI.

Inclusion and exclusion criteria Inclusion and exclusion criteria are reported the Online Supplementary Appendix.

Primary endpoint The primary endpoint was the total number of patients achieving an overall response (complete response [CR] + response [R]) at W52. CR was defined by platelet count >100x109/L and R by platelet count 30-100x109/L with at least a 2-fold increase from baseline according to international definitions.12 Patients who required any other treatment for ITP including rescue therapy more than 6 weeks after inclusion were considered non-responders regardless of the platelet count.

Secondary endpoints Secondary endpoints were the number of patients achieving an overall response (CR+R) initially and at W12, W24, W36, number of bleeding events, number of patients showing severe hypogammaglobulinemia (γ-globulin level <4 g/L at W24, W36, W52), duration of severe hypogammaglobulinemia, variation in γ-globulin subclass levels during the study, and number of severe infections requiring hospitalization during the study.

Adverse events Adverse effects were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) v5.0 .

Immunological analysis, antibody titer tests, free B-lymphocyte stimulator assay Phenotype of circulating T- and B-cell subpopulations was analyzed by flow cytometry at W0, W4, W12, W24, W36 and W52 for every included patient and in a prospective control cohort of ITP patients not included in this trial who received two infusions of fixed-dose RTX (1,000 mg) at W0 and W2 after premedication with 100 mg intravenous methylprednisolone as a standard of care for ITP (see Online Supplementary Methods, Online Supplementary Table S1, and Online Supplementary Figure S2 for gating strategy). Antibody titers for pneumococcal, tetanus, measles vaccine were measured by enzyme-linked immunosorbent assay (ELISA). Serum-free BAFF level was measured using the assay developed by Glaxo Smith Kline pharmaceuticals (see Online Supplementary Appendix).

Direct monoclonal antibody-specific immobilization of platelet antigen assay Anti-platelet antibodies on patient platelets were detected by using the monoclonal antibody-specific immobilization of platelet antigen (MAIPA) assay (ApDia, Turnhout, Belgium).

Study design and study drugs The study was a single-center, single-arm, prospective phase IIb trial (RITUX-PLUS, clinicaltrials gov. Identifier: NCT03154385) investigating the safety and efficacy of RTX at a

2450

Statistical analysis The statistical analysis was performed as described in the Online Supplementary Methods.

haematologica | 2021; 106(9)

Rituximab combined with belimumab for adult ITP

Results Baseline characteristics We included 15 patients (12 females) with median age 50 years (range, 20-70 years). All patients had previously received corticosteroids and/or IVIg (n=8) as first-line treatment, and ten had received a second-line treatment (Table 1). All but one had a previous transient response to corticosteroids. Within 1 month prior to inclusion, the median platelet count nadir was 16x109/L (range, 3-28 x109/L). All but one patient had cutaneous bleeding manifestations and four had mucosal bleeding. Six patients had positive anti-nuclear antibody titers >1/160 with no features of systemic lupus erythematosus. When receiving the first RTX infusion, the median duration of ITP was 11 months (range, 4-52 months). Nine (60%) patients had persistent ITP and six (40%) had chronic ITP.

Safety Overall, 31 adverse events were reported during the study (Table 2); five were infusion-related reactions during the first RTX administration (all grade I according to the CTCAE classification). No infusion-related reaction was reported with belimumab. All but one of 26 adverse events occurring during the study were grade I, and eight were possibly related to treatment (bronchitis, n=2; nasopharyngitis, n=3; arthralgia, n=1; candida vulvovaginitis, n=1; cystitis, n=1). One patient experienced grade II serum sickness with moderate arthralgia and rash one day after the second infusion of RTX. γ-globulin levels were systematically monitored during the study (Figure 1A to D). We observed no severe infection and no severe hypogammaglobulinemia (total serum immunoglobulin (Ig) <4 g/L or IgG <4.5 g/L). We observed a significant decrease in IgG and IgM titers (Figure 1; Online Supplementary Table S2) between baseline and W24 (0,98 g/L and 0,42 g/L decrease in median IgG and IgM titers, respectively). One patient experienced moderate hypogammaglobulinemia (total serum Ig titers 4.9 g/dL,

IgG 4.7 g/L) at W12, which was transient and recovered at W24. One patient had IgM titers <0.4 g/dl at W12 (IgM baseline 0.7 g/L) that did not recover at W52. IgA titers did not vary throughout the study.

Efficacy Thirteen (86.7%) patients achieved an initial overall response at W12, including nine (60%) with CR (Figure 2). Two patients had a response at W7 and W8, respectively; other patients achieved a response after W4. One nonresponder had bleeding symptoms at W4 and required thrombopoietin receptor agonists at W6. No other patient required ITP-directed therapy until W30. Among initial responders (R), one relapsed at W30, with moderate bleeding (cutaneous), and one eventually achieved CR at W36. At W52, the median platelet count among responders was 189x109/L (range, 69-416x109/L) and 12 of 15 (80%) patients achieved overall response (95% Confidence Interval [CI]: 52-96), including ten (66.7%) with CR (95% CI: 38-88). After a follow-up of 18 months, one patient with an initial CR eventually relapsed at 16 months (Online Supplemental Figure S3).

Vaccine response All patients had received vaccination with pneumococcal polysaccharide vaccine (Pneumovax-23®, n=2) or conjugate vaccine PCV13 (Prevenar 13®, n=13) at least 15 days before inclusion (range, 15-90 days). When considering a protective threshold ≥1 μg/mL for anti-pneumococcal antibodies,13 13 patients were protected for at least 11 of the 13 serotypes tested (1, 3, 4, 5, 6A, 6B, 7F, 9V, 14, 18C, 19A, 19F, 23F), and two patients (who had received Prevenar 13®) were protected for eight serotypes at baseline (Online Supplementary Table S3). At W52, eight (53%) patients had no change (n=6) or <2 serotypes (n=2) loss in protection among the 13 serotypes tested. Two patients who were protected for eight serotypes at baseline had lost two and four other serotypes, respectively, at W52. Finally, five patients (30%) had lost protection for a medi-

Table 1. Baseline characteristics of patients with immune thrombocytopenia receiving rituximab and belimumab.

Age/Sex

25/F 29/F 42/F 51/F 31/F 57/F 39/F 33/F 70/F 66/M 66/M 20/F 50/M 54/F 57/F

ITP duration (months)

Bleeding manifestations

Treatments received before inclusion

Nadir platelet count during the month before inclusion (x 109/L)

11 4 15 15 5 52 44 11 3 42 5 4 31 4 4

Cutaneous Cutaneous + mucosal No Cutaneous Cutaneous Cutaneous Cutaneous Cutaneous Cutaneous Cutaneous Cutaneous + mucosal Cutaneous Cutaneous + mucosal Cutaneous Cutaneous + mucosal

CST CST CST, IVIg, hydroxychloroquine CST, Dapsone CST, IVIg CST, IVIg CST, hydroxychloroquine CST, hydroxychloroquine CST, hydroxychloroquine CST, IVIg, dapsone, hydroxychloroquine CST, IVIg, dapsone CST, IVIg, romiplostim CST, IVIg CST, eltrombopag CST, IVIg, vinblastin, romiplostim, eltrombopag

16 3 3 28 16 27 18 18 18 9 15 15 6 7 17

CST: corticosteroids, IVIg: Intravenous immunoglobulin. F: female; M: male.

haematologica | 2021; 106(9)

2451

M. Mahevas et al.

Figure 1. Serum level of total γ-globulins and immunoglobulin isotypes (IgG, IgA, IgM) during the study of rituximab and belimumab combined. Serum level of total γ-globulins (A) and IgG (B), IgM (C), and IgA (D) were assessed by nephelometry at week 12 (W12), W24, W36, W52. Dotted line represents normal threshold for each isotype. *P<0.05, **P<0.01, ***P<0.001; ns: not significant.

an of seven serotypes (range, 4-8). Overall, at W52, 11 (73%) patients had protective titers for at least 11 serotypes. All patients had received vaccination with tetanus and measles at different times before enrollment. Fourteen patients (93%) had no significant change in antitetanus antibody titers between W0 and W52. There were also no significant changes in anti-measles antibody titers, which remained at protective level >200 UA/m (protective >16.5 UA/m) for all patients at W52.

Antiplatelet antibody testing Direct MAIPA was performed at inclusion in all but one patient and was positive in ten (71%) patients, including nine with glycoprotein IIb/IIIa (GpIIb/IIIa) specificity, and one with GpIb/IX specificity (Online Supplementary Table S4). Among these patients, seven (70%) achieved response and had negative MAIPA results at W52, two achieved response (one CR and one R) and still had antiGpIIb/IIIa antibodies at W52, and one did not respond but had negative MAIPA results at W52.

Immunological analysis In order to precisely assess the impact of blocking BAFF concomitantly with B-cell depletion on B- and T-cell subsets, we analyzed in parallel a prospective cohort of 12 ITP patients who received RTX without belimumab as a standard of care (Online Supplementary Table S5). As previously reported, we observed a significant increase in BAFF serum levels in patients receiving RTX alone at W12, W24, W36 and W52 (all P<0.001) after RTX, as compared to baseline. In the RITUX-PLUS study, belimumab treatment effectively reduced BAFF levels at W12 as compared to 2452

baseline (1,210±248 vs. 90±38 pg/mL, P<0.0001). BAFF levels started to return to baseline at W24 (730±294 pg/mL, P<0.01), then strongly increased at W36 (2,199±1,498 pg/mL, P<0.0001) and reached a plateau at W52 (2937±1561 pg/mL P<0.0001) (Figure 3A). BAFF levels at W36 and W52 did not significantly differ between patients receiving belimumab + RTX in the study and control patients receiving RTX alone. In order to evaluate the effect of the RTX and belimumab combination on B-cell depletion and re-appearance, we analyzed circulating CD19+ B cells in both cohorts at baseline, W4, W12, W24, W36, and W52 (Figure 3B). All patients showed complete depletion of circulating CD19+ B cells at W4 and W12 (Figure 3C). Reappearance of B cells in the peripheral blood varied among patients. We observed no significant delay in B-cell reconstitution in patients receiving belimumab plus RTX versus controls receiving RTX alone, despite a slight difference at W36. Transitional B cells (CD19+IgD+ CD24+CD38+CD10+), which are precursors of naïve B cells in peripheral blood,14,15 emerged early during B-cell reconstitution (Figure 3D to E). Indeed, two of 15 patients receiving belimumab showed transitional B cells in the peripheral blood at W24 as compared with five of 12 patients receiving RTX alone. All but one patient in both groups showed naïve and transitional B cells at W52. The absolute number of CD19+ cells remained significantly decreased at W52 in both groups as compared to baseline, and B-cell depletion mainly affected memory B cells and IgD+CD27+ B cells (Figure 3F and G). As previously reported, the number of circulating CD27highCD38high cells (mainly plasmablasts expressing haematologica | 2021; 106(9)

Rituximab combined with belimumab for adult ITP

Table 2. Adverse events reported in the study.

Patients

Infusion related reactions

Adverse events, CTCAE grade, imputability

1 2 3

Throat itching, grade I, (RTX W0) Sore throat, grade I, (RTX W0) No

Throat itching, grade I, (RTX W0)

7 8

Headache, grade I, (RTX W0) No

10 11 12

No No Abdominal discomfort grade I, (RTX W0)

14 15

No No

Arthralgia (W2- W8) grade I, possible No Transient hypereosinophilia < 1000 /mm3 (W40) treated with zentel/stromectol, grade 1, not related Gluteus medius tendinitis (W8-W12), grade I, not related Nasopharyngitis (W10), grade I possible Supraspinatus tendinitis (W24), grade I, not related Serum thickness (W2), grade I, related Candida vulvovaginitis (W8), grade I, not related Rhinorrhea (W2) and (W12), grade I, not related Bronchitis (W5), possible Pain extensor muscles of the forearm, (W24), grade I, not related Acute cystitis (W4), grade I, possible Arthralgia (W3), thoracic pain (W3), grade I, not related Viral conjunctivitis (W4), grade I, not related Arthralgia (W52), grade I, not related Rhinorrhea (W8), grade I, not related Pharyngitis (W4), grade I, possible Bronchitis (W50), grade I, possible Gout arthritis (W44), grade I, not related Increased number of monocytes after TPO-RA, grade I, not related Nasopharyngitis (W20), myalgia (W20), grade I, possible Knee pain (W52), grade I, not related Bronchitis (W9), grade I, possible Memory problems W14, grade I, not related Erectile dysfunction W36, grade I, not related Low back pain (W12), grade I, not related Bronchitis (W0-W2), grade I, possible

CTCAE: National Cancer Institute Common Terminology Criteria for Adverse Events; RTX: rituximab; W: week.

the Ki67 marker) was increased at baseline in ITP patients as compared with healthy donor controls (n=11, P<0.05, Online Supplementary Figure S4). A marked reduction of plasmablasts/PC was observed from W4 and lasted until W52, when the number of circulating plasmablasts was low and comparable to that in healthy donors (Figure 3H; Online Supplementary Figure S5). Because changes in T-cell homeostasis have been described with RTX, we investigated peripheral T-cell compartments before and after treatment. We observed no significant changes in the distribution of CCR7+CD45RA+ naive (TN), CCR7− CD45RA− memory (TEM), and CCR7+CD45RA− central memory (TCM) in CD4+ and CD8+ cells. The expression of CD38 and HLADR activation markers on CD4+ or CD8+ T cells was not modified with treatment. Finally, we observed no significant change in CD4 T-cell polarization (TH1/TH2/TH17) based on the expression of CXCR3, CXCR5, CCR6 (Online Supplementary Figure S6). By contrast, the activated subset of circulating follicular helper T cells (activated cTfh) identified as CD4+CD45RA−CXCR5+ inducible T-cell costimulator (ICOS)+ programmed death1 (PD1)+ cells16 (Figure 4A) was increased at baseline in ITP patients as compared with healthy controls (Online Supplementary Figure S4). While circulating Tfh percentage remained stable throughout the study, the percentage of activated cTfh cells was significantly decreased in patients receiving RTX plus belimumab but not RTX alone despite a trend at W4, and remained significantly decreased at W24 and W36 in patients receiving RTX plus belimumab versus RTX alone haematologica | 2021; 106(9)

(Figure 4B and C). There was no correlation between the response (initial response or at W52) and BAFF or cTfh (not shown).

Discussion The rationale of this study was based on three observations: i) the presence of anti-platelet long-lived PC in the spleen of patients who did not respond to RTX; ii) the increased BAFF level in the spleen and serum of these patients, and iii) the observation that combining an antiBAFF antibody with anti-CD20 treatment induced a major depletion of long-lived PC in a mouse model.6,9,10 The results of this prospective phase IIb pilot trial of ITP showed that combining RTX with five infusions of belimumab led to an overall response rate of 80% with a 66.7% CR rate at 1 year. Although the sample size was limited, these response rates were higher than expected with RTX alone (overall response rates of 40% to 50% at 1 year, with 30% of CR, according to most previous studies conducted in ITP).4,5,17,18 These response rates were also higher but closer to those obtained with RTX and dexamethasone19 or RTX associated with high-dose dexamethasone and ciclosporin.20 Hence, these results are promising and provide a real proof of concept for this new combination. Most consecutive patients included in this study were women with persistent ITP. Disease duration <12 months, young age and female sex have been found associated 2453

M. Mahevas et al.

with better outcomes,4,5,17,18 which could represent a bias although these factors were not associated with the overall long-term response in our large French prospective registry study.21 The combination strategy was well tolerated, with no severe adverse events and in particular no severe infection. Despite a significant decrease in IgG and IgM titers, we did not observe severe hypogammaglobulinemia. This finding contrasted with the results of the phase IIa study conducted in severe systemic lupus erythematosus, in which severe hypogammaglobulinemia developed in three of 16 patients, with IgG titers <4.5 g/L.22 However, these patients had previously received immunosuppressive therapies, which was not the case for our patients. No significant changes in IgG or IgM titers were observed in patients receiving two infusions of 1,000 mg of RTX or 375 mg/m2 once weekly for 4 weeks.5,23,24,25 Therefore, the slight decrease in IgG/IgM titers may reflect the impact on splenic PC, because anti-tetanus and anti-measles antibody titers, which are secreted by bone-marrow longlived PC, remained stable over time. The study was not specifically designed to assess the vaccine response; indeed, the timing of pneumococcal vaccination was heterogenous, and two patients received T-cell–independent vaccines. Half showed no decrease in serological protection for most serotypes with treatment, and only four patients had lost protection for more than seven serotypes. In the absence of a control cohort, it was not possible to measure the specific impact of combination therapy versus rituximab given alone. Vaccine-induced antibody titers against measles and tetanus toxoid were not reduced at 1 year, suggesting that bone marrow longlived PC were not affected by the combination therapy. Altogether, these results suggest that belimumab and RTX did not induce significant immunodepression. From an immunological perspective, achieving a sustained CR indicates that pathogenic PC were efficiently 2454

Figure 2. Efficacy of rituximab and belimumab combination in adults with persistent and chronic immune thrombocytopenia. (A) Outcome at week 4 (W4), W12, W24, W36, and W52 according to international recommendations. Complete response (CR) was defined by a platelet count >100x109/L and response (R) by a platelet count 30-100x109/L with at least a 2-fold increase from baseline. Non-responders (NR) are labeled in red; platelet counts were censured when an ITP-directed therapy was started. (B) Evolution of platelet count for each patient during the study. ITP: immune thrombocytopenia; NR: no response.

targeted. This is exemplified by the disappearance of platelet autoantibodies in all but one patient with an initial positive test. Despite the absence of a control cohort to clearly assess the impact of this combination on PC, our results support previous results obtained in mouse models showing that combination of RTX and belimumab inhibits the emergence of pathogenic splenic long-lived PC and anti-platelets antibodies.10 Of note, the addition of belimumab had no significant impact on residual circulating PC/plasmablasts, which were mainly expressing IgA and have been described as originating from resident mucosal B cells.26 The kinetics of B-cell repopulation seemed similar regardless of belimumab exposure. We observed a slight delay in the beginning of B-cell reconstitution (i.e., reappearance of naïve and transitional B cells), but at W52, all but one patient had detectable B cells in peripheral blood. As previously reported, the memory B-cell pool was profoundly depleted until W52 in both cohorts. Elevated BAFF levels may lower the stringency of the Bcell selection and allow for rescuing autoreactive cells.27 This hypothesis is supported by studies showing that negative selection of high-affinity DNA-reactive B cells was impaired by increased levels of BAFF during B-cell depletion in an auto-immune mouse model.28 This was the basis for studies conducted in systemic lupus erythematosus in which belimumab is maintained for more than 52 weeks after RTX. In the present study, the last belimumab infusion was administered at W12 and resulted in a complete blockade of BAFF at least until W24. Therefore, B-cell reconstitution occurred in a milieu with increased BAFF levels. Maintaining belimumab for 6 to 9 months after RTX may allow for dampening BAFF levels during Bcell reconstitution and improve the stringency of B-cell selection, thus limiting the risk of relapse after B-cell reconstitution. Our results also suggest that RTX followed by belimumhaematologica | 2021; 106(9)

Rituximab combined with belimumab for adult ITP

Figure 3. B-cell activating factor and B-cell subsets in immune thrombocytopenia patients receiving the combination therapy or rituximab alone. (A) B-cell activating factor (BAFF) concentrations were assessed by enzyme-linked immunosorbent assay in serum of patients receiving rituximab and belimumab (in blue) or rituximab alone (in red) at week 0 (W0), W12, W24, W36 and W52. Data are mean ± standard error of the mean (pg/mL). (B) Gating strategy of B-cell subpopulations. Single lymphoid cells on peripheral blood mononuclear cells (PBMC) were gated by using scatter parameters, and dead cells were eliminated by using zombie violet. Plasmablasts/plasma cells (PB/PC) were defined as CD27hiCD38hi cells among CD3-CD14-CD16- cells. After excluding CD3/CD14/CD16-positive cells and PB/PC from the CD19+ gate, B-cell subsets were separated according to their expression of CD27 and IgD and defined as memory B cells (CD27+IgD-), CD27+IgD+ B cells, and naïve B cells (CD27–IgD+). Transitional B cells were defined as CD38hiCD24hiCD10+ cells among naïve B cells. (C) Circulating B-cell subset count per million PBMC at W0, W4, W12, W24, W36 and W52. ****P<0.0001

ab had an unexpected effect on activated cTfh cells, which are essential for germinal center formation, B-cell affinity maturation and plasmablast generation.29 In humans, the majority of cTfh are central memory T cells expressing PD1 but no ICOS,16 but germinal center recruitment and support for B-cell differentiation requires up-regulation of ICOS.30 Splenic Tfh cells can contribute to ITP pathogeny and, as previously reported,7 the number of activated cTfh cells was increased at baseline in the peripheral blood of ITP patients. Although some cTfh cells were shown to express BAFF-R in systemic lupus erythematosus, this was not the case in ITP (data not shown),31 so their sensitivity to belimumab remains unexplained so far. Of note, local BAFF production by Tfh cells has been identified as an haematologica | 2021; 106(9)

important factor for promoting germinal center B-cell survival.32 Belimumab might also prevent, through such pleiotropic effects, the re-emergence of germinal centers in lymphoid organs at the time of B-cell repopulation, an effect that would be further strengthened by extending the duration of its administration. These preliminary results support a new rationale for the addition of BAFF blockade to B-cell depletion in auto-immune diseases. The main limitations of this pilot exploratory and single-center trial are the sample size and the open design. These results should be confirmed in a multi-center double-blind randomized prospective trial. In conclusion and despite these limitations, in adult ITP, adding belimumab to RTX at the initial phase of B-cell 2455

M. Mahevas et al.

Figure 4. Rituximab and belimumab combination affects activated circulating T- follicular helper cells. (A) Gating strategy for circulating T- follicular helper cells (cTfh) cells. After gating on CD4+CD45RA– memory T-CD4+ cells in whole blood, cTfh cells were defined as CXCR5+PD1+, and activated cTfh cells as CXCR5+PD1+ICOS+. (B) Percentages of cTfh and (C) activated Tfh cells at week 0 (W0), W4, W12, W24, W36 and W52 in patients receiving rituximab and belimumab or rituximab alone. *P<0.05, **P<0.01.

depletion seems to be a promising strategy with high efficacy and acceptable safety. Disclosures MMa received funds for research from GSK, and received fees from LFB; BG served as an expert for Amgen, Novartis, LFB and Roche: he received funds for research from Amgen and Roche; MMi received consultancy fees from Amgen, Novartis and Argenx; DG received consultancy fees from Novartis and Shire Takeda. Contributions MMa designed the study and initiated this work; MMa, BG, MMi, IA and EC wrote the report; all authors made substantial contributions to acquisition of data, revised the article critically and gave final approval of the manuscript to be submitted. 2456

Acknowledgments The authors thank Laura Smales for editorial assistance; Roxane Kaponou Johnson and Alexis Vandenberghe for technical assistance. E. Crickx was supported by a Poste d’accueil Inserm. Funding This study was initiated by the investigators and partially financed with an open grant from GSK, which played no role in designing the study, collecting and analyzing the data, or writing the article. The study was funded by a grant from Programme Hospitalier de Recherche Clinique - PRTS 2013 (Ministère de la Santé). The sponsor was Assistance Publique – Hôpitaux de Paris (Département de la Recherche Clinique et du Développement).

haematologica | 2021; 106(9)

Rituximab combined with belimumab for adult ITP

References 1. Cines DB, Cuker A, Semple JW. Pathogenesis of immune thrombocytopenia. Presse Médicale Paris Fr. 1983 2014;43(4 Pt 2):e49-59. 2. Audia S, Mahévas M, Samson M, Godeau B, Bonnotte B. Pathogenesis of immune thrombocytopenia. Autoimmun Rev. 2017;16(6):620-632. 3. Moulis G, Germain J, Comont T, et al. Newly diagnosed immune thrombocytopenia adults: Clinical epidemiology, exposure to treatments, and evolution. Results of the CARMEN multicenter prospective cohort. Am J Hematol. 2017;92(6):493-500. 4. Khellaf M, Charles-Nelson A, Fain O, et al. Safety and efficacy of rituximab in adult immune thrombocytopenia: results from a prospective registry including 248 patients. Blood. 2014;124(22):3228-3236. 5. Chugh S, Darvish-Kazem S, Lim W, et al. Rituximab plus standard of care for treatment of primary immune thrombocytopenia: a systematic review and meta-analysis. Lancet Haematol. 2015;2(2):e75-81. 6. Mahévas M, Patin P, Huetz F, et al. B cell depletion in immune thrombocytopenia reveals splenic long-lived plasma cells. J Clin Invest. 2013;123(1):432-442. 7. Audia S, Rossato M, Santegoets K, et al. Splenic TFH expansion participates in B-cell differentiation and antiplatelet-antibody production during immune thrombocytopenia. Blood. 2014;124(18):2858-2866. 8. Mahévas M, Michel M, Vingert B, et al. Emergence of long-lived autoreactive plasma cells in the spleen of primary warm auto-immune hemolytic anemia patients treated with rituximab. J Autoimmun. 2015;62:22-30. 9. Mahévas M, Michel M, Weill J-C, Reynaud C-A. Long-lived plasma cells in autoimmunity: lessons from B-cell depleting therapy. Front Immunol. 2013;4:494. 10. Thai L-H, Le Gallou S, Robbins A, et al. BAFF and CD4+T cells are major survival factors for long-lived splenic plasma cells in a B-cell-depletion context. Blood. 2018; 131(14):1545-1555. 11. Mackay F, Schneider P. Cracking the BAFF code. Nat Rev Immunol. 2009;9(7):491-502.

haematologica | 2021; 106(9)

12. Rodeghiero F, Stasi R, Gernsheimer T, et al. Standardization of terminology, definitions and outcome criteria in immune thrombocytopenic purpura of adults and children: report from an international working group. Blood. 2009;113(11):2386-2393. 13. Grabar S, Groh M, Bahuaud M, et al. Pneumococcal vaccination in patients with systemic lupus erythematosus: A multicenter placebo-controlled randomized double-blind study. Vaccine. 2017;35(37): 4877-4885. 14. Anolik JH, Friedberg JW, Zheng B, et al. B cell reconstitution after rituximab treatment of lymphoma recapitulates B cell ontogeny. Clin Immunol Orlando Fla. 2007;122(2):139145. 15. Sanz I, Wei C, Jenks SA, et al. Challenges and opportunities for consistent classification of human B cell and plasma cell populations. Front Immunol. 2019; 10:2458. 16. Locci M, Havenar-Daughton C, Landais E, et al. Human circulating PD-1+CXCR3CXCR5+ memory Tfh cells are highly functional and correlate with broadly neutralizing HIV antibody responses. Immunity. 2013;39(4):758-769. 17. Lucchini E, Zaja F, Bussel J. Rituximab in the treatment of immune thrombocytopenia: what is the role of this agent in 2019? Haematologica. 2019;104(6):1124-1135. 18. Zaja F, Volpetti S, Chiozzotto M, et al. Longterm follow-up analysis after rituximab salvage therapy in adult patients with immune thrombocytopenia. Am J Hematol. 2012;87 (9):886-889. 19. Bussel JB, Lee CS, Seery C, et al. Rituximab and three dexamethasone cycles provide responses similar to splenectomy in women and those with immune thrombocytopenia of less than two years duration. Haematologica. 2014;99(7):1264-1271. 20. Choi PY-I, Roncolato F, Badoux X, Ramanathan S, Ho S-J, Chong BH. A novel triple therapy for ITP using high-dose dexamethasone, low-dose rituximab, and cyclosporine (TT4). Blood. 2015; 126(4):500503. 21. Deshayes S, Khellaf M, Zarour A, et al. Long-term safety and efficacy of rituximab in 248 adults with immune thrombocytopenia: results at 5 years from the French prospective registry ITP-ritux. Am J

Hematol. 2019;94(12):1314-1324. 22. Kraaij T, Kamerling SWA, de Rooij ENM, et al. The NET-effect of combining rituximab with belimumab in severe systemic lupus erythematosus. J Autoimmun. 2018;91:4554. 23. Stasi R, Pagano A, Stipa E, Amadori S. Rituximab chimeric anti-CD20 monoclonal antibody treatment for adults with chronic idiopathic thrombocytopenic purpura. Blood. 2001;98(4):952-957. 24. Cooper N, Stasi R, Cunningham-Rundles S, et al. The efficacy and safety of B-cell depletion with anti-CD20 monoclonal antibody in adults with chronic immune thrombocytopenic purpura. Br J Haematol. 2004;125(2): 232-239. 25. Arnold DM, Heddle NM, Carruthers J, et al. A pilot randomized trial of adjuvant rituximab or placebo for nonsplenectomized patients with immune thrombocytopenia. Blood. 2012;119(6):1356-1362. 26. Mei HE, Frölich D, Giesecke C, et al. Steadystate generation of mucosal IgA+ plasmablasts is not abrogated by B-cell depletion therapy with rituximab. Blood. 2010;116(24):5181-5190. 27. Liu Z, Davidson A. BAFF and selection of autoreactive B cells. Trends Immunol. 2011;32(8):388-394. 28. Boneparth A, Woods M, Huang W, Akerman M, Lesser M, Davidson A. The effect of BAFF inhibition on autoreactive B cell selection in murine SLE. Mol Med. 2016;22:173-182. 29. Ueno H, Banchereau J, Vinuesa CG. Pathophysiology of T follicular helper cells in humans and mice. Nat Immunol. 2015; 16(2):142-152. 30. Heit A, Schmitz F, Gerdts S, et al. Vaccination establishes clonal relatives of germinal center T cells in the blood of humans. J Exp Med. 2017;214(7):2139-2152. 31. Coquery CM, Loo WM, Wade NS, et al. BAFF regulates follicular helper t cells and affects their accumulation and interferon-γ production in autoimmunity. Arthritis Rheumatol. 2015;67(3):773-784. 32. Goenka R, Matthews AH, Zhang B, et al. Local BLyS production by T follicular cells mediates retention of high affinity B cells during affinity maturation. J Exp Med. 2014;211(1):45-56.

2457

ARTICLE Ferrata Storti Foundation

Red Cell Biology & its Disorders

Effect of HBB genotype on survival in a cohort of transfusion-dependent thalassemia patients in Cyprus Petros Kountouris,1,2 Kyriaki Michailidou,1,2 Soteroula Christou,3 Michael Hadjigavriel,4 Maria Sitarou,5 Anita Kolnagou,6 Marina Kleanthous1,2 and Paul Telfer7 1

Haematologica 2021 Volume 106(9):2458-2468

The Cyprus Institute of Neurology and Genetics, Nicosia, Cyprus; 2Cyprus School of Molecular Medicine, Nicosia, Cyprus; 3Thalassemia Center, Archbishop Makarios Hospital, Nicosia, Cyprus; 4Thalassemia Center, Limassol General Hospital, Limassol, Cyprus; 5Thalassemia Center, Larnaca General Hospital, Larnaca, Cyprus; 6Thalassemia Center, Paphos General Hospital, Paphos, Cyprus and 7Center for Genomics and Child Health, Blizard Institute, Queen Mary University of London, London, UK

ABSTRACT

Correspondence: PETROS KOUNTOURIS petrosk@cing.ac.cy Received: May 25, 2020. Accepted: July 21, 2020. Pre-published: July 30, 2020.

nitiation of regular transfusion in transfusion-dependent thalassemia (TDT) is based on the assessment of clinical phenotype. Pathogenic HBB variants causing β-thalassemia are important determinants of phenotype and could be used to aid decision-making. We investigated the association of HBB genotype with survival in a cohort study in the four thalassemia centers in Cyprus. HBB genotype was classified as severe (β0/β0 or β+/β0), moderate (β+/β+), or mild (β0/β++ or β+/β++). Risk factors for mortality were evaluated using multivariate Cox proportional-hazards regression. Of the 537 subjects who were followed for a total of 20,963 person-years, 80.4% (95% confidence interval [95% CI]: 76.484.7) survived to 50 years of age with increasing rates of liver-, infectionand malignancy-related deaths observed during recent follow-up. We evaluated non-modifiable risk factors and found worse outcomes associated with male sex (hazard ratio 1.9, 95% CI: 1.1-3.0, P=0.01) and milder genotype (hazard ratio 1.6, 95% CI: 1.1-2.3, P=0.02). The effect of genotype was confirmed in a second model, which included treatment effects. Patients with a milder genotype initiated transfusion significantly later and had reduced blood requirements compared to those with moderate or severe genotypes, although pre-transfusion hemoglobin levels did not differ between genotypes. Our results suggest that early treatment decisions to delay transfusion and different long-term treatment strategies in individuals with milder genotypes have led to adverse longterm effects of under-treated thalassemia and worse survival. We propose that HBB genotype determination and use of this information to aid in decision-making can improve long-term outcomes of thalassemia patients.

https://doi.org/10.3324/haematol.2020.260224

2458

Standard care of transfusion-dependent thalassemia (TDT) has improved dramatically over the past five decades with the introduction of regular transfusion, a variety of iron-chelating drugs, improved evaluation of transfusion iron overload with magnetic resonance imaging modalities, and the development of specialist centers and regional networks.1-7 There has been a consensus on standard care disseminated in national and international guidelines, which are periodically updated,9-10 resulting in progressive improvement in survival in successive birth cohorts.2,5-11 One aspect of care that is not standardized concerns the decision to start regular transfusion. Individuals at the most severe end of the phenotypic spectrum require regular transfusion before the age of 2 years for survival. At the other end of the spectrum are non-TDT patients who do not require regular transfusion to survive, are less anemic, and have less bone marrow expansion. Clinicians have traditionally avoided

haematologica | 2021; 106(9)

HBB genotype in TDT in Cyprus

regular transfusion in these cases because of concerns about the consequences of long-term transfusion, notably transfusion iron overload. The result of not correcting the thalassemia phenotype in earlier life includes a variety of chronic complications which can become apparent later. These include liver disease, pulmonary hypertension, endocrine dysfunction, bone disease and splenomegaly leading to splenectomy. Quality of life is significantly impaired and life expectancy may be reduced.12-15 In the middle of the phenotypic spectrum are TDT patients who are not absolutely transfusion-dependent early in their life, and regular transfusion may be delayed in the age range 2– 10 years,16,17 during which interval they may be at risk of developing complications of untreated thalassemia. Over 400 HBB variants causing β-thalassemia have been reported to date18 and are generally classified as β0, in which no functional β-globin chains can be produced; β+, in which β-globin chain production is severely reduced; and β++, in which β-globin chain production is mildly reduced.19 A previous study evaluated the influence of HBB pathogenic variants and other genetic modifiers on time to initiate regular transfusion, and developed a predictive model derived from a composite score. HBB genotype was found to be the most powerful predictor of severity.16,17 In routine practice, genetic diagnosis is often not available to the clinician and is not currently used for clinical decision-making. Further evidence of different outcomes associated with different genotypes is needed to recommend routine determination of genotype at birth and to make use of genotypic information in clinical decision-making about the initiation of transfusion and changing other aspects of standard practice. Cyprus is a Mediterranean island with a relatively stable indigenous population at high risk of thalassemia. Following introduction of a national carrier screening program in 1974, all new diagnoses of infants on the government-controlled southern part of the island have been documented and standard care is given in four treatment centers,20 based on shared protocols for transfusion and chelation developed over the past 30 years. The decision on the start of transfusion has been based purely on clinical observation according to best standard practice. The patients have remained stable within each clinic with very little immigration or emigration. The Cyprus thalassemia cohort consists of TDT patients born between 1960 and 2000, and their follow-up to 2004 was published previously.1 Herein, we present updated information with longer follow-up from this cohort, in which we generate a more robust estimate of survival of thalassemia with standard treatment, we confirm changing causes of mortality, and we explore the association of genotype with long-term outcomes and survival.

lassemia screening and prevention program,20 which resulted in identification soon after birth of all affected children from 1974 onwards, year of birth was categorized as pre-1974 or 1974 onwards. Based on commonly agreed categories of severity of pathogenic variants,21 HBB genotype was classified as severe, moderate and mild, as shown in Table 1, assuming equally spaced severity across categories. Chelation treatment was categorized based on the predominant chelator used during follow-up from year 2000 onwards: (i) deferoxamine only, (ii) deferiprone alone or in combination with deferoxamine, and (iii) deferasirox. Deferoxamine was given by subcutaneous or intravenous infusion, whereas deferiprone and deferasirox were administered orally. For splenectomy, the effect on survival was assumed to be dependent on whether and when it had been performed, and was categorized as (i) before 16 years of age, and (ii) after 16 years of age or not done. To evaluate genotype associations with the transfusion regimens, we used data previously collected for 336 patients from the cohort, as part of the THALAMOSS project.22 These data comprised the age of the first regular transfusion, and annual blood usage and pre-transfusion hemoglobin levels, between 2014 and 2016.

Methods

Results

Patients and data preparation

A summary of the clinical characteristics of the 537 cohort patients are documented in Table 2. There were two fewer patients than in the original study. Both of these patients were found to have been ineligible, one because of birth prior to 1960, and the other with a diagnosis of Hb H disease. Eight patients had been treated with sibling allogeneic stem cell transplant, at a mean age of 12.7 years. Follow-up on these eight patients was censored at the date of transplantation. The total number of patient-years of follow up was

This study is a follow-up of a previously described cohort of patients.1 Briefly, we included all patients who were transfusiondependent at the time of the first study, who were born in Cyprus between 1960 and 2000, and who were treated using standard care in the four thalassemia centers. Data were collected and validated for each treatment center up to September 30, 2018. Causes of death were classified as cardiac, liver (including hepatocellular carcinoma), malignancy, infection and other, using standard clinical criteria. Considering the introduction of the Cyprus tha-

haematologica | 2021; 106(9)

Statistical analysis A univariate analysis was performed to estimate the overall survival and the association of different risk factors with mortality using Kaplan-Meier analysis. The log-rank test was used to evaluate the statistical significance of the effect of the risk factors on survival. For the multivariate analysis, we used Cox proportionalhazards regression with a backwards approach in which all potential factors that were significant in the univariate analysis were included in the initial regression model and then sequentially removed to obtain the best-fit model. All models were tested for whether they met the proportional hazards assumption for a Cox proportional-hazards model. Proportions were compared in contingency tables using the Pearson χ2 test. Mortality rates and confidence intervals were calculated overall and for the periods 1980-89, 1990-99, 2000-09, and 2010-18. The Kruskal-Wallis test and the Wilcoxon rank-sum test for post-hoc analyses were utilized to study the association between HBB genotype and transfusion data. All statistical analyses were performed using R (version 3.6.1) and R packages survival and survminer.

Ethical considerations The study was approved by the Cyprus National Bioethics Committee and was initiated by PT as part of a project commissioned by the Cyprus Ministry of Health. Subsequently, informed consent was obtained for clinical follow-up of the cohort as part of the THALAMOSS project (EU FP7 grant agreement 306201), which included collection of transfusion data.

2459

P. Kountouris et al.

Table 1. Pathogenic HBB variations in the cohort of thalassemia-dependent thalassemia patients.

Allele 1 IVS I-110 G>A IVS I-110 G>A IVS I-1 G>A IVS I-1 G>A IVS I-1 G>A IVS I-110 G>A IVS I-110 G>A IVS I-110 G>A IVS II-745 C>G IVS I-110 G>A IVS I-110 G>A IVS I-110 G>A IVS II-745 C>G IVS I-110 G>A IVS II-745 C>G IVS I-110 G>A IVS I-110 G>A IVS I-1 G>A IVS I-6 T>C CD 39 CAG>TAG Total

Allele 2 IVS I-1 G>A CD 39 CAG>TAG IVS II-745 C>G IVS I-1 G>A CD 44 -C IVS II-1 G>A CD 44 -C CD 5 -CT CD 39 CAG>TAG IVS I-110 G>A IVS II-745 C>G IVS II-848 C>A IVS II-745 C>G IVS I-6 T>C IVS I-6 T>C Hb Knossos -87 C>G IVS I-6 T>C CD 39 CAG>TAG Hb Lepore Boston-Washington

Allele phenotypes +

β /β β+/ β0 β0/ β+ β0/ β0 β0 β0 β+/β0 β+/ β0 β+/ β0 β+/ β0 β+/ β+ β+/β+ β+/ β+ β+/β+ β+/ β++ β+/ β++ β+/ β++ β+/ β++ β0/ β++ β++/ β0 β0/ β++

20,963. The majority of the patients were born between 1960 and 1979 (453 patients, 84.4%). Splenectomy status was significantly associated with clinic (P=0.02), with 21.8% of patients splenectomized during childhood in Nicosia, compared to 36.7%, 29.3% and 34.9% in, respectively, Larnaca, Limassol and Paphos, while there was no significant difference in choice of iron chelation therapy between clinics. Thirty-eight (7.1%) patients had evidence of hepatitis C infection (HCV antibody positive), of whom 9 (1.7%) were HCV RNA positive at the end of the study and 29 (7.3%) were HCV RNA negative. Three patients (0.6%) were positive for hepatitis B surface antigen. A summary of the genotypes of the study patients is shown in Table 1. HBB genotype data were not available for 24 patients, who had died before genotyping was introduced in Cyprus. An additional patient was removed from the genotype analysis because thalassemia was caused by α-locus duplications co-inherited with β-thalassemia trait.23 Homozygous IVS I-110 G>A (β+) was the commonest genotype (59.8%) and there was a large proportion of patients with a combination of IVS I-110 G>A (β+) and IVS I-6 T>C (β++). There was a significant trend on performing splenectomy in childhood with decreasing severity of genotype, specifically 19% in the severe genotype group, 27.8% in the moderate genotype group, and 38.1% in the mild genotype group (P=0.02). There was also a significant trend across the genotypic spectrum in chelation therapy, with increasing proportions of patients with a milder phenotype remaining on deferoxamine rather than switching to oral chelation, specifically 14.3% with severe genotype, 17.5% with moderate genotype; and 28.9% with mild genotype (P=0.02). Genotyping for the α-globin locus and the XmnI C/T polymorphism was almost complete for the study population (92.4% and 89.4%, respectively). The XmnI T allele, which is associated with increased Hb F production, 2460

Genotype category

Number (%)

Severe Severe Severe Severe Severe Severe Severe Severe Severe Moderate Moderate Moderate Moderate Mild Mild Mild Mild Mild Mild Mild

48 (9.4) 23 (4.5) 5 (1.0) 2 (0.4) 2 (0.4) 2 (0.4) 1 (0.2) 1 (0.2) 1 (0.2) 306 (59.8) 29 (5.7) 1 (0.2) 1 (0.2) 81 (15.8) 4 (0.8) 1 (0.2) 1 (0.2) 1 (0.2) 1 (0.2) 1 (0.2) 512

has a low prevalence and, thus, limited influence on clinical outcomes in the study population. In contrast, α+-thalassemia trait is present in about 19% of the population, whereas the prevalence of α0-thalassemia trait is lower (1.6%).24

Mortality rates and causes of death By September 2018, 94 (17.5%) patients had died. Crude comparisons of proportions are shown in Table 2, indicating that survival was significantly associated with sex, splenectomy in childhood, and receiving oral chelation compared to those chelated primarily with deferoxamine. However, 44 of 70 (57.1%) of the patients who received deferoxamine chelation died before the year 2000, and these patients would not have had the opportunity to receive prolonged oral chelation. Those with more severe HBB genotypes were also more likely to survive (P=0.03 for trend). The mean age of death in the severe genotype category was higher (39.2 years, 95% confidence interval [95% CI]: 28.4-50.1) than in the moderate (mean age 31.4, 95% CI: 28.3–34.4) and mild (35.0 years, 95% CI: 30.0-40.1) genotype categories. There were no significant differences in proportions surviving or not surviving with different α-globin gene numbers or with the XmnI polymorphism. Causes of death are shown in Table 3. The ten deaths caused by liver disease included six cases of liver failure and four cases of hepatocellular carcinoma. Other malignancies comprised acute myeloid leukemia (n=2), T-cell lymphoma (n=1), metastatic melanoma (n=1), renal carcinoma (n=1), carcinoma of colon (n=1), carcinoma of skull (n=1). Other causes of death were categorized as not known (n=7), accident (n=5), stroke (n=2), hypoglycemia (n=1), myasthenia (n=1), acute graft-versus-host disease (n=1), and pulmonary hypertension (n=1). Table 3 also shows mortality rates per 10,000 patient-years from different causes during sequential haematologica | 2021; 106(9)

HBB genotype in TDT in Cyprus

Figure 1. Kaplan-Meier curves for survival from birth. (A) Survival of the overall cohort, and categorized by (B) sex, (C) HBB genotype, and (D) year of birth.

haematologica | 2021; 106(9)

2461

P. Kountouris et al.

Table 2. Demographic and clinical features of the cohort of patients.

Sex Male Female Year of birth < 1974 ≥ 1974 Age, at end of study period or death (years, mean, SD) Clinic Larnaca Limassol Nicosia Paphos HBB genotypeb Severe Moderate Mild Not known α-globin gene number 1 2 3 4 5 6 Not known XmnI polymorphism CC CT Not known Splenectomy Childhood Adulthood or not splenectomised Not known Iron chelation therapy DFO only DFO switched to DFP-containing DFO switched to DFX

Alive N=443 (%)

Died N=94 (%)

Total N=537

Comparison of proportion alive and died (P)

214 (77.0) 229 (88.4)

64 (23.0) 30 (11.6)

278 259

χ2 = 11.4, 1 d.f., P<0.001

186 (73.5) 257 (90.5) 43.1 (8.1)

67 (26.5) 27 (9.5) 30.4 (11.0)

253 284

χ2 = 25.5, 1 d.f., P<0.001

99 (81.8) 115 (82.1) 186 (81.2) 43 (91.5)

22 (18.2) 25 (17.9) 43 (18.8) 4 (8.5)

121 140 229 47

χ2 = 2.9, 3 d.f., P=0.401

77 (90.6) 293 (86.9) 70 (77.8) 2 (8.3)

8 (9.4) 44 (13.1) 20 (22.2) 22 (91.7)

85 337 90 24

χ2 = 6.8, 2 d.f., P=0.0341

2 (100) 14 (93.3) 84 (89.4) 328 (86.5) 2 (100) 1 (100) 12 (27.3)

0 (0) 1 (6.7) 10 (10.6) 51 (13.5) 0 (0) 0 (0) 32 (72.7)

2 15 94 379 2 1 44

χ2 = 1.8, 5 d.f., P=0.8781

407 (88.3) 19 (100) 17 (29.8)

54 (11.7) 0 (0) 40 (70.2)

461 19 57

χ2 = 1.5, 1 d.f., P=0.2251

108 (74.5) 334 (91.0) 1 (4.2)

37 (25.5) 33 (9.0) 24 (95.8)

145 367 25

χ2 = 22.7, 1 d.f., P<0.001a

52 (42.6) 311 (93.7) 80 (96.4)

70 (57.4) 21 (6.3) 3 (3.6)

122 332 83

χ2 = 176, 2 d.f., P<0.001

a Analysis restricted to the population in which the variable was known. bOne patient was removed from the genotype analysis because thalassemia was caused by α-locus duplications co-inherited with β-thalassemia trait.23 d.f.: degrees of freedom; SD: standard deviation; DFO: deferoxamine; DFP: deferiprone; DFX: deferasirox.

time periods of follow-up. There was a lower incidence of overall mortality in the period 1980-89, but no significant differences in mortality rates in subsequent time periods and no overall trend. There was a significant decrease in cardiac mortality over the last three time periods, and an increased incidence of deaths due to liver disease, cancer and infections during successive decades. However, there were insufficient numbers of deaths from these causes in each individual decade for a statistical analysis. The proportion of deaths due to heart disease was lower in those with a mild genotype (6 out of 20, 30%) than in those with moderate (23 out of 44, 52%) and severe (4 out of 8, 50%) genotypes. These differences did not reach statistical significance. Of the 15 subjects who died due to infectious causes, six of nine with known splenectomy status had been splenectomized. Five of 94 (5.3%) patients who died were HCV RNA positive at the time of death. Three of these died of heart failure and two from liver disease. HCV infection could be a contributing factor for these deaths, but was likely to be the primary cause in only two (2.1%) cases. 2462

Survival analysis We constructed two models to explore risk factors associated with mortality. In both models, data were right-censored on September 30, 2018, at death or bone marrow transplantation. In the first model, we considered the entire follow-up period and evaluated non-modifiable risk factors. These comprised sex, year of birth, treatment clinic and genotype. To assess the impact of treatment factors (iron chelation therapy and splenectomy), a second model was constructed. This took into account the licensing of the oral iron chelator deferiprone in 1999 and general availability for prescription from 2000. Deferiprone was sometimes used as single therapy and frequently in combination with deferoxamine, depending on the clinician’s assessment of iron overload severity. A second oral chelator, deferasirox was licensed and available for prescription in 2006 and there was switching between different chelator regimes over time after 2006 to optimize therapy for each individual. Since it would be biased to compare chelator efficacy during the period haematologica | 2021; 106(9)

HBB genotype in TDT in Cyprus

Table 3. Incidence rates for mortality by cause and period of follow-up.

Heart Incidence (95% CI)

1980-1989 4 7.5 (2.8–20.1) 1990-1999 21 41.3 (26.9–63.3) 2000-2009 13 27.4 (15.9–47.1) 2010-2018 6 15.3 (6.9–34.1) 1980-2018 44 23.1 (17.2–31.0)

N 1 2 2 5 10

Liver Incidence (95% CI) 1.9 (0.27–13.3) 3.9 4.2 (1.1–16.8) 12.8 (5.3–30.7) 5.3 (2.0–8.5)

N 0 0 3 4 7

Cancer Incidence (95% CI) 0.00 0.00 6.3 (2.0–19.6) 10.2 (3.8–27.2) 3.7 (1.0–6.4)

Infection Incidence (95% CI)

2 3.77 (0.9–15.0) 4 4 7.9 (3.0–20.9) 6 4 8.4 (3.2–22.4) 4 5 12.8 (5.3–30.7) 4 15 7.9 (4.7–13.1) 18

Other Incidence (95% CI) 7.5 (2.8–20.1) 11.8 (5.3–26.2) 8.4 (3.2–22.4) 10.2 (3.8–27.2) 9.4 (6.0–15.0)

Overall Incidence Patient-years (95% CI) of follow-up

11 33 26 24 94

20.7 (11.5–37.4) 64.9 (46.2–91.2) 54.8 (37.3–80.3) 61.3 (41.1–91.3) 49.3 (40.3–60.3)

5,310.7 5,085.7 4,747.9 3,915.8 19,060.0

Incidence presented as events per 10,000 patient-years; 95% CI: 95% confidence interval.

haematologica | 2021; 106(9)

Figure 2. KaplanMeier curves for survival during follow-up 2000 to 2018 showing the effects of risk factors with statistically significant effects in univariate analysis. (A) Survival of the overall cohort, and categorized by (B) sex, (C) HBB genotype, (D) year of birth, (E) splenectomy during childhood, and (F) type of chelation therapy. DFO: deferoxamine; DFP: deferiprone; DFX: deferasirox.

2463

P. Kountouris et al.

Table 4. Univariate and multivariate analysis of risk factors for survival in the primary analysis.

Univariate % survival at age 50 (95% CI) Comparison of survival estimate Sex Female Male Year of birth <1974 ≥1974 Clinic Larnaca Limassol Nicosia Paphos HBB genotypea Severe Moderate Mild α-globin gene numberb <4 ≥4 Xmn polymorphismb CC CT

87.3 (82.9-91.8) 73.6 (66.9-81.0)

χ2 =13.62, 1 d.f., P<0.001

75.4 (70.1-81.1) Not yet reached

χ2 =10.2, 1 d.f., P=0.001

79.4 (71.6-88.0) 82.0 (75.1-89.5) 78.7 (72.0-86.0) 84.5 (68.9-100.0)

χ2 =2.7, 3 d.f., P=0.437

85.7 (75.4-97.5) 85.3 (80.6-90.2) 77.9 (68.7-88.3)

χ2 =4.4, 1 d.f., P=0.037

90.9 (84.5-97.7) 83.6 (78.8-88.8)

χ2 =2.01, 1 d.f., P=0.157

86.2 (82.1-90.5) 100

χ2 =2.8, 1 d.f., P=0.097

Multivariate

HR 1.9, 95% CI 1.2–3.0, P=0.011

HR 1.6, 95% CI 1.1–2.3, P=0.023

a The vector of trend weight for the three categories was 0, 1 and 2 for severe, moderate and mild genotype, respectively. bThe analysis was based on cases for which data were available (see Table 2 for missing data). d.f.: degrees of freedom; 95% CI: 95% confidence interval, HR: hazard ratio.

1980-2000, when the only option available was deferoxamine, follow-up was restricted to the period 2000-2018, when at least one option for oral chelation was available, and data were left-truncated at January 1, 2000. The results of the survival analysis are shown in Table 4 and Figure 1. Survival by age 50 was estimated at 80.4% (95% CI: 76.4-84.7). In the first model of survival, univariate analysis showed that male sex, milder HBB genotype and birth before 1974 were significantly associated with worse survival, while there was no significant effect of treatment clinic, α-globin gene number or XmnI polymorphism. When the survival analysis was applied separately to cardiac and non-cardiac deaths, the genotype had no significant effect on cardiac deaths, but a significant effect on non-cardiac deaths (P=0.02). In the multivariate model, independent predictors of survival were male sex (hazard ratio [HR] 1.9, 95% CI: 1.2-3.0, P=0.01) and genotype, with each decrement in severity being associated with a 1.6-fold increased risk of mortality (HR 1.6, 95% CI: 1.1-2.3, P=0.02). In the second model, 489 evaluable patients were followed for a total of 8,644 person-years. The estimated survival rate was 89.9% (95% CI: 87.3–92.6) as of September 30, 2018. Univariate and multivariate analyses of risk factors for survival over this time period are shown in Table 5 and Figure 2. In the univariate analysis, male sex, birth before 1974, milder genotype, splenectomy in childhood and deferoxamine-only chelation treatment were significantly associated with worse survival, with iron chelation treatment being the most significant factor. In multivariate analysis, the best-fit model included chelation, sex, splenectomy in childhood and HBB genotype. Importantly, the significant effects of male sex and milder genotype on survival were confirmed in the second multivariate model, while deferoxamine-only treatment and splenectomy in childhood were also associated with worse survival. 2464

Association of genotype with transfusion parameters Transfusion data were available for 336 patients of the cohort (62.6%) and were used to evaluate possible associations between HBB genotype and the transfusion regimens. The mean age of first regular transfusion was 29.4 months, the mean transfusion frequency per year was 28.7 times, the mean blood volume transfused was 179 mL/kg/year and the mean pre-transfusion hemoglobin level was 9.9 g/dL. The HBB genotype was significantly associated with age at first regular transfusion (P=0.001), annual transfusion frequency (P=0.04) and blood volume transfused (P=0.01), whereas no association was found for pre-transfusion hemoglobin levels (P=0.8). Post-hoc statistical analysis was subsequently performed to demonstrate differences in transfusion between different genotypic severity groups, shown in Figure 3. Statistically significant differences were identified between the severe and mild groups for age at first transfusion (P=0.004) and annual transfusion frequency (P=0.024). In addition, statistically significant differences were found between the moderate and mild groups for age at first regular transfusion (P=0.0006), annual transfusion frequency (P=0.025) and blood volume transfused (P=0.003). The was not a significant association between the severe and moderate groups for any of the transfusion parameters studied.

Discussion The first report on this cohort was right-censored at December 31, 2004.1 The current data cut-off allows a further 14 years of follow-up, and more robust estimates of overall survival, trends in mortality and risk factors for mortality. In a study of health outcomes and healthcare costs in the UK,11 survival of thalassemia patients treated according to haematologica | 2021; 106(9)

HBB genotype in TDT in Cyprus

Figure 3. Boxplots illustrating the distribution of transfusion data among groups with HBB genotypes of different severity. Statistically significant associations, determined with the Wilcoxon rank-sum non-parametric test, are shown.

standard guidelines was forward-estimated at 63% by the age of 50 years. This was based on a review of available data in which clinical follow-up was only reported up to 30 years of age. The estimate of 80.4% in this study is more robust and significantly more optimistic. Life expectancy in the general population of Cyprus has been reported recently using national statistics over the period 1986–2012.25 About 90–95% are expected to survive to 50 years, and our data would suggest that a diagnosis of thalassemia reduces the percentage surviving to this age by about 15%. These data should be helpful in discussing outcomes of thalassemia with new parents, in refining health economic models, comparing outcomes of new therapies with standard care, and in making treatment decisions about use of alternative donor transplantation, where overall survival with unrelated, or mismatched family donors are currently less than 80%.26 Causes of death appear to be changing with longer term follow-up of TDT patients. We have confirmed reduction in deaths due to cardiac causes over the past 20 years, as reported previously in this cohort1 and in other studies.5 We also noted increasing mortality related to liver disease and haematologica | 2021; 106(9)

infection in this cohort, a finding that has also been reported in recent registry studies from Sicily and Greece.6,27 The increasing numbers of deaths due to malignancies has also been noted previously.28 Although the incidence of cancer naturally increases with age, and it has not been established that the incidence is significantly different from that in the general population, the types of malignancy appear to be different, notably with a higher representation of hepatocellular carcinoma. Further follow-up of the cohort and evaluation in other similar studies will be needed to confirm this finding. The distribution of HBB genotypes in this study is consistent with the known distribution of thalassemia alleles in the Cypriot population.29,30 There is a predominance of homozygosity for IVS I-110 G>A, a severe β+ variant causing a cryptic splice site in the first intron of HBB which reduces functional β-globin mRNA to about 20% of normal.31 Next in prevalence is IVS 1-6 T>C, causing aberrant splicing at the first intronic junction of HBB, with a less severe reduction in β-globin mRNA. Being a milder variant, it has been classified as β++ or β+ in different studies. We have selected the former annotation for this study.21 The 2465

P. Kountouris et al.

Table 5. Univariate and multivariate analysis of risk factors for survival in the secondary analysis.

Number

Sex Female Male Year of birth <1974 ≥1974 Clinic Larnaca Limassol Nicosia Paphos HBB genotype Severe Moderate Mild α-globin gene number <4 ≥4 XmnI polymorphism CC CT Iron chelation DFO only DFO switched to DFP-containing DFO switched to DFX Splenectomy Childhood Adulthood or not splenectomized

Univariate % survival at right Comparison of censorship at survival estimate 30 Sept. 2018 (95% CI) (log rank)

245 244

93.0 (89.9-96.3) 86.7 (82.6-91.1)

χ2 =5.3, 1 d.f., P=0.021

215 274

86.0 (81.5-90.8) 93.0 (90.0-96.1)

χ2 =6.3, 1 d.f., P=0.012

107 128 209 45

90.3 (84.8-96.2) 89.0 (83.7-94.6) 89.0 (84.9-93.3) 95.6 (89.7-100)

χ2 =2.0, 3 d.f., P=0.578

81 321 84

93.8 (88.7-99.2) 90.9 (87.7-94.1) 82.1 (74.3-90.8)

χ2 =6.4, 1 d.f., P=0.011

108 369

92.5 (87.7-97.7) 89.1 (86.0-93.2)

χ2 =1.1, 1 d.f., P=0.287

452 19

89.5 (86.7-92.4) 100

χ2 =2.0, 1 d.f., P=0.158

74 332 83

64.5 (54.2-76.8) 93.7 (91.1-96.3) 96.4 (92.4-100)

χ2 =75.1, 2 d.f., P<0.001

130 354

82.2 (75.8-89.1) 93.7 (91.2-96.3)

χ2=14.7, 1 d.f., P<0.001

Multivariate

HR 2.34, 95% CI 1.26–4.34, P=0.007

HR 2.04, 95% CI 1.05–3.98, P=0.036

HR 0.10, 95% CI 0.05–0.19, P<0.001 HR 0.08, 95% CI 0.025–0.028, P<0.001 HR 7.92, 95% CI 2.03–30.75, P=0.003

d.f.: degrees of freedom; 95% CI: 95% confidence interval, HR: hazard ratio: DFO: deferoxamine; DFP: deferiprone; DFX: deferasirox.

most severe category (β0/β0) is rare in the Cypriot TDT population (0.8%), and this contrasts with some other TDT populations, such as in Sardinia where 92% of TDT patients have homozygous or compound heterozygous β0 mutations.16 Nevertheless, the two commonest variants in this cohort are common in the Mediterranean and the Middle East, and are also observed in many immigrant populations.18 For risk factor analysis, we introduced a genotypic severity categorization, with the assumption that severity increases linearly across the spectrum mild/ moderate/severe. This categorization could be explored in studies of severity, treatment and outcomes in other populations, even if the specific variants are different. The variability of risk factors during follow-up creates problems for the statistical modeling. We addressed this by evaluating two different models, the first of which incorporated only non-modifiable factors and did not include treatment factors, but assumed that standard care was applied uniformly to all subjects. The worse survival associated with male sex had been previously reported1,2 and could be linked to worse adherence to the recommended chelation therapy and a higher prevalence of specific complications, such as heart disease, in male patients. Moreover, the significantly worse survival associated with milder HBB genotype would suggest that standard care is not sufficiently effective to cancel the pathological effects of milder genotype. One recent confirmatory study has shown a convergence in mortality in a population of thalassemia major and thalassemia intermedia patients in Sicily with similar length 2466

of follow-up.27 The definition of thalassemia intermedia in the study from Sicily was regular transfusion after the age of 2 years. Based on the transfusion data of the current study, the above definition of thalassemia intermedia encompasses 44.4% (122 out of 275) of patients with moderate or mild genotypes in the Cyprus TDT patient cohort. The second model, which included treatment effects, confirmed that male sex and milder HBB genotype were significantly associated with mortality. Here, chelation therapy for each individual was categorized according to the predominant agent used. We confirmed that oral iron chelation therapy had a strong independent protective effect on survival compared to continuing with deferoxamine therapy. The positive effect on survival of switching from deferoxamine- to deferiprone-containing therapy has been previously reported from this cohort and in other studies.3,32 The survival benefit of deferasirox compared to deferoxamine has not previously been reported, and may be related to improved adherence to an oral rather than injected agent and its once-daily administration. It is important to note that the categorization of chelation therapy is a simplification, and the assumptions of the Cox proportional-hazards model are not fully met with regard to chelation. Therefore, the significant protective effects observed in this study should be interpreted with caution. We did not have comprehensive data on other genetic modifiers,33 while some genotype data were missing, particularly α-globin genotype (44/537, 8.2%) and XmnI polymorphism (57/537, 10.6%). However, the prevalence of haematologica | 2021; 106(9)

HBB genotype in TDT in Cyprus

severity-alleviating factors, such as α0 deletions and minor allele of the XmnI polymorphism, is low in this population and is unlikely to have affected the overall analysis of survival. In contrast, HBB genotype appears to be a significant determinant of severity, a finding confirmed in a study of likelihood of initiation of transfusion in patients with β-thalassemia, in whom transfusion was 30 times more likely in those with a severe genotype than in those with a mild genotype.15,16 The effect of genotype on survival could potentially be biased due to deaths of patients with severe genotypes at an earlier age and exclusion of patients who died early in the follow-up, before genotyping was introduced in Cyprus. Nevertheless, this is unlikely to have a major influence on our conclusions, because there was no significant difference between genotype frequencies and age at death in patients with available genotypes, while the year of birth was used in a multivariate survival analysis (data not shown) without materially changing the significant effect of genotype on survival. Survival in TDT is associated with a variety of risk factors, not all of which are quantifiable, and these risk factors are likely to differ according to the cause of mortality. The cause of mortality appears to be changing from cardiac iron overload to other etiologies more closely associated with the long-term effects of inadequate suppression of the underlying thalassemia phenotype. Our results suggest that HBB genotype could affect mortality through its causative effect on disease phenotype, which is now more apparent because of the reduction of cardiac deaths. The analysis of transfusion data shows that patients with mild genotypes are beginning transfusion significantly later in life and require less blood. However, once started on transfusion, there is no difference in pre-transfusion hemoglobin level between the genotypes, indicating that lower volumes of blood are required to suppress ineffective erythropoiesis in the group with mild genotypes. These observations in this study suggest that the delay in starting transfusion, along with the long-term transfusion strategy, are likely to have a negative impact on the survival of patients with mild genotypes. Notably, current treatment guidelines for TDT recommend that the decision about starting transfusion is based on clinical assessment. Our results suggest that determining the HBB geno-

References 1. Telfer P, Coen PG, Christou S, et al. Survival of medically treated thalassemia patients in Cyprus. Trends and risk factors over the period 1980-2004. Haematologica. 2006; 91(9):1187-1192. 2. Borgna-Pignatti C, Rugolotto S, De Stefano P, et al. Survival and complications in patients with thalassemia major treated with transfusion and deferoxamine. Haematologica. 2004;89(10):1187-1193. 3. Borgna-Pignatti C, Cappellini MD, De Stefano P, et al. Cardiac morbidity and mortality in deferoxamine- or deferiprone-treated patients with thalassemia major. Blood. 2006;107(9):3733-3737. 4. Modell B, Khan M, Darlison M, et al. A national register for surveillance of inherited disorders: beta thalassaemia in the United Kingdom. Bull World Health Organ. 2001; 79(11):1006-1013. 5. Modell B, Khan M, Darlison M, Westwood MA, Ingram D, Pennell DJ. Improved survival of thalassaemia major in the UK and

haematologica | 2021; 106(9)

type and using this information to aid in decision-making could improve long-term outcomes. The effect of genotype on mortality could be examined in other previously described thalassemia cohorts2,6,7 and in combined datasets, thus allowing the study of a more diverse spectrum of genotypes. This is a future direction of our work. If confirmed, the effect of genotype could have significant implications for the management of TDT. Firstly, it would provide evidence that patients diagnosed with β-thalassemia should initiate regular transfusion early, using a transfusion regimen which effectively suppresses erythropoiesis, to avoid long-term complications of uncontrolled anemia and marrow expansion. Secondly, it would contribute to the decision of whether to institute emerging strategies for treating TDT including enhancement of endogenous erythropoiesis,34 as well as lentiviral and gene editing strategies,35,36 both of which appear to be more effective in patients with milder genotypes and higher residual globin chain production. Disclosures PT was principal investigator in a phase III study of luspatercept, a product of Celgene Corporation; he has received research funding from and participated in advisory boards for Bluebird Bio, Inc; and has participated in advisory boards for Novartis and Apopharma. Contributions PK, MK and PT designed the study protocol, and wrote and edited the manuscript; SC, MH, MS and AK managed the patients according to national and international guidelines and collected clinical data; PK and MK performed the DNA studies; PK, KM and PT performed the statistical analysis. Acknowledgments We wish to thank the Cypriot patients and families attending the Greek Cypriot thalassemia clinics and the staff in these centers who have provided dedicated, long-term care over the study period. Funding This work was partially funded by the THALAMOSS project (FP7-HEALTH-2012-INNOVATION-1: grant agreement 306201).

relation to T2* cardiovascular magnetic resonance. J Cardiovasc Magn Reson. 2008; 10:42. 6. Voskaridou E, Kattamis A, Fragodimitri C, et al. National registry of hemoglobinopathies in Greece: updated demographics, current trends in affected births, and causes of mortality. Ann Hematol. 2019;98(1):55-66. 7. Ladis V, Chouliaras G, Berdoukas V, et al. Relation of chelation regimes to cardiac mortality and morbidity in patients with thalassaemia major: an observational study from a large Greek Unit. Eur J Haematol. 2010;85(4):335-344. 8. UK Thalassaemia Society. Standards for the Clinical Care of Children and Adults with Thalassaemia in the UK, 3rd Edition. 2016. 9. Cappellini MD, Cohen A, Porter J, Taher A, Viprakasit V, eds. Guidelines for the Management of Transfusion Dependent Thalassaemia (TDT). Thalassaemia International Federation. 2014 10. Musallam KM, Angastiniotis M, Eleftheriou A, Porter JB. Cross-talk between available guidelines for the man-

agement of patients with beta-thalassemia major. Acta Haematol. 2013;130(2):64-73. 11. Weidlich D, Kefalas P, Guest JF. Healthcare costs and outcomes of managing beta-thalassemia major over 50 years in the United Kingdom. Transfusion. 2016;56(5):10381045. 12. Cappellini MD, Porter JB, Musallam KM, et al. Development of a new disease severity scoring system for patients with non-transfusion-dependent thalassemia. Eur J Intern Med. 2016;28:91-96. 13. Ratip S, Skuse D, Porter J, Wonke B, Yardumian A, Modell B. Psychosocial and clinical burden of thalassaemia intermedia and its implications for prenatal diagnosis. Arch Dis Child. 1995;72(5):408-412. 14. Cappellini MD, Kattamis A, Viprakasit V, et al. Quality of life in patients with beta-thalassemia: a prospective study of transfusiondependent and non-transfusion-dependent patients in Greece, Italy, Lebanon, and Thailand. Am J Hematol. 2019;94(10):E261E264. 15. Cappellini MD, Musallam KM, Taher AT.

2467

P. Kountouris et al.

Insight onto the pathophysiology and clinical complications of thalassemia intermedia. Hemoglobin. 2009;33(Suppl 1):S145-159. 16. Danjou F, Anni F, Perseu L, et al. Genetic modifiers of β-thalassemia and clinical severity as assessed by age at first transfusion. Haematologica. 2012;97(7):989-993. 17. Danjou F, Francavilla M, Anni F, et al. A genetic score for the prediction of beta-thalassemia severity. Haematologica. 2015; 100(4):452-457. 18. Kountouris P, Lederer CW, Fanis P, Feleki X, Old J, Kleanthous M. IthaGenes: an interactive database for haemoglobin variations and epidemiology. PLoS One. 2014; 9(7):e103020. 19. Thein SL. Molecular basis of beta thalassemia and potential therapeutic targets. Blood Cells Mol Dis. 2018;70:54-65. 20. Angastiniotis MA, Hadjiminas MG. Prevention of thalassaemia in Cyprus. Lancet. 1981;1(8216):369-371. 21. Thein SL. The molecular basis of beta-thalassemia. Cold Spring Harb Perspect Med. 2013;3(5):a011700. 22. [No authors listed] Thalassemia Modular Stratification System for personalized therapy of beta-thalassemia (THALAMOSS). Hum Gene Ther Clin Dev. 2015;26(2):100102. 23. Clark B, Shooter C, Smith F, et al. Beta thalassaemia intermedia due to co-inheritance of three unique alpha globin cluster duplica-

2468

tions characterised by next generation sequencing analysis. Br J Haematol. 2018; 180(1):160-164. 24. Kyriacou K, Kyrri A, Kalogirou E, et al. Hb Bart's levels in cord blood and alpha-thalassemia mutations in Cyprus. Hemoglobin. 2000;24(3):171-180. 25. Agathokleous MN, Nena E, Chadolias D, et al. Estimating life expectancy of the population in Cyprus with the use of life tables. Hippokratia. 2016;20(2):99-103. 26. Baronciani D, Angelucci E, Potschger U, et al. Hemopoietic stem cell transplantation in thalassemia: a report from the European Society for Blood and Bone Marrow Transplantation Hemoglobinopathy Registry, 2000-2010. Bone Marrow Transplant. 2016;51(4):536-541. 27. Vitrano A, Calvaruso G, Lai E, et al. The era of comparable life expectancy between thalassaemia major and intermedia: Is it time to revisit the major-intermedia dichotomy? Br J Haematol. 2017;176(1):124-130. 28. Zanella S, Garani MC, Borgna-Pignatti C. Malignancies and thalassemia: a review of the literature. Ann N Y Acad Sci. 2016; 1368(1):140-148. 29. Baysal E, Indrak K, Bozkurt G, et al. The beta-thalassaemia mutations in the population of Cyprus. Br J Haematol. 1992; 81(4):607-609. 30. Kountouris P, Kousiappa I, Papasavva T, et al. The molecular spectrum and distribution

of haemoglobinopathies in Cyprus: a 20year retrospective study. Sci Rep. 2016; 6:26371. 31. Spritz RA, Jagadeeswaran P, Choudary PV, et al. Base substitution in an intervening sequence of a beta+-thalassemic human globin gene. Proc Natl Acad Sci U S A. 1981; 78(4):2455-2459. 32. Telfer PT, Warburton F, Christou S, et al. Improved survival in thalassemia major patients on switching from desferrioxamine to combined chelation therapy with desferrioxamine and deferiprone. Haematologica. 2009;94(12):1777-1778. 33. Stephanou C, Tamana S, Minaidou A, Papasavva P, Kleanthous M, Kountouris P. Genetic modifiers at the crossroads of personalised medicine for haemoglobinopathies. J Clin Med. 2019;8(11):1927. 34. 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. 35. Thompson AA, Walters MC, Kwiatkowski J, et al. Gene therapy in patients with transfusion-dependent beta-thalassemia. N Engl J Med. 2018;378(16):1479-1493. 36. Magrin E, Miccio A, Cavazzana M. Lentiviral and genome-editing strategies for the treatment of beta-hemoglobinopathies. Blood. 2019;134(15):1203-1213.

haematologica | 2021; 106(9)

ARTICLE

Red Cell Biology & its Disorders

Interleukin-1 receptor inhibition reduces stroke size in a murine model of sickle cell disease

Ferrata Storti Foundation

Jessica Venugopal,1 Jintao Wang,1 Jalal Mawri,2 Chiao Guo1 and Daniel T. Eitzman1 University of Michigan Internal Medicine - Cardiology Division and 2University of Michigan, Ann Arbor, MI, USA

ABSTRACT

Haematologica 2021 Volume 106(9):2469-2477

ickle cell disease (SCD) is associated with chronic hemolytic anemia and a heightened inflammatory state. The causal role of inflammatory pathways in stroke associated with SCD is unclear. Therefore, the hypothesis that deletion of the non-hematopoietic interleukin-1 receptor (IL-1R) pool may be beneficial in SCD was pursued. Since potential deleterious effects of IL-1R signaling in SCD could be mediated via downstream production of interleukin-6 (IL-6), the role of the nonhematopoietic IL-6 pool was also addressed. Bone marrow transplantation (BMT) from SCD to wild-type (WT) recipient mice was used to generate SCD mice (Wt,SCDbmt). In order to generate mice with nonhematopoietic deficiency of IL-1R or IL-6, SCD marrow was transplanted into IL-1R deficient (IL1R-/-,SCDbmt) or IL-6 deficient recipients (IL6-/-, SCDbmt). Blood counts, reticulocytes, soluble E-selectin (sEsel), and IL-6 levels were analyzed 14-15 weeks post-BMT. Ischemic stroke was induced by middle cerebral artery (MCA) photothrombosis at 16 weeks post-BMT. A separate group of Wt,SCDbmt mice was given the IL-1R inhibitor, anakinra, following stroke induction. Seventy-two hours after MCA occlusion, stroke volume was assessed by staining brain sections with 2,3,5-triphenyltetrazolium chloride. Formalin-fixed brain sections were also stained for macrophages with MAC3, for endothelial activation with ICAM-1, and for loss of blood brain barrier integrity with fibrin(ogen) staining. All SCD mice generated by BMT were anemic and the severity of anemia was not different between Wt,SCDbmt, IL1R-/-,SCDbmt, and IL-6-/-,SCDbmt mice. Three days following MCA occlusion, stroke volume was significantly reduced in IL1R-/-,SCDbmt mice compared to Wt,SCDbmt mice and IL6-/-,SCDbmt mice. Plasma levels of sEsel were lower in IL1R-/-,SCDbmt compared to Wt,SCDbmt and IL-6-/-,SCDbmt mice. Post-stroke treatment of Wt,SCDbmt mice with anakinra decreased stroke size, leukocyte infiltration, ICAM-1 expression, and fibrin(ogen) accumulation compared to vehicle-treated mice. Deficiency of non-hematopoietic IL-1R or treatment with an IL-1R antagonist is sufficient to confer protection against the increased stroke size associated with SCD. These effects of IL1R deficiency are associated with reduced endothelial activation, leukocyte infiltration, and blood brain barrier disruption, and are independent of non-hematopoietic IL-6 signaling.

Introduction Sickle cell disease (SCD) is associated with acute and chronic vascular complications leading to premature morbidity and mortality, including adverse cerebrovascular events, such as stroke.1 The stroke risk for a child with SCD is over 300 times greater than for a child without SCD2 with clinically apparent strokes occurring in 11% of SCD patients before the age of 20. Approximately two thirds of these patients experience recurrent cerebral infarction.3 Genotype greatly influences the

haematologica | 2021; 106(9)

Correspondence: DANIEL T. EITZMAN deitzman@umich.edu Received: March 10, 2020. Accepted: August 5, 2020. Pre-published: August 13, 2020. https://doi.org/10.3324/haematol.2020.252395

2469

J. Venugopal et al.

risk of stroke in SCD patients, with those with hemoglobin (Hb)SS having the highest risk and those with sickle β+-thalassemia having the lowest risk.2 In general, the most prevalent subtype of stroke associated with SCD patients is ischemic stroke, however between the ages of 20 and 29 hemorrhagic strokes are more prevalent.2 Stroke prevention in SCD patients is primarily accomplished through chronic blood transfusions4 and hydroxyurea treatment.5-7 Although erythrocyte sickling in response to stressors constitutes the primary underlying defect of SCD, subsequent inflammatory responses to vascular occlusive events contribute to organ damage and further vascular dysfunction.8-11 This heightened inflammatory milieu is characterized by leukocytosis and elevated levels of cytokines in SCD.12-14 Therapies shown to be beneficial in SCD such as hydroxyurea and anti-selectin antibodies may exert their beneficial effects, in part, via dampening of leukocyte-mediated inflammatory responses.15,16 Hemolysis in SCD may result in the activation of leukocytes via Toll-like receptors (TLR) and NOD-like receptor 3 (NLRP3) by free heme.17 TLR and NLRP3 inflammasome expression levels, including interleukin-1β (IL-1β ), are increased in peripheral blood monocytes from SCD patients.18,19 IL-1β is a particularly important mediator of acute and chronic inflammatory disease processes, as therapeutic targeting of IL-1β has proven beneficial in several inflammatory diseases.20-22 Additionally, some genetic polymorphisms of IL-1β have been shown to affect IL-1β transcription and are associated with arthritis, cardiovascular disease, and complications of SCD.9,23,24 While these studies suggest IL-1β signaling pathways are involved in some manifestations of SCD, the causal role of these pathways remains unclear. IL-1β may represent a particularly important modulator of stroke outcomes.25,26 IL-1β is rapidly upregulated during ischemic stroke and may contribute to ischemic injury.26 In a meta-analysis of 16 non-SCD animal studies, administration of the IL-1R antagonist, anakinra, produced a 36% reduction of infarct volume.27 IL-1β may promote neuronal death indirectly, via effects on astrocytes and endothelial cells.25 The binding of IL-1β to astrocyte IL-1R activates signaling cascades resulting in the production of IL-6, TNF-α, and other chemokines which influence central nervous system (CNS) inflammation28 and neurotoxicity.28 Neurotoxicity mediated by IL-1β may also occur through endothelial interleukin-1 receptor (IL-1R)-mediated activation of cerebral endothelial cells,30 leading to leukocyte infiltration31 and the loss of blood brain barrier integrity.32 The recruitment of peripheral leukocytes by IL-1β can sustain neuroinflammation,33 further promoting neurotoxicity,34 and blood brain barrier (BBB) disruption.35 IL-1β may also induce permeability of the BBB directly through endothelial cell signaling pathways.36 Mouse models of SCD have been developed that mimic the predominant features of SCD in humans.37-39 In general, these mice exhibit hemolysis, anemia, splenomegaly, and multi-organ infarcts.37-39 SCD mice have thus been a useful aid to identify mechanisms involved in vaso-occlusion and to test potential therapeutic interventions. Because of reduced fertility and complex genetics, generating SCD mice with complete deficiency of a diseasemodifying candidate gene through intercrosses is cumbersome, as is generation of suitable littermate controls. However, bone marrow transplantation (BMT) is an effi2470

cient means to generate SCD mice, and if a candidate gene of interest exerts its effects via non-bone marrow-derived cellular pools, then informative SCD mice can be readily generated by transplanting SCD marrow to recipient mice with deficiency of the candidate gene. In order to modify IL-1β signaling pathways using this strategy, transplantation of SCD marrow to mice lacking the receptor for IL-1β, (IL-1R), leads to lack of IL-1 signaling in nonhematopoietic IL-1R cellular pools. The endothelial IL-1R pool is responsible for mediating the upregulation of endothelial adhesion molecules and leukocyte-endothelial interactions in response to IL-1β stimulation,40 which might be particularly relevant to SCD pathogenesis. Therefore, to study IL-1β signaling pathways in SCD, mice were generated by transplanting SCD marrow into recipients with IL-1R deficiency and compared to control wild-type (WT) recipients on the same C57BL6/J strain background. The role of IL-1R signaling was then analyzed with regards to anemia and stroke in SCD mice. The effect of an IL-1R pharmacologic antagonist was also assessed.

Methods Animals Male C57BL/6J wild-type (WT), homozygous SCD ( SCD, Stock No:013071 Townes model), IL1R null mice (IL1R-/-, Stock No: 003245), interleukin-6 null mice (IL6-/-, Stock No: 002650) were purchased from Jackson Laboratory (Bar Harbor, Maine, USA). SCD and control experimental mice were then generated by BMT from SCD mice into WT, IL1R-/-, and IL6-/- recipients. Additional controls were generated by transplantation of WT marrow into WT recipients. Mice were housed under specific pathogen-free conditions in static microisolator cages with tap water ad libitum in a temperature-controlled room with a 12:12-hour light/dark cycle. Mice were fed a standard laboratory rodent diet (No. 5001, TestDiet, Richmond, IN, USA). All animal use protocols complied with the Principle of Laboratory and Animal Care established by the National Society for Medical Research and were approved by the University of Michigan Committee on Use and Care of Animals.

Bone marrow transplantation and blood parameter analysis SCD mice were generated by BMT as previously described.41,42 Briefly, 8 week-old male WT, IL1R-/- and IL6-/- mice were used as recipients that received bone marrow from 8 week-old SCD male donors. Bone marrow was harvested from the donor mice by flushing their femurs and tibias with RPMI medium (Gibco/Invitrogen, Carlsbad, CA, USA) containing 10% fetal bovine serum (Gibco/Invitrogen, Carlsbad, CA, USA). Cells were then centrifuged at 300g and resuspended in phosphate-buffered saline before injection. Each recipient mouse was irradiated (2×650 rad [0.02×6.5 Gy]) and injected with 4×106 bone marrow cells via the tail vein in a 200 μL bone marrow suspension in phosphate-buffered saline. Acid water (6 mM HCl, pH=2.5) was provided to animals beginning 4 days before BMT to 4 weeks following BMT. Transplant efficiency was determined by hemoglobin electrophoresis, as done previously.41,43,15 weeks following BMT, blood parameter analyses were performed with a Hemavet (Drew Scientific, Inc) on whole blood collected in EDTA-lined tubes via retro-orbital sampling from isofluorane-anesthetized mice (n=5 per group). Reticulocyte percentages were quantified by new methylene blue staining (n=5 per group) (Ricca Chemical

haematologica | 2021; 106(9)

Interleukin-1 receptor in sickle cell disease

Company, Arlington, TX, USA), according to the manufacturer's instructions and expressed as a percentage of total erythrocytes.

Stroke model Sixteen weeks after BMT, middle cerebral artery (MCA) occlusion was induced by photochemical injury as previously described,41,42 n=3-5. On day 3 following MCA occlusion, mice were anesthetized with pentobarbital, and blood drawn via cardiac puncture. Mouse bodies were perfused with PBS, then brains were excised and sliced into 2 mm segments before staining for 20 minutes with 2, 3, 5-triphenyltetrazolium chloride at room temperature while protected from light. The brain sections were imaged with a Nikon SMZ-2T microscope and Spot Idea camera model 29.2-13MP using Spot 5.1 software, and stroke size was then calculated as done previously.41 Brain macrophages were stained with an anti-mouse MAC3 antibody (1:200; #550292, BD Biosciences, San Jose, CA), n=3-5. Fibrin(ogen) was stained with an anti-mouse fibrin(ogen) antibody (1:4,000; ab189490, Abcam, Cambridge, MA) and ICAM-1 was stained with anti-mouse ICAM-1 antibody (2 ug/mL; #14-0542-85, ThermoFisher Scientific, Waltham, MA), followed by a biotin-conjugated secondary IgG (1:100), n=3-5. A Nikon Microphot-SA Epi-FL3 microscope, Nikon Ds Fi3 camera, and NIS Elements software were used to capture images. Quantification of fibrin(ogen) stained area was performed with Image J software. Quantification of MAC3positive cells was attained for each mouse by manually counting stained cells in 20 fields of view at 10x. For each field, the number of MAC3-positive cells was divided by the area of brain imaged in each field.

BMT. Compared to WT mice transplanted with WT bone marrow (Wt,WTbmt), WT recipients of SCD bone marrow (Wt,SCDbmt) were more anemic with elevated leukocyte and reticulocyte counts (Figure 1). IL-1R-/- and IL-6-/- recipients of SCD marrow (IL1R-/-,SCDbmt and IL6-/-, SCDbmt) displayed similar anemia and reticulocyte counts compared to Wt,SCDbmt mice (Figure 1).

Effect of IL-1R and IL-6 status on circulating levels of IL6 in SCD mice Plasma levels of IL-6 were detectable in Wt,SCDbmt mice (4.877±3.62 pg/mL) but undetectable in both IL-1R-/-,SCDbmt and IL6-/-,SCDbmt mice, consistent with a non-hematopoietic source for circulating IL-6 in SCD and a critical role for non-hematopoietic IL-1 receptor signaling towards IL-6 levels in SCD.

Effect of IL-1R and IL-6 status on stroke size in sickle cell disease following middle cerebral artery occlusion

Results

SCD mice have been shown to experience larger strokes following MCA occlusion due to vasocclusion by sickled erythrocytes in the penumbra.42 In order to determine whether IL1R-/-,SCDbmt mice would be protected from the increased stroke size associated with SCD, photochemicalmediated thrombosis was induced in the MCA in Wt,SCDbmt mice and IL-1R-/-,SCDbmt mice. Three days later, the stroke area was quantitated and IL-1R-/-,SCDbmt mice were found to have a similar stroke areas to Wt,WTbmt mice, both of which had reduced areas when compared to Wt,SCDbmt mice (Figure 2A to C). In contrast, stroke size in IL6-/-,SCDbmt mice was not reduced compared to Wt,SCDbmt mice (Figure 2D). Thus, although non-hematopoietic IL-1R signaling pathways regulate circulating IL-6 levels, this pathway does not account for the effects of nonhematopoietic IL-1R signaling on stroke size in SCD. Reduction in stroke size in IL-1R-/-,SCDbmt mice was also associated with reduced peri-infarct infiltration of macrophages, as denoted by staining of MAC3 (Figure 3). Since endothelial IL-1R signaling regulates expression of endothelial adhesion molecules40,44 which could affect the stroke phenotype, levels of sEsel were measured given its endothelial specificity. Plasma levels of sEsel were found to be significantly reduced in IL-1R-/-,SCDbmt compared to Wt,SCDbmt mice (32.12±2.08 ng/mL vs. 50.10±2.31 ng/mL; P=0.03). Circulating values of sEsel in IL6-/-,SCDbmt were not significantly different compared to Wt,SCDbmt (54.03±10.10 ng/mL, P=0.37). Fixed brain sections were also stained for ICAM-1. Expression of ICAM-1 was significantly decreased IL-1R-/-,SCDbmt compared to Wt,SCDbmt mice (0.008±0.002% area vs. 0.029±0.007% area, P<0.05). Increased infiltration of leukocytes may diminish blood brain barrier integrity, leading to leakage of fibrin(ogen)containing plasma into the brain from the vasculature.45-47 Fibrin(ogen) immunopositivity was significantly decreased in IL-1R-/-,SCDbmt compared to Wt,SCDbmt mice, whereas IL6-/-,SCDbmt were not significantly different than Wt,SCDbmt mice (Figure 4).

Effect of IL-1R and IL-6 status on hematological data in SCD mice

Effect of single dose anakinra on stroke size given post middle cerebral artery occlusion

In order to determine whether signaling through the IL1R or IL-6 in non-hematopoietic tissues would impact anemia in SCD mice, whole blood was analyzed for cell counts, platelets, and reticulocytes 15 weeks following

From a practical therapeutic standpoint, treatment with antagonists of IL-1β or IL-1 receptor signaling pathways may not be feasible in SCD patients prior to the onset of stroke, however therapies could be administered following

Circulating E-selectin and IL-6 measurements Soluble E-selectin (sEsel), and IL-6 enzyme-linked immunosorbent assays (ELISA) were performed according to the manufacturer’s instructions (R&D Systems, Inc.; Minneapolis MN, USA; Cat#: MES00 & MPS00), n=3-5. Blood for ELISA was collected via cardiac puncture at the time of sacrifice and plasma prepared by centrifugation at 8,500 rpm for 10 min.

Drug treatment Anakinra (Swedish Orphan Biovitrum AB, Stockholm, Sweden) (100 mg/kg via one intraperitoneal injection [i.p.]) or vehicle control (phosphate buffered saline) was administered 1 hour following induction of stroke (n=4 per group).

Statistical analysis Data are presented as mean ± standard deviation. Analysis was carried out using GraphPad Prism and tests for normality were performed using the Shapiro-Wilk test. Differences between groups were then analyzed using a one way ANOVA or an unpaired t-test for comparison between groups or Mann Whitney U test. Probability values of P < 0.05 were considered statistically significant.

Data sharing For original data, please contact deitzman@umich.edu.

haematologica | 2021; 106(9)

2471

J. Venugopal et al.

stroke onset. In order to determine whether stroke size could be reduced by pharmacologic blockade of the IL-1R, even when administered following MCA occlusion, the IL1R antagonist, anakinra, was administered 1 hour following stroke induction in Wt,SCDbmt mice. Compared to vehicle control, mice given anakinra after stroke onset experienced reduction in stroke volume when analyzed 3 days following MCA occlusion (Figure 5). A decrease in peri-infarct MAC3-positive cells was also observed (Figure 6), similar to what was seen in IL1R-/-,SCDbmt brain sections post-stroke.

Discussion SCD results from a missense mutation leading to an amino acid substitution in the β-globin gene.48 Although SCD is a monogenic disease, there is marked phenotypic heterogeneity in patients with SCD that applies to anemia,

cerebrovascular disease, acute chest syndrome, pain crises, and death.49,50 Differential activation of inflammatory pathways may be a mechanism which accounts for the observed phenotypic heterogeneity and may be a critical link between hemolysis and subsequent vascular complications.10 Multiple cytokines, including IL-1β and IL-6, have been postulated to play a role in SCD phenotypes, and as these cytokines are also known to regulate stroke volume in non-SCD populations, the hypothesis that the deletion of the endothelial IL-1R pool may be beneficial in SCD was pursued by utilizing an MCA occlusion model. This model of stroke leads to sustained occlusion of the MCA in the absence of treatment51 and the increased stroke size in sickle cell mice is likely due to vaso-occlusion in the penumbra microvasculature.42 Since potential deleterious effects of IL1R signaling in SCD could be mediated via downstream production of IL-6, this pathway was also studied.

Figure 1. Blood parameter analysis at 15 weeks post-bone marrow transplantation. Circulating erythrocytes (A), hemoglobin (B), hematocrit (C), reticulocytes (D), and leukocytes (E) of Wt,WTbmt, Wt,SCDbmt, IL1R-/-,SCDbmt and IL6-/-,SCDbmt mice (mean ± standard deviation). Total WBC: total white blood cells; NE: neutrophils; LY: lymphocytes; MO: monocytes, *P=< 0.05, **P<0.01, ***P<0.005 as determined by ANOVA. In Figure 1E, asterisks indicates significance to Wt,WTbmt and pound signs indicate significance to Wt,SCDbmt.

2472

haematologica | 2021; 106(9)

Interleukin-1 receptor in sickle cell disease

Results from this study support a critical role of nonhematopoietic IL-1R signaling in mediating acute brain tissue damage in SCD mice in to the setting of ischemic stroke. This effect was associated with IL-1R-mediated regulation of endothelial adhesive properties. Signaling via the endothelial IL-1R leads to upregulation of endothelial adhesion molecules with resultant increases in leukocyteendothelial interactions and tissue leukocyte infiltration.40 This signaling pathway involves enhanced NFκB signaling.40 Enhanced endothelial expression of adhesion molecules is detrimental in SCD, promoting vascular occlu-

sions and pain crises.52 Regulation of endothelial IL-1R responses to IL-1β has also been shown to occur indirectly by leukocyte interactions with selectins. For example, mice with leukocyte deficiency of p-selectin glycoprotein ligand-1 (Psgl-1) are resistant to IL-1β-mediated stimulation of endothelial adhesion molecule expression and show reduced leukocyte-endothelial interactions.40 This is potentially relevant to SCD as an antibody to p-selectin, crizanlizumab, has been shown in human clinical trials to reduce the frequency of vaso-occlusive events.53 Preclincal studies have also shown that SCD mice treated with an

Figure 2. Stroke area following middle cerebral artery occlusion. Representative brain sections stained with 4% 2,3,5-triphenyltetrazolium chloride (TTC) to assess stroke size (in white) of (A) Wt,WTbmt, (B) Wt,SCDbmt, (C) IL1R-/-,SCDbmt mice, and (D) IL6-/-,SCDbmt mice. (E) Quantification of stroke volume (mean ± standard deviation). The brain sections were imaged with a Nikon SMZ-2T microscope and Spot Idea camera model 29.2-13MP using at Nikon 0.45x TV lens and Spot 5.1 software. *P<0.05 as determined by ANOVA.

Figure 3. Post-stroke macrophage infiltration. Representative images of MAC3-positive cells in the peri-infact area of Wt,WTbmt, Wt,SCDbmt, IL1R-/-,SCDbmt and IL6-/-,SCDbmt mouse brains, and quantification (mean ± standard deviation). A Nikon SE upright microscope and a Nikon DS-Fi3 camera was used to capture 10x and 20x images of ipsilateral brain and 10x images of contralateral brain. Dotted line denotes transition between infarcted area and heathy tissue. *P<0.05 as determined by ANOVA.

haematologica | 2021; 106(9)

2473

J. Venugopal et al.

Figure 4. Blood brain barrier integrity poststroke. Representative staining immunopositive to Fibrinogen in the peri-infact area of Wt,WTbmt, Wt,SCDbmt, IL1R-/-,SCDbmt and IL6-/-,SCDbmt mouse brains, and quantification (mean ± standard deviation). A Nikon SE upright microscope and a Nikon DS-Fi3 camera was used to capture 10x and 20x images of ipsilateral brain and 10x images of contralateral brain. Dotted line denotes transition between infarcted area and heathy tissue. *P<0.05, **P<0.01, ***P<0.005 as determined by ANOVA.

Figure 5. Infarct area in response to anakinra. Representative brain sections stained with 4% 2,3,5-triphenyltetrazolium chloride (TTC) in Wt,SCDbmt given one intraperitoneal injection of phosphate buffered saline (PBS) (A) or anakinra (B) immediately post-stroke induction, and quantification of stroke volume (mean ± standard deviation) (C). The brain sections were imaged with a Nikon SMZ-2T microscope and Spot Idea camera model 29.2-13MP using at Nikon 0.45x TV lens and Spot 5.1 software. *P<0.05 as determined by student’s t-test.

antibody to Psgl-1 displayed reduced leukocyte-endothelial interactions and reduced levels of circulating selectins.54 Circulating monocytes express Psgl-1 and use this receptor to engage E-selectin when undergoing extravasion from the vasculature to the damaged tissue.55 In agreement with the known role of IL-1β in endothelial activation,4,54 IL1R-/-,SCDbmt mice had lower circulating levels of sEsel post-stroke than Wt,SCDbmt or IL6-/-,SCDbmt mice. The lower sEsel concentrations of IL1R-/-,SCDbmt mice correlated with decreased MAC3-positive macrophages present in the peri-infarct area. A similar decrease in MAC3-positive cells was also attained with post-stroke administration of anakinra. ICAM-1 may also contribute to the leukocyte infiltration as immuno-stain2474

ing for ICAM-1 was also reduced in IL1R-/-,SCDbmt mice compared to Wt,SCDbmt mice in the peri-infarct region. This is consistent with a previous in vitro study in which ICAM-1 was upregulated on endothelial cells following exposure to sickled erythrocytes and was further increased in the presence of IL-1β.56 IL-1β has also been shown to reduce BBB integrity.36 A decrease in infarct-associated fibrin(ogen) immunostaining, was observed to be significantly decreased in IL1R-/-,SCDbmt mice relative to both Wt,SCDbmt and IL6-/-,SCDbmt mice, suggesting less BBB disruption. As disruption of the BBB can lead to a greater influx of leukocytes,57 inhibition of IL-1β-mediated actions on endothelial cells may reduce leukocyte accumulation by both haematologica | 2021; 106(9)

Interleukin-1 receptor in sickle cell disease

Figure 6. Post-stroke macrophage infiltration in response to anakinra. Representative staining immunopositive to MAC3 in the peri-infact area of Wt,SCDbmt given one intraperitoneal injection of phosphate buffered saline (PBS) or anakinra immediately post-stroke induction, and quantification of MAC3-positive cells (mean ± standard deviation). A Nikon SE upright microscope and a Nikon DS-Fi3 camera was used to capture 10x and 20x images of ipsilateral brain and 10x images of contralateral brain. Dotted line denotes transition between infarcted area and heathy tissue. **P<0.01 as determined by student’s t-test.

decreasing adhesion molecules such as E-selectin and ICAM-1,56 and also by preservation of BBB integrity. Although some studies have shown BBB integrity may be preserved by the action of macrophages,35 this study demonstrated that macrophages recruited acutely through IL-1R pathways may be deleterious. Long term human studies targeting IL-1β in SCD will be informative. A clinical study with the IL-1β antagonist, canakinumab, is currently underway to determine safety and efficacy of IL-1β inhibition in SCD patients (clincialtrials gov. Identifier: NCT02961218). The beneficial effects of IL-1R inhibition observed in this study are independent of non-hematopoietic IL-6. In response to ischemia, neurons and other cell types in the brain produce IL-6,58,59 and circulating IL-6 concentrations have been positively associated both with stroke size in patients,60 and with worsening outcomes within 48 hours post-stroke.61 While IL-6 may adversely affect stroke acutely.23,59 IL-6 may show beneficial effects towards resolution of stroke damage.62 The assessment of stroke volume 72 hours after stroke induction in this study may have been too early to observe the full effects of IL-6 towards stroke repair. However, the failure of IL-6-/-,SCDbmt mice to phenocopy IL1R-/-,SCDbmt mice in relation to stroke size illustrates the lack of dependence on the downstream induction of IL-6 toward the acute detrimental action of IL-1R signaling. While chronic treatment of SCD patients with anticytokine therapies may increase susceptibility to infections, short term treatment could be administered to patients presenting with acute complications. Remarkably, the IL-1R antagonist, anakinra, was beneficial even when administered following onset of MCA occlusion. These findings suggest that targeting the IL-1R may be beneficial in SCD patients presenting with stroke or other vascular complications. Limitations of this work include the use of BMT to generate chimeric mice. It is possible the irradiation procedure used to ablate the bone marrow could have affected the vascular phenotypes, however, this strategy has been widely employed and greatly facilitates the generation of chimeric SCD mice.63-65 It is also possible that other beneficial effects of chimeric IL-1R deficiency in this model are

haematologica | 2021; 106(9)

at play. Although there were no differences in anemia between different recipient transgenic mice receiving sickle cell marrow, we cannot rule out differences in engraftment related to recipient IL-1R status. Future experiments with mice generated by crossbreedings to produce complete and tissue-specific gene deletions will be useful to confirm and expand these studies. The stroke model used in this study is induced by MCA thrombotic occlusion. A more relevant model would include spontaneous stroke due to vasoocclusion triggered by sickled erythrocytes. However, well validated models of spontaneous stroke are not available in SCD mice to our knowledge. Additionally, we cannot rule out differences in blood flow following MCA occlusion due to difference in IL-1R signaling. These studies would require a time course analysis after stroke induction in addition to laser doppler imaging. While there is no standard drug for treatment of acute stroke associated with sickle cell disease, future experiments comparing anakinra with tissue plasminogen activator or emergent blood transfusion would be interesting. Finally, longer periods between stroke induction and stroke volume measurement may be informative. In conclusion, non-hematopoietic deficiency of the IL-1R is sufficient to reduce stroke size in SCD. Therapies targeting this pathway may be beneficial towards the treatment of stroke and possibly other complications of SCD. Disclosures No conflicts of interest to disclose.. Contributions DTE contributed to the conception and experimental design of this work; JV, JW, JM and CG contributed to the acquisition of data; JV and JM were responsible for data analysis; DTE and JV contributed to the interpretation of the data, drafting and revision of the manuscript; DTE approved the final version of the work and agrees to be accountable for aspects of the work. Funding This work was supported by the National Institutes of Health (T32-HL007853 to JV. and a VA Merit Award (BX002776 to DTE).

2475

J. Venugopal et al.

References 1. Fitzhugh CD, Lauder N, Jonassaint JC, et al. Cardiopulmonary complications leading to premature deaths in adult patients with sickle cell disease. Am J Hematol. 2010; 85(1):36-40. 2. Ohene-Frempong K, Weiner SJ, Sleeper LA, et al. Cerebrovascular accidents in sickle cell disease: rates and risk factors. Blood. 1998;91(1):288-294. 3. Strouse JJ, Lanzkron S, Urrutia V. The epidemiology, evaluation and treatment of stroke in adults with sickle cell disease. Expert Rev Hematol. 2011;4(6):597-606. 4. Webb J, Kwiatkowski JL. Stroke in patients with sickle cell disease. Expert Rev Hematol. 2013;6(3):301-316. 5. Sumoza A, de Bisotti R, Sumoza D, Fairbanks V. Hydroxyurea (HU) for prevention of recurrent stroke in sickle cell anemia (SCA). Am J Hematol. 2002;71(3):161-165. 6. Ware RE, Mueller BU. The second century of sickle cell disease. Pediatr Blood Cancer. 2012;59(2):352. 7. Hasson C, Veling L, Rico J, Mhaskar R. The role of hydroxyurea to prevent silent stroke in sickle cell disease: systematic review and meta-analysis. Medicine (Baltimore). 2019;98(51):e18225. 8. Teixeira RS, Terse-Ramos R, Ferreira TA, et al. Associations between endothelial dysfunction and clinical and laboratory parameters in children and adolescents with sickle cell anemia. PLoS One. 2017;12(9): e0184076. 9. Zhou Z, Yee DL, Guchhait P. Molecular link between intravascular hemolysis and vascular occlusion in sickle cell disease. Curr Vasc Pharmacol. 2012;10(6):756-761. 10. Conran N, Belcher JD. Inflammation in sickle cell disease. Clin Hemorheol Microcirc. 2018;68(2-3):263-299. 11. Kaul DK, Finnegan E, Barabino GA. Sickle red cell-endothelium interactions. Microcirculation. 2009;16(1):97-111. 12. Vilas-Boas W, Cerqueira BA, Zanette AM, Reis MG, Barral-Netto M, Goncalves MS. Arginase levels and their association with Th17-related cytokines, soluble adhesion molecules (sICAM-1 and sVCAM-1) and hemolysis markers among steady-state sickle cell anemia patients. Ann Hematol. 2010;89(9):877-882. 13. Zennadi R, Chien A, Xu K, Batchvarova M, Telen MJ. Sickle red cells induce adhesion of lymphocytes and monocytes to endothelium. Blood. 2008;112(8):3474-3483. 14. Pitanga TN, Oliveira RR, Zanette DL, et al. Sickle red cells as danger signals on proinflammatory gene expression, leukotriene B4 and interleukin-1 beta production in peripheral blood mononuclear cell. Cytokine. 2016;83:75-84. 15. Platt OS. Hydroxyurea for the treatment of sickle cell anemia. N Engl J Med. 2008; 358(13):1362-1369. 16. Steinberg MH, McCarthy WF, Castro O, et al. The risks and benefits of long-term use of hydroxyurea in sickle cell anemia: A 17.5 year follow-up. Am J Hematol. 2010; 85(6):403-408. 17. Dutra FF, Alves LS, Rodrigues D, et al. Hemolysis-induced lethality involves inflammasome activation by heme. Proc Natl Acad Sci U S A. 2014;111(39):E4110E4118. 18. Wun T, Cordoba M, Rangaswami A, Cheung AW, Paglieroni T. Activated monocytes and platelet-monocyte aggregates in

2476

patients with sickle cell disease. Clin Lab Haematol. 2002;24(2):81-88. 19. Singhal R, Chawla S, Rathore DK, et al. Development of pro-inflammatory phenotype in monocytes after engulfing Hb-activated platelets in hemolytic disorders. Clin Immunol. 2017;175:133-142. 20. Kalliolias GD, Ivashkiv LB. Targeting cytokines in inflammatory diseases: focus on interleukin-1-mediated autoinflammation. F1000 Biol Rep. 2009;1:70. 21. Dinarello CA, Simon A, van der Meer JW. Treating inflammation by blocking interleukin-1 in a broad spectrum of diseases. Nat Rev Drug Discov. 2012;11(8):633-652. 22. Ridker PM, Everett BM, Thuren T, et al. Antiinflammatory therapy with canakinumab for atherosclerotic disease. N Engl J Med. 2017;377(12):1119-1131. 23. Huang CM, Tsai FJ, Wu JY, Wu MC. Interleukin-1beta and interleukin-1 receptor antagonist gene polymorphisms in rheumatoid arthritis. Scand J Rheumatol. 2001;30(4):225-228. 24. Vicari P, Adegoke SA, Mazzotti DR, Cancado RD, Nogutti MA, Figueiredo MS. Interleukin-1beta and interleukin-6 gene polymorphisms are associated with manifestations of sickle cell anemia. Blood Cells Mol Dis. 2015;54(3):244-249. 25. Denes A, Pinteaux E, Rothwell NJ, Allan SM. Interleukin-1 and stroke: biomarker, harbinger of damage, and therapeutic target. Cerebrovasc Dis. 2011;32(6):517-527. 26. Sobowale OA, Parry-Jones AR, Smith CJ, Tyrrell PJ, Rothwell NJ, Allan SM. Interleukin-1 in stroke: from bench to bedside. Stroke. 2016;47(8):2160-2167. 27. McCann SK, Cramond F, Macleod MR, Sena ES. Systematic review and metaanalysis of the efficacy of interleukin-1 receptor antagonist in animal models of stroke: an update. Transl Stroke Res. 2016; 7(5):395-406. 28. Parker LC, Luheshi GN, Rothwell NJ, Pinteaux E. IL-1 beta signalling in glial cells in wildtype and IL-1RI deficient mice. Br J Pharmacol. 2002;136(2):312-320. 29. Thornton P, Pinteaux E, Allan SM, Rothwell NJ. Matrix metalloproteinase-9 and urokinase plasminogen activator mediate interleukin-1-induced neurotoxicity. Mol Cell Neurosci. 2008;37(1):135-142. 30. Giles JA, Greenhalgh AD, Denes A, et al. Interleukin-1 and platelets as key drivers of cerebrovascular inflammation. Immunology. 2011;135:154. 31. Ching S, Zhang H, Belevych N, et al. Endothelial-specific knockdown of interleukin-1 (IL-1) type 1 receptor differentially alters CNS responses to IL-1 depending on its route of administration. J Neurosci. 2007;27(39):10476-10486. 32. Bolton SJ, Anthony DC, Perry VH. Loss of the tight junction proteins occludin and zonula occludens-1 from cerebral vascular endothelium during neutrophil-induced blood-brain barrier breakdown in vivo. Neuroscience. 1998;86(4):1245-1257. 33. Basu A, Krady JK, Levison SW. Interleukin1: a master regulator of neuroinflammation. J Neurosci Res. 2004;78(2):151-156. 34. Allen C, Thornton P, Denes A, et al. Neutrophil cerebrovascular transmigration triggers rapid neurotoxicity through release of proteases associated with decondensed DNA. J Immunol. 2012;189(1):381-392. 35. Li Y, Zhu ZY, Huang TT, et al. The peripheral immune response after stroke-A double edge sword for blood-brain barrier integrity. CNS Neurosci Ther. 2018; 24(12):

1115-1128. 36. Fahey E, Doyle SL. IL-1 Family cytokine regulation of vascular permeability and angiogenesis. Front Immunol. 2019; 10:1426. 37. Paszty C, Brion CM, Manci E, et al. Transgenic knockout mice with exclusively human sickle hemoglobin and sickle cell disease. Science. 1997;278(5339):876-878. 38. Ryan TM, Ciavatta DJ, Townes TM. Knockout-transgenic mouse model of sickle cell disease. Science. 1997;278(5339):873876. 39. Wu LC, Sun CW, Ryan TM, Pawlik KM, Ren J, Townes TM. Correction of sickle cell disease by homologous recombination in embryonic stem cells. Blood. 2006; 108(4):1183-1188. 40. Russo HM, Wickenheiser KJ, Luo W, et al. P-selectin glycoprotein ligand-1 regulates adhesive properties of the endothelium and leukocyte trafficking into adipose tissue. Circ Res. 2010;107(3):388-397. 41. Wang J, Tran J, Wang H, et al. mTOR Inhibition improves anaemia and reduces organ damage in a murine model of sickle cell disease. Br J Haematol. 2016; 174(3):461-469. 42. Luo W, Su EJ, Wang J, et al. Increased stroke size following MCA occlusion in a mouse model of sickle cell disease. Blood. 2014; 123(12):1965-1967. 43. Campbell AD, Cui S, Shi L, et al. Forced TR2/TR4 expression in sickle cell disease mice confers enhanced fetal hemoglobin synthesis and alleviated disease phenotypes. Proc Natl Acad Sci U S A. 2011; 108(46):18808-18813. 44. Ching S, He L, Lai W, Quan N. IL-1 type I receptor plays a key role in mediating the recruitment of leukocytes into the central nervous system. Brain Behav Immun. 2005;19(2):127-137. 45. Akassoglou K, Strickland S. Nervous system pathology: the fibrin perspective. Biol Chem. 2002;383(1):37-45. 46. Adams RA, Passino M, Sachs BD, Nuriel T, Akassoglou K. Fibrin mechanisms and functions in nervous system pathology. Mol Interv. 2004;4(3):163-176. 47. Petersen MA, Ryu JK, Akassoglou K. Fibrinogen in neurological diseases: mechanisms, imaging and therapeutics. Nat Rev Neurosci. 2018;19(5):283-301. 48. Bunn HF. Pathogenesis and treatment of sickle cell disease. N Engl J Med. 1997; 337(11):762-769. 49. Steinberg MH. Predicting clinical severity in sickle cell anaemia. Br J Haematol. 2005; 129(4):465-481. 50. Steinberg MH, Rodgers GP. Pathophysiology of sickle cell disease: role of cellular and genetic modifiers. Semin Hematol. 2001;38(4):299-306. 51. Su EJ, Geyer M, Wahl M, et al. The thrombomodulin analog Solulin promotes reperfusion and reduces infarct volume in a thrombotic stroke model. J Thromb Haemost. 2011;9(6):1174-1182. 52. Telen MJ. Role of adhesion molecules and vascular endothelium in the pathogenesis of sickle cell disease. Hematology Am Soc Hematol Educ Program. 2007:84-90. 53. Ataga KI, Kutlar A, Kanter J, et al. Crizanlizumab for the prevention of pain crises in sickle cell disease. N Engl J Med. 2017;376(5):429-439. 54. Luo W, Campbell A, Wang H, et al. Pselectin glycoprotein ligand-1 inhibition blocks increased leukocyte-endothelial interactions associated with sickle cell dis-

haematologica | 2021; 106(9)

Interleukin-1 receptor in sickle cell disease

ease in mice. Blood. 2012;120(18):38623864. 55. An G, Wang H, Tang R, et al. P-selectin glycoprotein ligand-1 is highly expressed on Ly-6Chi monocytes and a major determinant for Ly-6Chi monocyte recruitment to sites of atherosclerosis in mice. Circulation. 2008;117(25):3227-3237. 56. Shiu YT, Udden MM, McIntire LV. Perfusion with sickle erythrocytes up-regulates ICAM-1 and VCAM-1 gene expression in cultured human endothelial cells. Blood. 2000;95(10):3232-3241. 57. Obermeier B, Daneman R, Ransohoff RM. Development, maintenance and disruption of the blood-brain barrier. Nat Med. 2013; 19(12):1584-1596. 58. Schwaninger M, Neher M, Viegas E, Schneider A, Spranger M. Stimulation of

haematologica | 2021; 106(9)

interleukin-6 secretion and gene transcription in primary astrocytes by adenosine. J Neurochem. 1997;69(3):1145-1150. 59. Suzuki S, Tanaka K, Suzuki N. Ambivalent aspects of interleukin-6 in cerebral ischemia: inflammatory versus neurotrophic aspects. J Cereb Blood Flow Metab. 2009;29(3):464-479. 60. Hotter B, Hoffmann S, Ulm L, Meisel C, Fiebach JB, Meisel A. IL-6 Plasma levels correlate with cerebral perfusion deficits and infarct sizes in stroke patients without associated infections. Front Neurol. 2019; 10:83. 61. Vila N, Reverter JC, Yague J, Chamorro A. Interaction between interleukin-6 and the natural anticoagulant system in acute stroke. J Interferon Cytokine Res. 2000; 20(3):325-329.

62. Doll DN, Barr TL, Simpkins JW. Cytokines: their role in stroke and potential use as biomarkers and therapeutic targets. Aging Dis. 2014;5(5):294-306. 63. Wood KC, Hebbel RP, Granger DN. Endothelial cell NADPH oxidase mediates the cerebral microvascular dysfunction in sickle cell transgenic mice. FASEB J. 2005;19(8):989-991. 64. Ghosh S, Adisa OA, Chappa P, et al. Extracellular hemin crisis triggers acute chest syndrome in sickle mice. J Clin Invest. 2013;123(11):4809-4820. 65. Chang KH, Nayak RC, Roy S, et al. Vasculopathy-associated hyperangiotensinemia mobilizes haematopoietic stem cells/progenitors through endothelial AT(2)R and cytoskeletal dysregulation. Nat Commun. 2015;6:5914.

2477

ARTICLE Ferrata Storti Foundation

Red Cell Biology & its Disorders

Oxidative stress activates red cell adhesion to laminin in sickle cell disease Maria Alejandra Lizarralde-Iragorri,1,2,3* Sophie D. Lefevre,1,2,3* Sylvie Cochet,1,2,3 Sara El Hoss,1,2,3 Valentine Brousse,1,2,3,4 Anne Filipe,1,2,3.5 Michael Dussiot,6 Slim Azouzi,1,2,3 Caroline Le Van Kim,1,2,3 Fernando Rodrigues-Lima,5 Olivier Français,7 Bruno Le Pioufle,8 Thomas Klei,9 Robin van Bruggen9 and Wassim El Nemer1,2,3

Haematologica 2021 Volume 106(9):2478-2488

*MALI and SDL contributed equally as co-first authors. 1 Université de Paris, UMR S1134, BIGR, INSERM, Paris, France; 2Institut National de la Transfusion Sanguine, Paris, France; 3Laboratoire d’Excellence GR-Ex, Paris, France; 4 Service de Pédiatrie Générale et Maladies Infectieuses, Hôpital Universitaire Necker Enfants Malades, Paris, France; 5Université de Paris, BFA, UMR 8251, CNRS, Paris, France; 6Institut Imagine, INSERM U1163, CNRS UMR8254, Université Paris Descartes, Hôpital Necker Enfants Malades, Paris, France; 7ESYCOM, Université Gustave Eiffel, CNRS UMR 9007, ESIEE Paris, Marne-la-Vallee, France; 8Université Paris-Saclay, ENS Paris-Saclay, CNRS Institut d'Alembert, LUMIN, Gif sur Yvette, France and 9Department of Blood Cell Research, Sanquin Research and Lab Services and Landsteiner Laboratory, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands

ABSTRACT

V Correspondence: WASSIM EL NEMER wassim.el-nemer@efs.sante.fr Received: June 2, 2020. Accepted: August 12, 2020. Pre-published: August 27, 2020.

aso-occlusive crises are the hallmark of sickle cell disease (SCD). They are believed to occur in two steps, starting with adhesion of deformable low-dense red blood cells (RBC), or other blood cells such as neutrophils, to the wall of post-capillary venules, followed by trapping of denser RBC or leukocytes in the areas of adhesion because of reduced effective lumen-diameter. In SCD, RBC are heterogeneous in terms of density, shape, deformability and surface proteins, which accounts for the differences observed in their adhesion and resistance to shear stress. Sickle RBC exhibit abnormal adhesion to laminin mediated by Lu/BCAM protein at their surface. This adhesion is triggered by Lu/BCAM phosphorylation in reticulocytes but such phosphorylation does not occur in mature dense RBC despite firm adhesion to laminin. In this study, we investigated the adhesive properties of sickle RBC subpopulations and addressed the molecular mechanism responsible for the increased adhesion of dense RBC to laminin in the absence of Lu/BCAM phosphorylation. We provide evidence for the implication of oxidative stress in post-translational modifications of Lu/BCAM that impact its distribution and cis-interaction with glycophorin C at the cell surface activating its adhesive function in sickle dense RBC.

Introduction https://doi.org/10.3324/haematol.2020.261586

2478

Sickle cell disease (SCD) is an autosomal recessive disorder caused by a single mutation in the sixth codon of the β-globin gene resulting in the expression of an abnormal hemoglobin that polymerizes under hypoxic conditions driving red blood cell (RBC) sickling.1 SCD is a multisystem disease characterized by hemolytic anemia, recurrent painful vaso-occlusive crises (VOC), stroke, acute chest syndrome, organ failure and high susceptibility to infections.2,3 On the cellular level, SCD is characterized by dehydration and RBC sickling, which decrease cell deformability and increase rigidity resulting in altered blood rheology and microcirculatory flow.2-6 In addition, RBC are known to be highly adhesive in SCD.7-9 This abnormal adhesion to the endothelium is a contributing factor of the VOC and is believed to be triggered by signaling cascades that activate adhesion proteins at the red cell surface.10 A two-step model, based on in vivo vaso-occlusion observations in SCD mouse models, postulates that adhesion of deformable low-dense RBC and stress reticulocytes reduces effective lumen-diameter of post-capillary venules promoting selective trapping of the denser and misshapen RBC in the areas

haematologica | 2021; 106(9)

Activation of SCD RBC adhesion by oxidative stress

of adhesion. However, random precapillary obstruction by a small number of dense RBC also contributes to VOC as well as the entrapment of leukocytes and platelets.11-14 Sickle RBC are very heterogeneous in terms of age, shape and surface proteins. These variabilities account for the differences observed in cell adhesion and resistance to shear stress under flow conditions.11 In SCD, among other proteins and mechanisms, adhesion proteins LW/ICAM-4 (Landsteiner-Wiener/intercellular adhesion molecule-4) and Lu/BCAM (Lutheran/basal cell adhesion molecule) are abnormally activated and believed to prime adhesion of RBC to endothelial cells and/or subendothelial matrix proteins exposed to the bloodstream following vascular damage, contributing to microvasculature blockade.10,15-21 Lu/BCAM is an adhesion molecule with wide tissue distribution.22,23 Lu/BCAM-mediated cell adhesion to laminin can be triggered either by the phosphorylation of its serine 62117,24 or by the dissociation of its cytoplasmic domain from the spectrin-based skeleton.25,26 In SCD, phosphorylation of Lu/BCAM was shown to occur in low-density (LD) RBC,27 mainly reticulocytes, consistent with the adhesion of these cells to laminin.27,28 However, despite firm adhesion to laminin of high-density (HD) RBC, Lu/BCAM phosphorylation is very minor in this subpopulation and these cells do not respond to cAMP inducers such as forskolin.28 The mechanism underlying this increased adhesion is still unknown. In this study, we investigated the molecular mechanism responsible for the increased adhesion of sickle HD RBC to laminin. We provide evidence for the implication of oxidative stress in post-translational modifications of Lu/BCAM that impact its distribution and cis-interactions at the cell surface and activate its adhesive function.

Methods

of 10, 8, 7, 6 or 5 μm (Figure 1A) or four parallel rows with slits of 5, 4, 3 or 2 μm (Online Supplementary Figure S1). Side flow is rendered possible in the device, the U form filter zone comprises pillars with a 5 μm gap between them. In order to reduce the hydraulic resistance of the full design, the microchannel network is 25 μm-high compared to the 5 μm height of each filtering unit (Figure 1A). The microfluidic device was made of polydimethylsiloxane (PDMS, Sylgard), a silicone elastomer,29 using standard microfabrication and molding. The mold was fabricated by the micro-patterning of two successive SU8 photoresist layers (Microchem, Newton, MA) to obtain a two-levels negative mold on a 4-inch Silicon substrate. The SU8 layers thicknesses were 5 μm and 25 μm, corresponding respectively to the height of the filtering units and the microchannels network. A mixture of PDMS and curing agent was poured on the SU8 mold, and reticulated at 75°C for 2 hours. Access through-holes were then punched, using biopsy punchers (diameter of 1.5 mm). The PDMS device, with open channels formed on one of its sides, was then assembled to a microscope coverslip, using O2 plasma activation (30 W, 300 mT, 20 s) to achieve a covalent bonding. Luer (TM) connectors were then inserted at the inlet and outlet of this microfluidic device, to achieve the sample injection with the flow controller. For each assay, 10 μL of RBC pellet were stained with either PKH67 fluorescent Cell Linker Kit (green) or PKH26 fluorescent Cell Linker Kit (red) according to the manufacturer’s instructions (Sigma Aldrich). A 1% hematocrit solution in CellStab containing equal concentration of green and red stained RBC were loaded in the input well of the chip and perfused at constant pressure (250 mBar) using an MFCS™-EZ-1C pump (Fluigent). RBC trapping within each filtering unit was monitored over time by sequential fluorescence images acquired using an inverted AxioObserver Z1 microscope coupled with a high resolution AxioCam MRm Rev.3 camera (Carl Zeiss). Green and red fluorescent RBC were visualized using the 470 and 555 nm Colibri LED (Carl Zeiss), respectively. Images were then analyzed using ImageJ software.30

Patients

Flow adhesion assays and red blood cell counting

The study was conducted in accordance with the Declaration of Helsinki and was approved by the Regional Ethics Committee (n°3215 CPP Ile de France III). Blood samples were recovered from blood tubes drawn for medical care at Necker Hospital (Paris) after written informed consent. Blood samples were collected on ethylenediaminetetraacetic acid (EDTA) from a total of 39 patients with sickle cell anemia (SS and Sβ° genotypes) (females and males; median age: 8 years [Min-Max, 2-53 years]), and from 26 healthy donors (age range, 18-70, as per Etablissement Français du Sang [EFS] criteria). Sickle patients were not on a regular transfusion program nor under hydroxyurea (HU) treatment. All experiments were performed with fresh blood samples, within 2 hours after blood was drawn.

RBC adhesion to Laminin 521 was evaluated under flow conditions using capillary flow chambers. Recombinant human Laminin 521 (BioLamina, Sundbyberg, Sweden) at 5 ng/μL was immobilized in Vena8 Endothelial+TM biochips (internal channel dimensions: length 20 mm, width 0.8 mm, height 0.12 mm). RBC were perfused at 5.107 RBC/mL for 10 min at 0.5 dyn/cm2 and 6 min washouts were performed at 1, 2, 3, 4, 5, 6, 7, 8, 9 and 10 dyn/cm2 using the ExiGoTM pump (Cellix Ltd, Dublin, Ireland). After each wash, images of adherent RBC were taken using the AxioObserver Z1 microscope (10x objective) (Carl Zeiss, Le Pecq, France). Adherent RBC were counted on each field using Image J. The number of immobile RBC was assessed by using the Image Calculator option of the Image J software.30 The picture of one area at 2 dyn/cm² was combined to the picture of the same area taken at 3 dyn/cm². On the newly created image, immobile cells appeared in dark grey whereas cells present on only one of the two combined images appeared in light grey. Dark grey objects were counted with Image J software after setting an appropriate threshold.

Percoll fractionation RBC subpopulations were obtained from sickle whole blood fractionation as previously described,27 using a Percoll triple-density fractionation (densities: 1.076, 1.096, and 1.11). Three different density layers were obtained and collected as follows: LD (low density, rich in reticulocytes), D (dense), and HD (high density, rich in irreversibly sickled cells).

Microfluidic assays The microfluidic filtering design is based on eight mechanical filtering units associated in parallel and connected together with a microchannel network. Each filtering unit is composed of five parallel rows comprising pillars of 15 μm diameter separated by slits

haematologica | 2021; 106(9)

Control blood sample oxidation Control RBC were washed with phospate buffered saline (PBS) 1X (Thermo Fisher), suspended at 20% hematocrit in either PBS or 270 μM cumene hydroperoxide (SIGMA-ALDRICH) and incubated for 2 hours with constant mild shaking. After incubation, RBC suspensions were centrifuged, the supernatant was discarded and the RBC pellets were used for subsequent experiments.

2479

M.A. Lizarralde-Iragorri et al.

Confocal microscopy Imaging was performed on the Confocal LSM 510 META-TIRF (Zeiss, Oberkochen, Germany). LASX software was used to set up and analyze the experiments (Leica microsystems, Wetzlar, Germany).

Imaging flow cytometry assays Expression of Lu/BCAM on the RBC surface was analyzed using F241 mouse monoclonal antibody. After 1-hour incubation

with F241 (dilution [d]: 1/10), the secondary anti-mouse APC-conjugated antibody (d: 1/100) (Beckman Coulter) was added for 1 hour, then RBC were washed and suspended in 200 µL of thiazole orange (TO) dye (Retic-CountTM, Becton-Dickinson) for 30 minutes (min) to label reticulocytes. RBC were analyzed using ImageStream®X Mark II Imaging Flow Cytometer (Merck Millipore) (60x magnification) and the IDEAS software (version 6.2). Lu/BCAM-positive mature RBC (Lu APC) were gated, excluding the reticulocytes (TO-positive events). Using the

Figure 1. Analysis of red blood cell deformability using a microfluidic biomimetic chip. (A) Left panel: the microfluidic device comprises eight filtering units arranged in parallel. Right panel: each filtering unit is 5 μm-high and has a U shape composed of a series of 15 μm pillars separated by 5 μm slits, with two 10 μm-wide side channels. Inside the U shape, four rows are disposed in parallel with decreasing slit width (10, 8, 7 and 6 μm). (B) Microscopy image showing SS red blood cells (RBC) (green) and AA RBC (red) trapped into the filtering unit slits. (C) Retention percentage of AA and SS RBC in the 5 μm slits. Mann-Whitney test, ***P<0.0001 (D) Microscopy image showing low-density (LD) RBC (green) and high-density (HD) RBC (red) trapped into the peripheral 5 μm slits of the filtering unit. The majority of the other cells are in motion in the space separating two consecutive walls (E) Retention percentage of LD and HD RBC in the peripheral 5 μm slits. Wilcoxon test, *P<0.05. In the graphs C and E the data is expressed as the percentage of cells from each RBC type trapped into the 5 μm slits, considering the total number of cells trapped into the 5 μm slits as 100% (n=7).

2480

haematologica | 2021; 106(9)

Activation of SCD RBC adhesion by oxidative stress

C D

Figure 2. Legend on following page.

haematologica | 2021; 106(9)

2481

M.A. Lizarralde-Iragorri et al.

Figure 2. Adhesion to laminin and shear stress resistance of SS red blood cell subpopulations. (A). Adhesion assays under flow conditions were done by perfusing low-density (LD) and high-density (HD) red blood cells (RBC) from seven sickle cell disease (SCD) patients through a biochip coated with Laminin 521. The amount of RBC, of each subpopulation and each blood sample, attached at 1 dyn/cm2 was considered as 100% and used to determine the percentage of attached RBC at increasing shear stress, as indicated on the x-axis. The percentage of HD RBC (■) attached to laminin was higher than the percentage of LD RBC (▲). Paired t-test, **P<0.005. (B) Tracking of mobile cells during a flow adhesion assay for LD (upper panel) and HD (lower panel) RBC at 3 dyn/cm². (C) Computational treatment of brightfield images of the same field to identify immobile cells between 2 and 3 dyn/cm² during a flow adhesion assay. Overlaid image (bottom panel) revealing immobile cells in dark grey and mobile cells in light grey (D) Quantification of immobile cells for LD (▲) and HD (■) RBC from blood samples of four SCD patients. Results are expressed as percentage of immobile cells between two consecutive shear stress values. Paired t-test, *P<0.05. (E) Representative images of Lu/BCAM detected by immunofluorescence as well as tether length of LD and HD RBC in a flow adhesion assay at high shear stress (7 dyn/cm²). (F) Percentage of tether-containing cells in each RBC subpopulation at 7 dyn/cm². Paired t-test, *P<0.05. (G) Tether length in LD and HD RBC. Mann-Whitney test, ****P<0.0001.

Modulation Feature, that measures the intensity range of an image, normalized between 0 and 1 (formula: Modulation = Max Pixel - Min Pixel / Max Pixel + Min Pixel), reflecting the fluorescent signal distribution, we defined two subpopulations of mature Lu/BCAM RBC: Low-Modulation (Spots) and High-Modulation (Patches). Based on the x-axis Modulation_M11_Ch11 APC and y-axis Mean Pixel_M11_Ch11 APC, the “Spots” population was between -0.039 and 0.231, and the “Patches” population was between 0.235 and 0.552.

Flow cytometry assays Protein sialylation was measured by incubating RBC suspensions with biotinylated lectin (35 ng/mL) (Maackia amurens Lectin II, VECTOR) for 1 hour with constant shaking. After washes, Streptavidin-488 (10 μg/mL) (Streptavidin Alexa Fluor 488 conjugate, Invitrogen) was added to the pellet, and incubated for 30 min in the dark. Glycophorin-C (GPC) sialylation on mature RBC was determined by incubating the RBC suspensions with the antiBRIC 4 (dilution [d]: 1/100) antibody or anti-BRIC 10 (d: 1/200) antibody (IBGRL Research Products). After 1-hour incubation at room temperature and several washes, the RBC pellet was incubated with the secondary anti-mouse APC antibody (d: 1/100) (Beckman Coulter) for 45 min in the dark. RBC were analyzed using a BD FACS Canto II (BD Biosciences), the data obtained was analyzed using the FCS Express 6 software (De Novo).

Statistical analyses Data was analyzed by two-tailed Mann-Whitney or Wilcoxon test, and Paired t-test using the GraphPad Prism 7.00 software. *P≤0.05, **P≤ 0.01, ***P≤0.001 and ****P≤ 0.0001 were considered significant.

Results Validating high-density and low-density red blood cell isolation Deformability of AA and SS red blood cells HD RBC are known to be rigid cells with reduced deformability, which contributes to capillary blockade in vivo. In order to validate the fractionation method used to isolate LD and HD RBC from SCD blood samples, we assessed their deformability after isolation at the single cell level using a microfluidic approach based on perfusing RBC in a spleen-like biomimetic chip with filtering units comprising slits from 5 down to 2 μm31 (Online Supplementary Figure S1). First, we analyzed the deformability of total RBC using this biochip. AA and SS RBC were fluorescently labeled with PKH26 (red) and PKH67 (green), respectively, and mixed at a 1:1 ratio into a suspension at 0.1% hematocrit. Perfusing this RBC suspension led to a total blockade of the biochip indicating that the slit dimensions were not suitable for testing deformability of SS RBC. As the blockade occurred at the 5 μm wall we designed a new biochip with slit dimensions of 10 to 5 μm (Figure 1A). Perfusing the 1:1 AA-SS RBC suspen2482

sion into this biochip showed preferential trapping of SS RBC (Figure 1B), with these cells showing a higher retention rate than AA RBC (Figure 1C), indicating that the biochip was a good tool to assess RBC deformability in the SCD context.

Deformability of low-density and high-density sickle red blood cells RBC suspensions were prepared with PKH67-labeled LD RBC and PKH26-labeled HD RBC (1:1) and perfused in the biochip. Both LD and HD RBC were retained in the biochip, mainly at the 5 µm wall, indicating the presence of rigid cells in both populations (Figure 1D). Quantification of both RBC types retained in the 5 µm peripheral slits showed more HD than LD RBC (Figure 1E) indicating that RBC from the HD fraction were less deformable than those from the LD fraction, thus validating our fractionation method based on cell density.

High-density red blood cells are more resistant to shear stress and adhere more firmly to laminin than low-density red blood cells Red blood cell adhesion to laminin under flow conditions We assessed the adhesive properties and resistance to shear stress of RBC sub-populations by performing adhesion assays under flow conditions with LD and HD RBC from seven SCD patients using channels coated with Laminin 521. Both RBC types showed significant adhesion to laminin but there was a difference between them regarding resistance to shear stress. This resistance was first assessed by calculating the percentage of cells adhering at a given shear stress considering the number of adhering cells at 1 dyn/cm² as 100%. HD RBC were more resistant than LD RBC at physiological shear stresses for capillaries or post-capillary venules (2-5 dyn/cm²), while at high shear stress (10 dyn/cm²) both cell types showed no significant difference (Figure 2A). Exploring the cellular dynamics, we observed that LD RBC comprised a higher proportion of rolling cells than HD RBC (Online Supplementary Video S1), with a higher proportion of mobile cells covering a greater distance within the same time frame (Figure 2B). We assessed cell dynamics by determining the percentage of immobile cells between two consecutive shear stress steps. In order to do so, computational treatment of brightfield images of the same fields at two consecutive shear stresses was done to identify and count the cells that remained at the same spot between the two time-points (Figure 2C; Online Supplementary Figure S2A). The percentage of immobile cells was constantly higher for HD than LD RBC (Figure 2D), indicating that HD RBC were more firmly attached to laminin than LD RBC even at high shear stress.

Lu/BCAM expression pattern on adherent low-density and high-density red blood cells haematologica | 2021; 106(9)

Activation of SCD RBC adhesion by oxidative stress

Figure 3. Effect of in vitro oxidation on AA red blood cell adhesion to laminin. (A) Left panel: typical microscopy images of non-oxidized and cumene hydroperoxideoxidized AA red blood cells (RBC) adhering to Laminin 521 at 3 dyn/cm2. Right panel: quantification of cell adhesion showing the mean number of adherent RBC/mm2 in seven oxidized and non-oxidized AA RBC samples at 3 dyn/cm2. Wilcoxon test, *P<0.05. (B) Flow cytometry analysis of Lu/BCAM expression at the RBC surface expressed as percentage of Lu/BCAM-positive RBC (left panel) and mean fluorescence intensity (MFI) of these RBC (right panel) under non-oxidized (phosphate buffered saline [PBS]) and oxidized conditions (cumene). No significant differences were observed, Wilcoxon test, P=0.0714.

In order to gain insight into the potential mechanism underlying this difference between LD and HD RBC, cells were fixed after the 7 dyn/cm² step, stained fluorescently for Lu/BCAM and analyzed by confocal microscopy. There was a difference in the expression pattern of Lu/BCAM between LD and HD RBC at the interface with laminin. HD RBC showed a homogeneous distribution of Lu/BCAM with some cells showing intense staining and the presence of bigger spots suggestive of potential Lu/BCAM aggregates (Figure 2E; Online Supplementary Figure S2B). LD RBC showed cells with a smaller surface contact area, large fluorescent patches and very fine fluorescent membrane extensions tethering the cells to the surface (Figure 2E). The proportion of RBC exhibiting membrane tethers was higher in LD (55.6%) than in HD RBC (8.2%) (Figure 2F), and tethers were also longer in LD RBC (Figure 2G) suggesting a more dynamic lipid bilayer in LD RBC.

Oxidation activates AA red blood cell adhesion to laminin We have previously shown that Lu/BCAM-mediated adhesion to laminin is activated by phosphorylation of serine 621 of its cytoplasmic tail.17 In SCD, this phosphorylation takes place in reticulocytes, with very low levels haematologica | 2021; 106(9)

of phosphorylation detected in HD RBC.27 Considering the high levels of adhesion of HD RBC, we hypothesized that Lu/BCAM might be activated by post-translational modifications triggered by oxidative stress. In order to test this hypothesis, we assessed adhesion of control (AA) RBC under oxidative conditions after incubation with cumene hydroperoxide (270 μM), an agent that induces membrane lipid peroxidation. AA RBC showed the expected residual adhesion to laminin, that was significantly increased after incubation with cumene hydroperoxide (Figure 3A; Online Supplementary Figure S3). This was not due to a difference in the Lu/BCAM expression level between oxidative and control conditions as determined by flow cytometry measuring the percentage of Lu/BCAM-positive RBC and their mean fluorescence intensity (MFI) (Figure 3B). Moreover, this increased adhesion was not due to increased phosphorylation of Lu/BCAM as determined by western blot using an antiphosphoSerine antibody (not shown).

Oxidative stress alters Lu/BCAM distribution at the red blood cell surface Analysis of Lu/BCAM membrane distribution by confocal microscopy We evaluated the impact of oxidation on the distribu2483

M.A. Lizarralde-Iragorri et al.

Figure 4. Expression pattern of Lu/BCAM and ICAM-4 on red blood cells by confocal microscopy and imaging flow cytometry. Confocal microscopy images of red blood cells (RBC) expressing (A) Lu/BCAM and (B) ICAM-4 under non-oxidized and oxidized conditions. (C) Imaging flow cytometry analysis steps of Lu/BCAM expression on the RBC surface. After gating the single cells, gating is set on RBC facing the camera (circular cells), then on those expressing Lu/BCAM. In the final step a modulation feature is applied to measure the intensity range and distribution of Lu/BCAM, and define two main patterns: Low Modulation (Spots) and High Modulation (Patches). (D) Percentage of Lu/BCAM-positive RBC with Spots pattern (left) and Patches pattern (right) (n=10). Wilcoxon test, **P<0.01. (E) Representative flow cytometry plots of Lu/BCAM expression in a non-oxidized (phosphate buffered saline [PBS]) and oxidized (cumene hydroperoxide) AA RBC sample. (F) Percentage of ICAM-4-positive RBC with Spots pattern (left) and Patches pattern (right) (n=8). Wilcoxon test, P=0.8125 and P=0.3828. (G) Representative flow cytometry plots of ICAM-4 expression in a non-oxidized (PBS) and oxidized (cumene hydroperoxide) AA RBC sample. (H) Percentage of Lu/BCAM-positive RBC with Spots pattern (left) and Patches pattern (right) in low-density (LD) and high-density (HD) SS RBC (n=8); Wilcoxon test, **P<0.01, *P<0.05. (I) Representative flow cytometry plots of Lu/BCAM expression in LD and HD RBC from one SS blood sample.

2484

haematologica | 2021; 106(9)

Activation of SCD RBC adhesion by oxidative stress

Figure 5. Sialic acid levels, glycophorin C sialylation and Lu/BCAM distribution in AA and SS red blood cells. (A) Impact of neuraminidase (N’ase) treatment on Lu/BCAM Spots and Patches patterns on AA red blood cells (RBC). Wilcoxon test, *P<0.05. (B) Representative flow cytometry plots of Lu/BCAM expression in RBC treated or not with neuraminidase. (C) Left panel: sialic acid levels expressed as mean fluorescence intensity (MFI) on the RBC surface of 11 SS and nine AA samples. No significant difference was found between both groups. Mann-Whitney test, P=0.224. Right panel: representative flow cytometry plots of sialic acid distribution on one AA and one SS RBC samples. (D) Left panel: sialic acid levels expressed as MFI on the RBC surface of eight low-density (LD) and high-density (HD) samples. Wilcoxon test, **P<0.01. Right panel: representative flow cytometry plots of sialic acid distribution on LD and HD RBC from the same SS blood sample. (E) Left panel: glycophorin C (GPC) sialylation levels expressed as mean fluorescence intensity (MFI) on the RBC surface of eight LD and HD samples. Wilcoxon test, *P<0.05. Right panel: representative flow cytometry plots of silalylated GPC on LD and HD RBC from the same SS blood sample. (F) Left panel: GPC expression represented as MFI on the RBC surface of eight LD and HD samples. No significant differences were observed. Wilcoxon test, P=0.64. Right panel: representative flow cytometry plots of GPC expression on LD and HD RBC from the same SS blood sample.

haematologica | 2021; 106(9)

2485

M.A. Lizarralde-Iragorri et al.

tion of Lu/BCAM at the red blood cell surface using confocal microscopy. As expected and previously shown,26 Lu/BCAM showed a punctuated expression pattern at the red blood cell surface (Figure 4A left panel; Online Supplementary Figure S4). Under oxidative conditions, there was more RBC with large fluorescent patches suggestive of Lu/BCAM aggregation (Figure 4A right panel; Online Supplementary Figure S4). In order to check if this was a general feature of membrane proteins under oxidative conditions we stained for ICAM-4, another member of the immunoglobulin superfamily, and found no difference in its membrane distribution between both conditions (Figure 4B) indicating that oxidation targets Lu/BCAM membrane distribution through a specific mechanism.

Analysis of Lu/BCAM membrane distribution by imaging flow cytometry We assessed Lu/BCAM membrane distribution using a high throughput approach based on imaging flow cytometry. RBC were classified based on the expression pattern of Lu/BCAM using the modulation feature that measures the intensity range and distribution of a fluorescent signal (see methods) (Figure 4C). Two main patterns were distinguished in this analysis: the “Low modulation” for weak intensity dispersed spots (Spots) and “High modulation” for strong intensity big patches (Patches) (Figure 4C, last panel). Both patterns were concomitantly present in all blood samples and all conditions. We compared the proportion of each pattern under oxidative and control conditions and found less RBC with Spots and more with Patches in the presence of cumene hydroperoxide (Figure 4D; Online Supplementary Figure S5A), with no difference in Lu/BCAM global expression (Figure 4E), indicating that oxidation induces the formation of Lu/BCAM aggregates at the cell surface. As expected, based on the confocal microscopy results, oxidation did not impact the proportions of ICAM-4 Spots and Patches subpopulations (Figure 4F; Online Supplementary Figure S5A) or its global expression at the RBC surface (Figure 4G). Finally, we assessed Lu/BCAM distribution on SS LD and HD RBC in which both the Spots and the Patches patterns were found. Similar to oxidized AA RBC, the percentage of cells with the Patches pattern was higher in HD RBC in comparison with LD RBC, while the Spots pattern prevailed in the LD subpopulation (Figure 4H; Online Supplementary Figure S5B). As expected and already reported, Lu/BCAM expression was lower in HD than in LD RBC (Figure 4I).

Less glycophorin C sialylation on SS red blood cells Impact of sialic acid removal on Lu/BCAM membrane distribution We have recently shown that Lu/BCAM can establish cis-interactions with GPC sialic acids at the RBC surface keeping it in a “locked” conformation impeding its interaction with laminin.32 In order to test if such interactions modulate the Lu/BCAM expression pattern, AA RBC were treated with neuraminidase (N’ase) in order to eliminate protein sialic acids, labeled with an anti-Lu/BCAM antibody and analyzed by imaging flow cytometry. We found that loss of sialic acids resulted in less Spots and more Patches RBC (Figure 5A) with no impact on Lu/BCAM global expression level (Figure 5B), indicating that lateral interactions of Lu/BCAM with sialic acid residues impede its capacity to aggregate at the cell sur2486

face. Altogether, these results suggest that oxidation may trigger Lu/BCAM release from sialic acid lateral interactions leading to its aggregation and activated adhesive function.

Total sialylation and specific glycophorin C sialylation levels As HD RBC showed higher percentages of cells with the Patches pattern and that treatment of AA RBC with N’ase increases also the proportions of this sub-population, we hypothesized that increased HD RBC adhesion to laminin might result from aggregation of Lu/BCAM molecules subsequent to altered GPC sialylation. Using biotinylated lectins, we measured the sialic acid levels at the surface of RBC from nine AA and nine SS blood samples by flow cytometry and found no significant difference between the two groups (Figure 5C). We performed the same analysis on seven SCD fractionated blood samples and found less sialic acid at the surface of all HD RBC when compared to LD RBC (Figure 5D), suggesting that increased adhesion to laminin of HD RBC might result from the partial loss of interaction between Lu/BCAM and GPC at the cell surface. In order to specifically address GPC sialylation levels, we used an antibody directed against sialylated forms of GPC. Flow cytometry analysis showed less GPC sialylation levels in HD than in LD RBC (Figure 5E), while no significant difference was observed in total amounts of GPC at the cell surface as determined by flow cytometry using a sialic acid-independent antiGPC antibody (Figure 5F). This result supported our hypothesis of increased HD RBC adhesion to laminin following altered cis-interactions of Lu/BCAM with GPC at the cell surface.

Discussion Oxidative stress is an important feature of SCD and plays an important role in the pathophysiology of hemolysis, vaso-occlusion and ensuing organ damage. In this study we investigated the relationship between oxidative stress and adhesion of RBC to laminin. We report altered protein cis-interactions at the surface of sickle dense RBC that may account for the activation of RBC adhesion in the absence of signaling events and contribute to vasoocclusion. Using a microfluidic biomimetic chip, we confirm the importance of the mechanical parameter in the preferential trapping of HD RBC at a single cell level, confirming the hypothesis of the two-step model in which dense RBC contribute to the obstruction of fine blood vessels because of reduced deformability.13 In addition, we show that HD RBC adhere more firmly to laminin and are more resistant to shear stress than LD RBC suggesting that they would also contribute to initiate VOCs in vivo by adhering to the vessel wall even at high shear stress. This difference is probably partly due to the increased rigidity of HD RBC, as well as to the difference in cell shape between both subtypes, with a majority of very young and round reticulocytes in the LD fraction having a smaller contact surface with the capillary wall than HD RBC that are flatter cells with a larger contact surface and a smaller section facing the flow after adhesion is initiated. This is supported by the presence of long cellular tethers and of big patches of Lu/BCAM on several adhering LD RBC indicative of important membrane dynamics in this subpopulation, which is a characteristic of reticulocytes haematologica | 2021; 106(9)

Activation of SCD RBC adhesion by oxidative stress

known to undergo skeletal and membrane remodeling during maturation.33,34 Several studies have investigated the dynamics and rheology of SS RBC under flow conditions using microfluidic devices. Alapan et al. assessed sickle RBC adhesion in fibronectin-coated microfluidic chips and observed significantly greater numbers of adhered non-deformable than deformable RBC.35 Our study extends these findings by comparing the adhesion of deformable and nondeformable RBC within both the young and mature populations. As a matter of fact, fibronectin is the substrate of integrin α4β1 that is expressed only in very young reticulocytes, restricting the analysis to a very small subpopulation of RBC, while the laminin receptor Lu/BCAM is expressed on RBC at all the maturation stages. We show that oxidation can activate RBC adhesion to laminin by inducing post-translational modifications of Lu/BCAM that modify its distribution at the cell surface generating aggregates with high binding potential to laminin. This mechanism targets and abolishes Lu/BCAM cis-interaction with GPC at the cell surface32 and seems to be specific for Lu/BCAM and to target GPC primarily, as another IgSF member, i.e., LW/ICAM-4, was not impacted by in vitro oxidation and did not show altered distribution at the surface of SS RBCs. Abnormal RBC adhesion to laminin was reported in several pathologies15 with two triggering mechanisms including Lu/BCAM phosphorylation17,24,27 and Lu/BCAM dissociation from the spectrinbased skeleton.25,26,36 Here, the oxidation-driven mechanism seems to be at the origin of increased adhesion of HD RBC in the absence of Lu/BCAM phosphorylation and seems in line with the high adhesion levels of HD RBC in the absence of responsiveness to cAMP inducers.28 Moreover, although abnormal actin oxidation has been reported in irreversibly sickled cells affecting cytoskeletal dynamics,37 in our study oxidation and the subsequent loss of interaction with GPC do not alter Lu/BCAM binding to the skeleton as we did not see a difference in Lu/BCAM Triton extractability between in vitro oxidized and non-oxidized AA RBC (data not shown). This is supported by the unchanged mobility of Lu/BCAM at the surface of neuraminidase-treated RBC as measured by fluorescence recovery after photobleaching assay.32 As a matter of fact, oxidation may impact the interactions of membrane proteins with the skeleton as it was reported for Band 338-41 but here Lu/BCAM activation seems to be triggered by modifications at the extracellular rather than the intracellular side. During erythrocyte lifespan, sialic acid levels gradually

References 1. Pauling L, Itano HA, Singer SJ, Wells IC. Sickle cell anemia, a molecular disease. Science. 1949;110(2865):543-548. 2. Piel FB, Steinberg MH, Rees DC. Sickle cell disease. N Engl J Med. 2017;376(16):15611573. 3. Ware RE, Montalembert MD, Tshilolo L, Abboud MR. Sickle cell disease. Lancet. 2017;6736(17):1-13. 4. Barabino GA, Platt MO, Kaul DK. Sickle cell biomechanics. Annu Rev Biomed Eng. 2010;12:345-367. 5. Connes P, Lamarre Y, Waltz X, et al. Haemolysis and abnormal haemorheology in sickle cell anaemia. Br J Haematol. 2014;

haematologica | 2021; 106(9)

decrease.42-44 The sialic acid levels were lower on HD than on LD RBC for all blood samples, corroborating the finding that sialoglycoproteins are enriched in membranes of young reticulocytes45 and indicating an accelerated aginglike phenotype between the two stages despite the very short lifespan of RBC in SCD.46,47 This suggests increased damage of the RBC surface in SCD, targeting the glycocalyx, probably mediated by high levels of oxidative stress effectors in the plasma including free hemoglobin.48 Our study extends our recent findings on the novel mechanism activating Lu/BCAM-mediated RBC adhesion and suggests that this mechanism could be triggered by oxidative stress activating the adhesion of dense sickle RBC in the absence of Lu/BCAM phosphorylation. It would be interesting to determine the impact of anti-oxidant drugs on this specific mechanism and to evaluate their potential of reducing or attenuating RBC adhesion in SCD. Disclosures No conflicts of interests to disclose. Contributions MALI and SDL conducted experiments, acquired and analyzed data and wrote the manuscript; VB provided blood samples and discussed data; SC, SEH, AF, MD and SA conducted experiments, analyzed data and edited the manuscript; CLVK and FRL discussed data and edited the manuscript; OF, BLP, TK and RvB conducted experiments, discussed data and edited the manuscript; WEN designed research, analyzed data and wrote the manuscript. Acknowledgments We thank Mr Mickaël Marin, Mr Harvey Nagy and Dr Jean-Philippe Semblat for technical support. Funding The work was supported by the Institut National de la Santé et de la Recherche Médicale (INSERM), the Institut National de la Transfusion Sanguine, the Laboratory of Excellence GR-Ex, reference ANR-11-LABX-0051, and the Laboratory of Excellence LaSIPS (ANR-10-LABX-0040-Lasips). The labex GR-Ex is funded by the IdEx program “Investissements d’avenir” of the French National Research Agency, reference ANR-18-IDEX-0001. MALI and SEH were funded by the Ministère de l’Enseignement Supérieur et de la Recherche (Ecole Doctorale BioSPC); they received financial support from: Club du Globule Rouge et du Fer and Société Française d’Hématologie.

165(4):564-572. 6. Stuart MJ, Nagel RL. Sickle cell disease. Lancet. 2004;364(9442):1343-1360. 7. Hebbel RP. Beyond hemoglobin polymerization: the red blood cell membrane and sickle disease pathophysiology. Blood. 1991;77(2):214-237. 8. Hebbel RP. Adhesive interactions of sickle erythrocytes with endothelium. J Clin Invest. 1997;99(11):2561-2564. 9. Hebbel RP, Yamada O, Moldow CF, Jacob HS, White JG, Eaton JW. Abnormal adherence of sickle erythrocytes to cultured vascular endothelium. Possible mechanism for microvascular occlusion in sickle cell disease. J Clin Invest. 1980;65(1):154-160. 10. Cartron J, Elion J. Erythroid adhesion mole-

cules in sickle cell disease: effect of hydroxyurea. Transfus Clin Biol. 2008;15(1-2):3950. 11. Kaul DK, Fabry ME. In vivo studies of sickle red blood cells. Microcirculation. 2004; 11(2):153-165. 12. Kaul DK, Fabry ME, Nagel RL. Microvascular sites and characteristics of sickle cell adhesion to vascular endothelium in shear flow conditions: pathophysiological implications. Proc Natl Acad Sci U S A. 1989;86(9):3356-3360. 13. Kaul DK, Finnegan E, Barabino Ga. Sickle red cell-endothelium interactions. Microcirculation. 2009;16(1):97-111. 14. Kaul DK, Nagel RL. Sickle cell vasoocclusion: many issues and some answers.

2487

M.A. Lizarralde-Iragorri et al. Experientia. 1993;49(1):5-15. 15. Colin Y, Le van Kim C, El Nemer W. Red cell adhesion in human diseases. Curr Opin Hematol. 2014;21(3):186-192. 16. El Nemer W, Gane P, Colin Y, et al. The Lutheran blood group glycoproteins, the erythroid receptors for laminin, are adhesion molecules. J Biol Chem. 1998; 273(27):16686-16693. 17. Gauthier E, Wautier MP, Nemer WE, et al. Protein kinase a-dependent phosphorylation of lutheran/basal cell adhesion molecule glycoprotein regulates cell adhesion to laminin α5. J Biol Chem. 2005; 280(34): 30055-30062. 18. Telen MJ. Sickle cell anemia role of adhesion molecules and vascular endothelium in the pathogenesis of sickle cell disease. Hematology Am Soc Hematol Educ Program. 2007;84-90. 19. Udani M, Zen Q, Cottman M, et al. Basal cell adhesion molecule/lutheran protein: the receptor critical for sickle cell adhesion to laminin. J Clin Invest. 1998; 101(11):2550-2558. 20. Zennadi R, Hines PC, De Castro LM, Cartron JP, Parise LV, Telen MJ. Epinephrine acts through erythroid signaling pathways to activate sickle cell adhesion to endothelium via LW-alphavbeta3 interactions. Blood. 2004;104(12):3774-3781. 21. Zennadi R, Moeller BJ, Whalen EJ, et al. Epinephrine-induced activation of LWmediated sickle cell adhesion and vasoocclusion in vivo. Blood. 2007;110(7):27082717. 22. Parsons SF, Mallinson G, Holmest CH, et al. The Lutheran blood group glycoprotein, another member of the immunoglobulin superfamily, is widely expressed in human tissues and is developmentally regulated in human liver. Blood. 1995;92(12):5496-500. 23. Rahuel BC, Kim CLV, Mattei MG, Cartron JP, Colin Y. Unique gene encodes spliceoforms of the B-cell adhesion molecule cell surface glycoprotein of epithelial cancer and of the Lutheran blood group glycoprotein. Blood. 1996;88(5):1865-1872. 24. De Grandis MD, Cambot M, Wautier M-p, Cassinat B, Chomienne C, Colin Y, et al. JAK2V617F activates Lu/BCAM-mediated red cell adhesion in polycythemia vera through an EpoR-independent Rap1/Akt pathway. Blood. 2007;121(4):658-66. 25. An X, Gauthier E, Zhang X, et al. Adhesive activity of Lu glycoproteins is regulated by interaction with spectrin. Blood. 2008; 112(13):5212-5219.

2488

26. Gauthier E, El W, Wautier MP, et al. Role of the interaction between Lu/BCAM and the spectrin-based membrane skeleton in the increased adhesion of hereditary spherocytosis red cells to laminin. Br J Haematol. 2009;148(3):456-465. 27. Bartolucci P, Chaar V, Picot J, et al. Decreased sickle red blood cell adhesion to laminin by hydroxyurea is associated with inhibition of Lu/BCAM protein phosphorylation. Blood. 2010;116(12):1-4. 28. Hines PC, Zen Q, Burney SN, et al. Novel epinephrine and cyclic AMP-mediated activation of BCAM/Lu-dependent sickle (SS) RBC adhesion. Blood. 2003;101(8):32813287. 29. McDonald JC, Whitesides GM. Poly (dimethylsiloxane) as a material for fabricating microfluidic devices. Acc Chem Res. 2002;35(7):491-499. 30. Schneider CA, Rasband WS, Eliceiri KW, Instrumentation C. NIH image to imageJ: 25 years of image analysis. Nat Methods. 2012;9(7):671-675. 31. Picot J, Ndour PA, Lefevre SD, et al. A biomimetic microfluidic chip to study the circulation and mechanical retention of red blood cells in the spleen. Am J Hematol. 2015;90(4):339-345. 32. Klei TRL, Back DZD, Asif PJ, et al. Glycophorin-C sialylation regulates Lu/BCAM adhesive capacity during erythrocyte aging. Blood. 2018;2(1):14-24. 33. Chasis JA, Prenant M, Leung A, Mohandas N. Membrane assembly and remodeling during reticulocyte maturation. Blood. 1989;74(3):1112-1120. 34. Mohandas N, Groner W. Cell membrane and volume changes during red cell development and aging. Ann N Y Acad Sci. 1989;554:217-224. 35. Alapan Y, Matsuyama Y, Little JA, Gurkan UA. Dynamic deformability of sickle red blood cells in microphysiological flow. Technology (Singap World Sci). 2016; 4(2):71-79. 36. Wandersee NJ, Olson SC, Holzhauer SL, Hoffmann RG, Barker JE, Hillery CA. Increased erythrocyte adhesion in mice and humans with hereditary spherocytosis and hereditary elliptocytosis. Blood. 2004; 103(2):710-717. 37. Shartava A, Monteiro CA, Bencsath FA, et al. A posttranslational modification of betaactin contributes to the slow dissociation of the spectrin-protein 4.1-actin complex of irreversibly sickled cells. J Cell Biol. 1995; 128(5):805-818.

38. Arashiki N, Kimata N, Manno S, Mohandas N, Takakuwa Y. Membrane peroxidation and methemoglobin formation are both necessary for Band 3 clustering: mechanistic insights into human erythrocyte senescence. Biochemistry. 2013;52(34):57605769. 39. Mannu F, Arese P, Cappellini MD, et al. Role of hemichrome binding to erythrocyte membrane in the generation of Band-3 alterations in B-Thalassemia intermedia erythtrocytes. Blood. 1995;86(5):20142020. 40. Pantaleo A, Giribaldi G, Mannu F, Arese P, Turrini F. Naturally occurring anti-band 3 antibodies and red blood cell removal under physiological and pathological conditions. Autoimm Rev. 2008;7(6):457-462. 41. Noomuna P, Risinger M, Zhou S, et al. Inhibition of Band 3 tyrosine phosphorylation: a new mechanism for treatment of sickle cell disease. Br J Haematol. 2020;190 (4):599-609. 42. Hadengue AL, Del-pino M, Simon A, Levenson J. Erythrocyte disaggregation shear stress, sialic acid, and cell aging in humans. Hypertension. 1998;32(2):324330. 43. Huang Y-x, Tuo W-w, Wang D, Kang L-l, Chen X-y, Luo M. Restoring the youth of aged red blood cells and extending their lifespan in circulation by remodelling membrane sialic acid. J Cell Mol Med. 2016;20 (2):294-301. 44. Shinozuka T. Changes in human red blood cells during aging in vivo. Kelo J Med. 1994; 43(3):155-163. 45. Skutelsky E, Farquhar M. Variations in distribution of Con A receptor sites and anionic groups during red blood cell differentiation in the rat. J Cell Biol. 1976;71(1):218231. 46. Franco RS, Lohmann J, Silberstein EB, Mayfield-pratt G, Palascak M, Nemeth TA. Time-dependent changes in the density and hemoglobin F content of biotin-labeled sickle cells. J Clin Invest. 1998;101(12): 2730-2740. 47. Quinn CT, Smith EP, Arbabi S, et al. Biochemical surrogate markers of hemolysis do not correlate with directly measured erythrocyte survival in sickle cell anemia. Am J Hematol. 2016;91(12):1195-1201. 48. Rifkind JM, Mohanty JG, Nagababu E. The pathophysiology of extracellular hemoglobin associated with enhanced oxidative reactions. Front Physiol. 2015;5:500.

haematologica | 2021; 106(9)

LETTERS TO THE EDITOR

Survival and causes of death in 2,033 patients with non-transfusion-dependent β-thalassemia Non-transfusion-dependent β-thalassemia (NTDT) is a broad term encompassing patients who do not require lifelong transfusion therapy for survival. NTDT patients commonly, but not exclusively, present to medical care later in childhood (commonly >2 years of age) and with milder anemia and clinical symptoms compared to patients with transfusion-dependent forms. Our understanding of the disease process in NTDT has evolved significantly over the past two decades and it is now established that a diagnosis of NTDT can be associated with greater morbidity than previously recognized.1 Ineffective erythropoiesis and peripheral hemolysis lead to a state of chronic anemia, which can impact organ function in the long-term. There is a significant correlation between the degree of anemia and morbidity development in this NTDT patient population.2 Ineffective erythropoiesis is also linked to other pathogeneses manifesting as extramedullary hematopoiesis, bone disease, hypercoagulability and vascular disease, as well as primary iron overload due to increased intestinal iron absorption.1 Iron overload can be cumulative and leads to end-organ damage, especially in the liver.3,4 Despite advances in realizing risk factors and morbidity in NTDT, data on mortality and causes of death remain limited.5 There are currently no approved drugs for the management of ineffective erythropoiesis, or anemia, in NTDT. Thus, despite the terminology, many patients are given sporadic transfusions or are even placed on regular transfusion programs later in their disease course. This is commonly undertaken in situations of acute stress (during pregnancy, surgery, or infection), in the context of supporting growth and development, or for the management and prevention of complications in adulthood.6 Decisions are often based on the physician’s judgement, since management guidelines have only become available since

2013 and do not necessarily provide specific recommendations for transfusion therapy or for other erythropoiesis modulators when data from clinical trials is absent.7 Iron chelation therapy has been used in patients with NTDT for decades, but this have primarily based on expert opinion since data from dedicated clinical trials and management guidelines have only become available in the last 10 years.8,9 Against this background, the aim of the current study was to evaluate survival and causes of death in a large cohort of patients with NTDT. Considering the high variability in management practices, the impacts of transfusion and iron chelation therapy on mortality outcomes were also examined. Data were retrieved from an International Health Repository (IHR) established and approved on 25 May 2017 by the Italian Ethical Committee (EudraCT and Sponsor’s Protocol Code Numbers: 2017-004457-17 and 143AOR2017). All data were anonymized and added to the repository following informed consent by patients, or their legal representatives in case of death. The database included all β-thalassemia patients attending participating centers from 1 January 1997 onwards, and historic data were retrieved for all patients from birth up to 31 December 2020, in case of death, or loss to follow-up. The database included 13 international thalassemia centers of excellence from eight countries: Italy, Iran, Pakistan, USA, Oman, Egypt, Greece, and Saudi Arabia. For the current analysis, we gathered data on 2,033 patients identified by the centers as NTDT; a β-thalassemia diagnosis was confirmed by clinical and molecular studies at all participating centers. The definition of NTDT was based on the absence of dependence on transfusions for survival, delayed presentation, mildmoderate anemia, and clinician’s judgement of disease severity at diagnosis and during follow-up. Patients had homozygous or compound heterozygous β-thalassemia mutations, or heterozygous β-thalassemia mutations combined with α-globin gene duplications (and hence an

Table 1. Causes of death in 2,033 patients with non-transfusion-dependent β-thalassemia.

Cause

% of deaths (n = 113)

% of population (n = 2,033)

Median age at death (min-max), years

Cardiovascular disease (iron-related cardiomyopathy, n = 2; other cardiomyopathy, n = 14; myocardial infarction, n = 1; valvular disease, n = 1; pulmonary hypertension, thrombosis or peripheral vascular disease, n = 23) Hepatic disease (fibrosis or cirrhosis, n = 10; HCC, n = 13) Cancer (solid or hematologic malignancy excluding HCC) Infection Unclassified thalassemia-related complications Non-thalassemia related causes

36.3

2.0

34.2 (19-85)

20.4

1.1

55.4 (26-76)

12.4

0.7

54.0 (12-85)

13 17 5

11.5 15.0 4.4

0.6 0.8 0.2

44.1 (12-68) 19.8 (7-64) 62.0 (27-73)

HCC: hepatocellular carcinoma.

haematologica | 2021; 106(9)

2489

Letters to the Editor

α/β-globin imbalance leading to clinically-significant disease). For each patient, data were retrieved for gender, age at last observation, status (living or dead) at last observation and eventual cause of death. Data was also gathered on whether the patient was transitioned to a regular transfusion program (and the date) or started receiving iron chelation therapy (and date). Overall, 113 out of 2,033 patients (5.6%, 95.0% confidence interval [CI]: 4.6-6.6) died during the observation period. The median follow-up time was 33.9 years (interquartile range [IQR]: 23.7-46.8). The median age at death was 46.3 years (IQR: 28.3-61.9; 43.4% females), while the median age for patients alive at the last observation was 33.7 years (IQR: 23.7-45.9; 52.4% females). The Kaplan-Meier survival curve for all-cause mortality is illustrated in Figure 1A. Cumulative survival estimates at 18, 50, 65, 75, and 85 years were 99.4%, 93.4%, 81.8%, 66.2%, and 25.4%, respectively. By comparison, survival probability estimates at 50, 65, and 75 years in the normal population of Italy in 2019 were 98.5%, 94.0% and

82.9% (http://dati.istat.it/). Survival was significantly shorter in patients from the Middle East and Asia (n = 922) compared with the US and Europe (n = 1111), with a cumulative survival at 50 years of 74.9% vs 96.3%; Log-rank test χ2: 82.581, p <0.001). Causes of death are summarized in Table 1. Cardiovascular disease was the leading cause of early death (36.3%, at a median age of 34.2 years), while hepatic disease was the leading cause of death in older patients (20.4%, at a median age of 55.4 years). A subset of 254 patients (12.5%) were eventually placed on regular transfusion programs, starting at a median age of 10 years (IQR: 4-28.3). The remaining 1,779 patients (87.5%) received only sporadic or no transfusions at all. Survival was significantly worse in non-regularly transfused patients compared to regularly transfused patients for all-cause mortality (Log-rank test Chi-square: 13.298, P<0.001, Figure 1B). Cumulative survival estimates at 18, 50, 65, 75, and 85 years were 99.3% vs. 100%, 92.6% vs. 97.1%, 79.5% vs. 95.0%,

Figure 1. Kaplan-Meier survival curves. (A) all-cause mortality, (B) all-cause mortality according to regular transfusion therapy status, (C) mortality from cardiovascular disease according to regular transfusion therapy status, and (D) mortality from hepatic disease according to iron chelation therapy status.

2490

haematologica | 2021; 106(9)

Letters to the Editor

62.2% vs. 95.0%, and 18.5% vs. 95.0%, respectively. Survival was also significantly worse in non-regularly transfused patients compared to regularly transfused patients when considering mortality from cardiovascular disease (Log-rank test χ2: 4.571, P=0.033, Figure 1C) or hepatic disease (Log-rank test Chi-square: 4.989, P=0.026). Survival was comparable between splenectomized (n = 886) and non-splenectomized (n = 1,147) patients. Iron chelation therapy was concomitantly used in all regularly-transfused patients and in 1,168 (67.7%) nonregularly-transfused patients, beginning at a median age of 15 years (IQR: 7-26). Survival was comparable in chelated and non-chelated patients for all-cause mortality (Log-rank test χ2: 0.717, P=0.397). Survival was also comparable in chelated and non-chelated patients when considering mortality from cardiovascular disease (Log-rank test χ2: 0.001, P=0.0969), but was significantly worse in non-chelated compared with chelated patients when considering mortality from hepatic disease (Log-rank test χ2: 11.489, P=0.001, Figure 1D). On multivariate Cox regression analysis, including regular transfusion and iron chelation as explanatory variables, regular transfusion therapy was associated with a reduction of approximately 80% in the risk of all-cause mortality (hazard ratio [HR]: 0.202, 95.0%CI: 0.0800.509, P=0.001) and mortality from cardiovascular disease (HR: 0.199, 95.0%CI: 0.046-0.869, P=0.032); while iron chelation therapy was associated with a reduction of around 73% in the risk of mortality from hepatic disease (HR: 0.277, 95.0%CI: 0.093-0.830, P=0.022). This is the first study to provide mortality estimates in a large cohort of NTDT patients. Cardiovascular disease was the leading cause of death, but unlike in patients with TDT, this cannot be fully explained by cardiac siderosis and subsequent heart failure secondary to chronic transfusions given that the cohort is not so transfusion-dependent. Chronic anemia and hypercoagulability can play a considerable role in the development of vascular disease in NTDT (large- and micro-vessel thrombosis, pulmonary hypertension, peripheral and renal vascular disease) with or without cardiac dysfunction. In fact, and as seen in regularly-transfused patients in this cohort, transfusions in this context may have a protective effect by halting ineffective erythropoiesis and subsequent pathogeneses; an observation made in previous crosssectional studies.6 Improvement in hemolysis markers, nucleated red cells and cardiac index have also been reported in longitudinal studies of NTDT patients who were started on regular transfusions in adulthood.10 This also explains why iron chelation did not seem to have a role in preventing cardiovascular deaths in this cohort. Thus, a trial of chronic transfusion in patients at risk of significant morbidity may be justified but this needs to be weighed against the eventual risk of secondary siderosis and the elevated need and high cost of iron chelation therapy in a regular transfusion setting. We await data from various novel therapies targeting ineffective erythropoiesis and anemia in NTDT.11 Iron overload in non-regularly-transfused NTDT patients is attributed to hepcidin dysregulation and increased intestinal iron absorption.12 Observational studies indicate that hepatic siderosis is the main consequence, with no evidence of iron deposition in the heart (unlike in transfusional siderosis).1 Several reports have linked iron overload to hepatic fibrosis and hepatocellular carcinoma in NTDT;13-15 these were a common cause of death in this cohort, although at older ages considering they require more time to manifest. Iron chelation was haematologica | 2021; 106(9)

associated with a lower risk of death from hepatic disease, adding further evidence to data from clinical trials showing significant decline in liver iron concentration in NTDT patients receiving iron chelation.8,9 Our work merits further evaluation in prospective birth cohorts to address missing information and loss to follow-up bias typical of long-term retrospective studies, a factor that can lead to an over-estimation of survival risk. The study could also include additional subsets of nontransfusion-dependent patients including hemoglobin E/β-thalassemia and α-thalassemia, and further explore the role of genotype and environment in geographical variations in outcomes. Khaled M. Musallam,1* Angela Vitrano,2* Antonella Meloni,3 Sebastiano Addario Pollina,4 Mehran Karimi,5 Amal ElBeshlawy,6 Mahmoud Hajipour,7 Vito Di Marco,8 Saqib Hussain Ansari,9 Aldo Filosa,10 Paolo Ricchi,10 Adriana Ceci,11 Shahina Daar,12 Efthymia Vlachaki,13 Sylvia Titi Singer,14 Zaki A. Naserullah,15 Alessia Pepe,3 Salvatore Scondotto,4 Gabriella Dardanoni,4 Fedele Bonifazi,11 Vijay G. Sankaran,16 Elliott Vichinsky,14 Ali T. Taher17 and Aurelio Maggio2 International Working Group on Thalassemia (IWG-THAL) *KMM and AV contributed equally as co-first authors. 1

Thalassemia Center, Burjeel Medical City, Abu Dhabi, UAE; Campus of Haematology Franco and Piera Cutino, AOOR Villa Sofia-V. Cervello, Palermo, Italy; 3MRI Unit, Fondazione G. Monasterio CNR-Regione Toscana, Pisa, Italy; 4D.A.S.O.E, Regione Siciliana, Palermo, Italy; 5Haematology Research Center, Shiraz University of Medical Sciences, Shiraz, Iran; 6Department of Pediatric Haematology, Faculty of Medicine, Cairo University, Cairo, Egypt; 7 Pediatric Gastroenterology, Hepatology and Nutrition Research Center, Research Institute for Children’s Health, Shahid Beheshti University of Medical Sciences, Tehran, Iran; 8Department of Promozione della Salute, Materno Infantile, Medicina Interna e Specialistica di Eccellenza (PROMISE), University of Palermo, Palermo, Italy; 9 Department of Pediatric Haematology & Molecular Medicine, National Institute of Blood Diseases and Bone Marrow Transplantation, Karachi, Pakistan; 10Rare Blood Cell Disease Unit, "Cardarelli" Hospital, Naples, Italy; 11Fondazione per la Ricerca Farmacologica Gianni Benzi Onlus, Valenzano (BA), Italy; 12 Department of Haematology, College of Medicine and Health Sciences, Sultan Qaboos University, Sultanate of Oman, Wallenberg Research Centre, Stellenbosch Institute for Advanced Study, Stellenbosch University, Stellenbosch, South Africa; 13Thalassaemia Unit, Ippokratio University Hospital, Thessaloniki, Greece; 14 Division of Hematology-Oncology, Department of Pediatrics, University of California San Francisco, UCSF Benioff Children's Hospital Oakland, Oakland, CA, USA; 15Dammam Maternity and Child Hospital, Dammam, Saudi Arabia; 16Division of Hematology/Oncology, Boston Children's Hospital and Department of Pediatric Oncology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, USA; Broad Institute of MIT and Harvard, Cambridge, MA, USA; Harvard Stem Cell Institute, Cambridge, MA, USA and 17Department of Internal Medicine, American University of Beirut Medical Center, Beirut, Lebanon Correspondence: AURELIO MAGGIO - md.amaggio@gmail.com doi:10.3324/haematol.2021.278684 Received: March 1, 2021. Accepted: April 13, 2021. Pre-published: April 22, 2021. Disclosures: KMM has been or is a consultant for Novartis, Celgene Corp (Bristol Myers Squibb), Agios Pharmaceuticals, CRISPR Therapeutics and Vifor Pharma; AM received speakers’ honoraria from Chiesi Farmaceutici S.p.A.; EV received hononaria from DEMO S.A. 2

2491

Letters to the Editor

Pharmaceutical Industry and Novartis; ATT has been or is a consultant for Novartis, Celgene Corp (Bristol Myers Squibb), Vifor Pharma, Silence Therapeutics and Ionis Pharmaceuticals; and received research funding from Novartis, Celgene Corp (Bristol Myers Squibb), La Jolla Pharmaceutical Company, Roche, Protagonist Therapeutics and Agios Pharmaceuticals; AM has been or is a member of advisory boards for Novartis, Celgene Corp (Bristol Meyers Squibb) and Bluebird Bio. The remaining authors have no conflicts of interest to disclose. Contributions: KMM, AV, AM: study design; AV, AM, SAP, MK, AE-B, MH, VDM, SHA, AF, PR, AC, SD, EV, STS, ZAN, EV: data collection; KMM, AV: data analysis; KMM, AV, AM: manuscript drafting; data interpretation and manuscript review for intellectual content: all authors. Final approval for submission: all authors. Acknowledgments: the authors would like to thank all patients for agreeing to participate in this study. The support of the Foundation Franco and Piera Cutino is appreciated. Data-sharing: data were collected and stored on the IHR electronic platform (www.sanitasicilia.eu/IWG), and are available from the corresponding author upon request.

References 1. Musallam KM, Rivella S, Vichinsky E, Rachmilewitz EA. Non-transfusion-dependent thalassemias. Haematologica. 2013;98(6):833844. 2. Musallam KM, Cappellini MD, Taher AT. Variations in hemoglobin level and morbidity burden in non-transfusion-dependent beta-thalassemia. Ann Hematol. 2021;100(7):1903-1905. 3. Musallam KM, Cappellini MD, Daar S, et al. Serum ferritin level and morbidity risk in transfusion-independent patients with beta-thalassemia intermedia: the ORIENT study. Haematologica. 2014; 99(11):e218-221. 4. Musallam KM, Cappellini MD, Wood JC, et al. Elevated liver iron concentration is a marker of increased morbidity in patients with beta thalassemia intermedia. Haematologica. 2011;96(11):16051612. 5. Vitrano A, Calvaruso G, Lai E, et al. The era of comparable life

2492

expectancy between thalassaemia major and intermedia: Is it time to revisit the major-intermedia dichotomy? Br J Haematol. 2017; 176(1):124-130. 6. Taher AT, Musallam KM, Karimi M, et al. Overview on practices in thalassemia intermedia management aiming for lowering complication rates across a region of endemicity: the OPTIMAL CARE study. Blood. 2010;115(10):1886-1892. 7. Taher A, Vichinsky E, Musallam K, Cappellini MD, Viprakasit V. Guidelines for the Management of non transfusion dependent thalassaemia (NTDT). Nicosia, Cyprus: Thalassaemia International Federation; 2013. PMID: 24672826 8. Taher AT, Porter J, Viprakasit V, et al. Deferasirox reduces iron overload significantly in nontransfusion-dependent thalassemia: 1-year results from a prospective, randomized, double-blind, placebo-controlled study. Blood. 2012;120(5):970-977. 9. Calvaruso G, Vitrano A, Di Maggio R, et al. Deferiprone versus deferoxamine in thalassemia intermedia: Results from a 5-year longterm Italian multicenter randomized clinical trial. Am J Hematol. 2015;90(7):634-638. 10. Ricchi P, Meloni A, Pistoia L, et al. Longitudinal follow-up of patients with thalassaemia intermedia who started transfusion therapy in adulthood: a cohort study. Br J Haematol. 2020;191(1):107114. 11. Musallam KM, Rivella S, Taher AT. Management of non-transfusion-dependent beta-thalassemia (NTDT): The next 5 years. Am J Hematol. 2021;96(3):E57-E59. 12. Rivella S. beta-thalassemias: paradigmatic diseases for scientific discoveries and development of innovative therapies. Haematologica. 2015;100(4):418-430. 13. Moukhadder HM, Halawi R, Cappellini MD, Taher AT. Hepatocellular carcinoma as an emerging morbidity in the thalassemia syndromes: a comprehensive review. Cancer. 2017; 123(5):751-758. 14. Borgna-Pignatti C, Garani MC, Forni GL, et al. Hepatocellular carcinoma in thalassaemia: an update of the Italian Registry. Br J Haematol. 2014;167(1):121-126. 15. Musallam KM, Motta I, Salvatori M, et al. Longitudinal changes in serum ferritin levels correlate with measures of hepatic stiffness in transfusion-independent patients with beta-thalassemia intermedia. Blood Cells Mol Dis. 2012;49(3-4):136-139.

haematologica | 2021; 106(9)

Letters to the Editor

Circulating histones play a central role in COVID-19-associated coagulopathy and mortality COVID-19 has highlighted the lethal consequences of immunothrombosis; i.e., the cross-talk between coagulation, inflammation and the innate immune system. Patients with immunothrombosis have significant immune cell death,1 which can release pro-coagulant2 and cytotoxic3 histones. Histones are small, positivelycharged proteins that are typically found within the cell nucleus and which bind to negatively-charged DNA. We hypothesize that circulating histones play a central role in critically-ill COVID-19 patients. This translational study demonstrates that admission histone levels are significantly elevated with increasing severity of COVID-19 infection (Mild, median=2.6 μg/mL [IQR=0.7-7.6], Moderate, 10.5 μg/mL [3.5-27.2], Critical, 20.0 μg/mL [6.2-33.0], Non-survivors, 29.6 μg/mL [11.2-60.0]; P<0.001). Circulating histones associated with severe coagulopathy, inflammation and organ injury markers, including cardiac troponin. Extracellular histone levels on admission are associated with poor outcomes and independently predict 28-day mortality of hospitalized COVID-19 patients. This is the first report to indicate that circulating histones, released following immune cell death, may play a central pathological role in severe SARS-CoV-2 infection. COVID-19 was the cause of more than two million deaths worldwide by February 2021,4 resulting from respiratory and multi-organ failure,5 with evidence of pulmonary thrombosis at post-mortem.6 These patients have extensive immune cell death,1 a strong acute-phase inflammatory response and coagulopathy, as well as cardiac injury.1,5 Cell death can release histones, and extracellular histones are cytotoxic, pro-inflammatory7 and pro-coagulant,2 leading to pulmonary thrombosis.8 Extracellular histones also trigger interleukin-6 (IL-6) release to induce an acute phase response, including elevation of C-reactive protein (CRP), which, in turn, reduces histone toxicity.9 High levels of circulating histones initiate an alternative coagulation pathway during sepsis,2 mediate multiple organ injury3 and correlate with adverse clinical outcomes, including death.10 We therefore hypothesized that high levels of histones are present in severe SARS-CoV-2 infection, and act as major mediators of coagulopathy and mortality in COVID-19 disease. In this study, adult COVID-19 patients (n=113) were recruited at the Royal Liverpool University Hospital from 30th March 2020 to 16th May 2020. Patients were selected using the ISARIC WHO Clinical Characterisation Protocol for Severe Emerging Infections in the UK. Inclusion criteria were: (1) swab positive or high likelihood of infection or (2) ≥1 of the following symptoms: fever ≥38°C, new cough, dyspnea or tachypnea and admitted to a healthcare facility.11 Patients were categorized into four groups: 1) Mild (minor respiratory symptoms to exclude shortness of breath OR incidental finding, where the patient required admission to hospital for reasons other than COVID-19 (such as for frailty) and was otherwise asymptomatic of COVID-19); 2) Moderate (dyspnea, i.e., patient symptomatic with shortness of breath OR hypoxia, defined by oxygen saturations on pulse oximeter of ≤93% or requiring supplementary oxygen to maintain oxygen saturations ≥96%); 3) Critical disease (respiratory failure requiring the administration of continuous positive airway pressure (CPAP) to maintain oxygen saturations ≥96% OR invasive ventilation in a critical care setting); 4) Non-survivors (patients haematologica | 2021; 106(9)

who died within 28 days of hospital admission). Circulating histones were quantified in patient plasma on admission (as described previously)8,12 and associations with severity of infection, coagulation, inflammatory and organ injury markers were analyzed. Severity of infection was determined by the patient’s most severe clinical state throughout hospital admission, according to the previously described definitions. Cytokines were measured using a Luminex-based bead array, as per manufacturer’s instructions (Thermo-Fisher Scientific). Outcome measures included ventilator-support days, length of hospital stay, and 28-day mortality. Ethical approval was provided by the South Central - Oxford C Research Ethics Committee in England (Ref 13/SC/0149), the Scotland A Research Ethics Committee (Ref 20/SS/0028), and the WHO Ethics Review Committee (RPC571 and RPC572, 25 April 2013). Local approval was granted by the North West - Haydock Research Ethics Committee (REC reference 20/NW/0332). The Kruskall-Wallis test was used to compare continuous variables, presented as median (interquartile range; IQR); the Fisher Exact/χ2 test for comparison of categorical variables, presented as counts (percentage). Circulating histone levels were measured by Western Blot, using purified histone as the standard, and analyzed either as continuous variables or categorized based on a previously-determined threshold for cytotoxicity (30 μg/mL).3,7 The Mann-Whitney U test was used to compare categorical histone levels to continuous clinical variables. Correlation analysis was performed using Spearman’s rank. A Receiver Operating Characteristic (ROC) curve analysis and multivariate regression (adjusted for age, gender, ethnicity and co-morbidities) assessed admission histone levels in predicting 28-day mortality. Kaplan-Meier survival curve analysis was performed to analyze the probability of mortality over time. Statistical tests were performed on SPSS (IBM, version 25). A 2-tailed P value of <0.05 was considered significant. The study involved 113 COVID-19 patients (Table 1): median age 65.0 years (IQR=51.0-78.0 years), 65 patients were male (57.5%), 96 of white ethnicity (85.0%). Disease severity was associated with coagulation activation (Table 1), characterized by elevated D-dimer (P=0.017) and prolonged prothrombin time (P=0.005), and a pro-inflammatory phenotype characterized by elevated CRP (P<0.001) and IL-6 (P=0.002) on hospital admission, as well as with hypoxia and cardiac injury (Table 1). The median hospital stay was 10 days (IQR, 3-20 days) and 25 patients (22.1%) died within 28 days. Circulating histone levels on admission were significantly elevated in COVID-19 patients compared to normal controls and were associated with increasing severity of infection (Figure 1A and B; Healthy controls, median=2.9 μg/mL [IQR=1.5-3.3]; Mild, 2.6 μg/mL [0.7-7.6]; Moderate, 10.5 μg/mL [3.5-27.2]; Critical, 20.0 μg/mL [6.2-33.0]; Non-survivors, 29.6 μg/mL [11.2-60.0]; P<0.001). Circulating histone levels strongly correlated with D-dimer levels (R=0.606), indicating the potential involvement of extracellular histones in COVID-19 coagulopathy. Positive association with organ injury markers, including bilirubin (R=0.531), creatinine (R=0.501) and cardiac troponin (R=0.486), indicates the possible role of histone-induced cytotoxicity in multiple organ injury. Strong associations with fibrinogen (R=0.632), CRP (R=0.735) and IL-6 (R=0.677) confirmed histone-initiated acute phase response.9 Negative correlation with lymphocyte count (R=-0.446) suggests that lymphocyte and other immune cell death might be a major source of circulating histones in COVID-19 infection. 2493

Letters to the Editor

Adopting a 30 μg/mL cytotoxic histone threshold,3,7 patients over the threshold (n=29) had significantly higher D-dimer (2267.0 ng/mL [1227.0-5235.0] vs. 1128.0 ng/ml [589.0-1844.3], P=.001), fibrinogen (6.6 g/L [4.67.6] vs. 4.8 g/L [3.9-5.7], P=0.012), IL-6 (226.2 pg/mL [90.6-518.9] vs. 71.8 pg/mL [35.2-111.4], P<0.001) and CRP levels (186 mg/L [108.5-247.5] vs. 48.0 mg/L [10.0107.5], P<0.001) than those patients below the threshold

(Table 2). These patients also had significantly reduced SpO2 compared to those with circulating histones <30 μg/mL (oxygen saturations 92.0% [85.8-94.0] vs. 95.0% [93.5-97.0], P=0.001), required critical care admission (P<0.001), with a longer duration of mechanical ventilation (R=0.635) and longer hospital stay (R=0.654). Circulating histone levels were significantly higher in non-survivors than those who survived (29.6 μg/mL

Table 1. Demographics, peripheral blood measurements and outcomes for disease severity groups in COVID-19 infection.

Total Total number 113 Demographics & Comorbidities Age (years), Median [IQR] 65.0 [51.0, 78.0] Male, No. [%] 65 [57.5] White ethnicity, No. [%] 96 [85.0] Smoking history, No. [%] 38 [33.6] Hypertension, No. [%] 36 [31.9] Asthma/COPD, No. [%] 29 [25.7] Diabetes mellitus, No. [%] 29 [25.7] Ischemic heart disease, No.[%] 16 [14.2] Chronic kidney disease, No. [%] 15 [13.3] Histones (μg/mL), Median [IQR] 10.8 [3.2, 29.9] Peripheral blood cell counts White blood cells (x109/L), Median [IQR] 8.5 [5.8, 11.8] Neutrophils (x109/L), Median [IQR] 6.4 [4.0, 9.3] Lymphocytes (x109/L), Median [IQR] 1.0 [0.7, 1.6] Haemoglobin (g/L), 129.0 Median [IQR] [117.8, 145.3] Platelets (x109/L), 236.5 Median [IQR] [170.3, 296.0] Coagulation parameters PT (seconds), Median [IQR] 13.2 [12.1, 14.4] aPTT (seconds), Median [IQR] 30.6 [28.2, 33.6] Fibrinogen (g/L), Median [IQR] 4.8 [3.9, 6.5] D-dimer (ng/mL), 1227.0 Median [IQR] [687.0, 2141.5] Antithrombin (%), 80.0 Median [IQR] [61.0, 100.0] Pro-inflammatory markers IL-6 (pg/ml), 79.0 Median [IQR] [40.5, 131.9] C-reactive protein (mg/L), 61.0 Median [IQR] [21.0, 153.5] Organ injury markers Troponin T (ng/L), Median [IQR] 12.0 [5.0, 35.0] Bilirubin (µmol/L), Median [IQR] 9.0 [6.0, 14.0] ALT (U/L), Median [IQR] 25.5 [14.5, 45.0] Creatinine (µmol/L), Median [IQR] 77.0 [63.0, 105.0] SpO2 (%), Median [IQR] 95.0 [92.0, 97.0] Outcomes Length of stay (days), Median [IQR] 10.0 [3.0, 20.0] Ventilator support (days), Median [IQR] 0.0 [0.0, 0.0]

Mild

Moderate

Critical

Non-survivors

P valuea

63.5 [42.0, 70.0] 15 [50.0] 26 [86.7] 10 [33.3] 8 [26.7] 14 [46.7] 5 [16.7] 3 [10.0] 3 [10.0] 2.6 [0.7, 7.6]

67.0 [57.5, 81.5] 51.0 [42.8, 54.5]*,¥ 76.0 [66.0, 86.0]*,† 20 [52.6] 14 [70.0] 16 [64.0] 35 [92.1] 11 [55.0] 24 [96.0] 16 [42.1] 4 [20.0] 8 [32.0] 12 [31.6] 5 [25.0] 11 [44.0] 10 [26.3] 1 [5.0] 4 [16.0] 10 [26.3] 5 [25.0] 9 [36.0] 8 [21.1] 0 [0.0] 5 [20.0] 10 [26.3] 0 [0.0] 2 [8.0] * 10.5 [3.5, 27.2]* 20.0 [6.2, 33.0] 29.6 [11.2, 60.0]*,¥

8.2 [6.6, 10.7] 5.9 [3.8, 8.0] 1.2 [0.8, 1.7] 126.0 [119.0, 145.0] 253.0 [177.0, 311.0]

9.8 [5.9, 12.3] 7.0 [4.1, 9.8] 1.1 [0.8, 1.4] 123.0 [113.8, 139.8] 243.5 [113.8, 139.8]

8.1 [6.5, 10.8] 6.4 [4.0, 9.0] 1.1 [0.9, 2.1] 134.5 [131.0, 146.0] ¥ 250.5 [207.3, 299.3]

8.1 [5.2, 11.3] 7.2 [4.0, 11.2] 0.7 [0.4, 1.1]*,¥,† 136.0 [107.0, 147.0] 174.0 [124.0, 250.0]*,¥,†

12.1 [11.2, 13.0] 31.0 [28.9, 32.7] 4.2 [2.8, 5.4]† 755.5 [431.5, 1744.0] 81.0 [57.5, 98.5]

13.1 [12.1, 14.4]* 30.5 [28.3, 32.6] 4.8 [4.4, 6.7] 1315.0 [832.5, 2176.3] 80.0 [61.5, 97.5]

13.4 [13.1, 14.2]* 32.0 [29.1, 33.7] 6.5 [5.4, 6.6]* *950.0 [602.0, 1728.0] 98.0 [80.3, 114.8]*,¥

14.1 [12.4, 20.7] * 30.0 [28.2, 37.6] 4.5 [3.1, 4.9]† 1630.0 [1117.0, 4334.0]*,† 70.0 [59.0, 87.0]†

53.2 [15.0, 83.1] 16.0 [3.5, 53.8]

70.5 [41.9, 115.0] 52.0 [23.3, 146.3]*

166.7 [75.6, 214.7]* 145.0 [97.0, 202.5]*,¥

107.7 [81.3, 269.8]*,¥ 105.0 [71.0, 192.0]*,¥

8.0 [5.0, 16.0] 8.0 [4.5, 13.0] 21.0 [11.5, 55.0] 74.5 [62.0, 82.3] 97.0 [95.0, 98.0] 2.0 [1.0, 13.8] 0.0 [0.0, 0.0]

<0.001 0.428 0.001 0.033 0.474 0.005 0.443 0.116 0.025 <0.001 0.623 0.748 0.009 0.122 0.026

0.005 0.775 0.010 0.017 0.024

0.002 <0.001

16.0 [6.8, 47.3]* 6.5 [5.0, 10.5]¥ 35.0 [17.0, 58.0]*,† <0.001 8.0 [6.0, 15.0] 9.0 [6.0, 12.5] 12.0 [8.0, 16.5]* 0.142 ¥ 19.0 [11.5, 38.0] 33.5 [29.0, 59.5] 28.5 [15.8, 44.3] 0.163 78.0 [60.8, 104.3] 80.0 [57.8, 96.0] 102.0 [71.0, 180.0]* 0.125 94.5 [92, 96]* 94.0 [92.0, 96.5]* 92.0 [78.5, 96.0]* <0.001 10.0 [6.0, 22.0]* 0.0 [0.0, 0.0]

17.0 [9.5, 43.8]* 2.0 [0.0, 9.3]

0.0 [0.0, 8.0]

<0.001 <0.001

P value for comparisons mild vs. moderate vs. critical disease vs. non-survivors, collectively. Performed using Kruskall-Wallis test. *Significant vs. mild disease. ¥Significant vs moderate disease. †Significant vs. critical disease.

2494

haematologica | 2021; 106(9)

Letters to the Editor

[11.2-60.0] vs. 8.6 µg/ml [3.1-24.8], P=0.002), and, accordingly, patients with histones >30 μg/mL were more likely to die (13/29 [44.8%] vs. 12/84 [14.3%], P=0.001). Patients who died were significantly older than those who survived (Table 2, 76 years [66-86] vs. 59 years [46-72] P<0.001). Compared to survivors, non-survivors had evidence of consumptive coagulopathy with lower platelet counts (P=0.003), prolonged prothrombin time

(P=0.028), elevated D-dimer (P=0.017) and reduced antithrombin levels (P=0.048). Furthermore, in non-survivors, lymphocyte counts (P=0.001), and oxygen saturations (P=0.005) were significantly reduced, and IL-6 (P=0.021), CRP (P=0.013), troponin (P<0.001), bilirubin (P=0.041) and creatinine (P=0.024) were elevated when compared to survivors (Table 2). Univariate analysis using continuous circulating his-

D C

E Figure 1. High levels of circulating histones on hospital admission are associated with disease severity and mortality in COVID-19. Typical Western Blots (A) and quantification (B) of histone levels in healthy controls (n=12), mild (n=30), moderate (n=38), critical disease (n=20) and non-survivors (n=25) with COVID-19 infection. Circulating histone levels were higher with increasing disease severity (P<0.001). Histone levels were higher in non-survivors compared to the moderate (P=0.023), mild groups (P<0.001) and to normal healthy controls (P<0.001). Histone levels were higher in the critical group compared to mild groups (P<0.001) and normal healthy controls (P<0.001). Histone levels were higher in the moderate group compared to the mild group (P=0.007) and normal healthy controls (P=0.002). (C) Multivariate analysis of crude and adjusted odds ratios (with patients adjusted for age, gender, Black and Ethnic Minorities (BAME) and comorbidities including smoking, hypertension, asthma/COPD, diabetes, ischemic heart disease and chronic kidney disease). Circulating histone levels ≥30 μg/mL were independently associated with 28-day mortality. (D) Kaplan-Meier survival curve for the probability of mortality during the 28-day period. Patients were stratified based on circulating histones levels on admission (<30 μg/mL vs. ≥ 30 μg/mL). (E) Diagram to propose that circulating histones play a central pathological role in the development of severe COVID-19.

haematologica | 2021; 106(9)

2495

Letters to the Editor

Table 2. Demographics, peripheral blood measurements and outcomes of COVID-19 patients.

Total number (n) Demographics & Comorbidities Age (years), n. Median [IQR] Male, No. [%] White ethnicity, No. [%] Smoking history, No. [%] Hypertension, No. [%] Asthma/COPD, No. [%] Diabetes mellitus, No. [%] Ischemic heart disease, No. [%] Chronic kidney disease, No. [%] Histones (μg/mL), Median [IQR] Peripheral blood cell counts White blood cells (x109/L), Median [IQR] Neutrophils (x109/L), Median [IQR] Lymphocytes (x109/L), Median [IQR] Haemoglobin (g/L), Median [IQR] Platelets (x109/L), Median [IQR] Coagulation parameters PT (seconds), Median [IQR] aPTT (seconds), Median [IQR] Fibrinogen (g/L), Median [IQR] D-dimer (ng/mL), Median [IQR] Antithrombin (%), Median [IQR] Pro-inflammatory markers IL-6 (pg/mL), Median [IQR] C-reactive protein (mg/L), Median [IQR] Organ injury markers Troponin T (ng/L), Median [IQR] Bilirubin (µmol/L), Median [IQR] ALT (U/L), Median [IQR] Creatinine (μmol/L), Median [IQR] SpO2 (%), Median [IQR] Outcomes Length of stay (days), Median [IQR] Ventilator support (days), Median [IQR] Mortality at 28 days, No. [%]

Survivors

Non-survivors

P valuea

Histones <30 μg/mL

Histones ≥30 μg/mL

P valueb

59.0 [45.8, 72.3] 49 [55.7] 72 [81.8] 30 [34.1] 25 [28.4] 25 [28.4] 20 [22.7] 11 [12.5] 13 [14.8] 8.6 [3.1, 24.8]

76.0 [66.0, 86.0] 16 [64.0] 24 [96.0] 8 [32.0] 11 [44.0] 4 [16.0] 9 [36.0] 5 [20.0] 2 [8.0] 29.6 [11.2, 60.0]

<0.001

63.0 [47.8, 76.0] 48 [57.1] 73 [86.9] 28 [33.3] 28 [33.3] 25 [29.8] 21 [25.0] 13 [15.5] 11 [13.1] 6.1 [2.0, 13.5]

66.0 [57.0, 80.0] 17 [58.6] 23 [79.3] 10 [34.5] 8 [27.6] 4 [13.8] 8 [27.6] 3 [10.3] 4 [13.8] 51.6 [38.2, 72.8]

0.224

8.7 [6.1, 11.8] 6.2 [4.0, 8.9] 1.1 [0.8, 1.7] 128.0 [118.0, 144.0] 248.0 [181.0, 299.0]

8.1 [5.2, 11.3] 7.2 [4.0, 11.2] 0.7 [0.4, 1.1] 136.0 [107.0, 147.0] 174.0 [124.0, 250.0]

0.387

8.0 [5.7, 11.0] 5.7 [3.6, 8.2] 1.2 [0.8, 1.7] 128.0 [118.0, 145.0] 237.5 [174.3, 295.8]

9.8 [6.7, 13.3] 9.1 [6.1, 12.2] 0.8 [0.5, 1.1] 131.0 [116.0, 147.0] 215.0 [155.8, 296.8]

13.0 [11.8, 14.1] 30.9 [28.4, 32.9] 5.3 [4.1, 6.5] 1166.0 [619.0, 2038.0] 83.0 [62.5, 102.5]

14.1 [12.4, 20.7] 30.0 [28.2, 37.6] 4.5 [3.1, 4.9] 1630.0 [1117.0, 4334.0] 69.5 [55.8, 81]

0.028 0.858 0.091 0.017

12.8 [11.8, 14.0] 30.7 [28.7, 34.0] 4.7 [3.9, 5.7] 1128.0 [589.0, 1844.3] 82.0 [59.0, 100.4]

13.8 [13.3, 15.6] 29.5 [28.0, 32.6] 6.6 [4.6, 7.6] 2267.0 [1227.0, 5235.0] 77.0 [69.0, 99.0]

0.005 0.268 0.012 0.001

73.9 [36.6, 125.4] 50.0 [15.3, 149.0]

107.7 [81.3, 269.8] 105.0 [71.0, 192.0]

0.021

71.8 [35.2, 111.4] 48.0 [10.0, 107.5]

226.2 [90.6, 518.9] 186.0 [108.5, 247.5]

<0.001

<0.001 0.041 0.727 0.024 0.005

10.0 [5.0, 24.0] 8.0 [5.0, 13.0] 20.5 [12.8, 38.3] 76.0 [62.5, 99.3] 95.0 [93.5, 97.0]

25.0 [9.8, 57.3] 11.0 [85.0, 16.3] 36.5 [25.5, 55.3] 96.0 [65.0, 154.0] 92.0 [85.8, 94.0]

0.011 0.016 0.062 0.127 0.001

0.347 <0.001

28.0 [13.0, 41.5] 0.0 [0.0, 8.0] 13 [44.8]

<0.001

0.0 [0.0, 0.0] 25 [100]

8.0 [2.5, 15.5] 0.0 [0.0, 0.0] 12 [14.3]

5.0 [10.0, 23.0] 35.0 [17.0, 58.0] 8.0 [5.0, 13.0] 12.0 [8.0, 16.5] 25.0 [12.8, 45.0] 28.5 [15.8, 44.3] 76.0 [61.0, 96.8] 102.0 [71.0, 180.0] 95.0 [93.0, 97.0] 92.0 [78.5, 96.0] 10.0 [3.0, 20.0] 0.0 [0.0, 0.0] 0 [0]

0.458 0.113 0.845 0.140 0.301 0.180 0.343 0.515 0.002

0.563 0.001 0.740 0.003

0.048

0.013

0.890 0.324 0.910 0.567 0.138 0.783 0.758 >0.999 <0.001

0.084 0.001 0.007 0.740 0.410

0.971

<0.001

<0.001 0.001

a P value for survivors vs.non-survivors. bP value for toxic histone levels vs.non-toxic. Performed using the Mann-Whitney U test for continuous variables and Fisher Exact/χ2 tests for categorical variables.

2496

haematologica | 2021; 106(9)

Letters to the Editor

tones demonstrated that rising histone levels were associated with mortality (odds ratio =1.031 (95% CI=1.0131.049, P=0.001). Using categorical data where patients were stratified based on a ≥30 μg/mL threshold,3,7 similar results were obtained (Figure 1C, OR=4.875 (95% CI=1.879-12.649, P=0.001), demonstrating that patients with high circulating histone levels on admission had a higher risk of mortality. Subsequent multivariate analysis demonstrated that histones were independently associated with mortality after adjustment for age, gender, ethnicity and co-morbidity when histone levels were treated as either continuous (odds ratio=1.032; 95% CI=1.0131.051, P=0.001) or categorical variables (odds ratio=5.404; 95% CI=1.852-15.770, P=0.002). ROC curve analysis shows an area under the curve [AUC] of 0.708 (95% CI=0.589-0.827, P=0.002). A Kaplan-Meier survival curve demonstrated a significant increase in the probability of mortality during the 28-day period in patients with histones ≥30 μg/mL (Figure 1D, P<0.001). Coagulopathy has emerged as a key feature of severe COVID-19 and has been linked to increased mortality.13 It has been documented that extracellular histones, released following cell death, are drivers of coagulation by activating platelets,7 generating thrombin2 and damaging endothelial cells8 to induce coagulopathy in critical illness.3 This is the first report to demonstrate high levels of circulating histones in SARS-CoV-2 infection, with levels strongly associated with coagulopathy. This suggests their involvement in thrombosis in severe cases.14 High levels of circulating histones reflect the extent of cellular death, such as lymphopenia or NETosis,15 which may be a major source of circulating histones in COVID-19. Histone release following cell death triggers IL-6 release to induce an acute-phase response.8 We found that circulating histone levels significantly correlated with IL-6 and acute-phase protein levels, including fibrinogen and CRP, indicating histone-induced acute phase response in patients with COVID-19. Extracellular histones disrupt cell membranes through phospholipid binding to induce cytotoxic effects on cells, including endothelial cells8 and cardiomyocytes.12 This study demonstrates circulating histones associated with cardiac injury, which is frequently observed in severe COVID-19 and associated with poor outcomes.5 Therefore, the cytotoxic and pro-coagulant properties of circulating histones may be an underlying molecular mechanism contributing to disease severity and poor outcomes (Figure 1E). In conclusion, this is the first report to quantify high levels of circulating histones in viral infection and demonstrate that extracellular histones play a central role in the development of immunothrombosis and critical illness in COVID-19. Rebecca J. Shaw,1,2* Simon T. Abrams,1* James Austin,1* Joseph M. Taylor,3 Steven Lane,4 Tina Dutt,2 Colin Downey,3 Min Du,1 Lance Turtle,1,5 J. Kenneth Baillie,6,7 Peter J.M. Openshaw,8 Guozheng Wang,1 Malcolm G. Semple1,9# and Cheng-Hock Toh1,2# on behalf of the ISARIC4C investigators *RJS, STA and JA contributed equally as co-first authors. # MGS and C-HT contributed equally as-senior authors. 1

Department of Clinical Infection, Microbiology and Immunology, University of Liverpool, Liverpool; 2Roald Dahl Haemostasis and Thrombosis Centre, Liverpool University Hospitals NHS Foundation Trust, Liverpool; 3Liverpool Clinical Laboratories, Liverpool University Hospitals NHS Foundation Trust, Liverpool; 4Department of Biostatistics, University of Liverpool, Liverpool; 5Infectious Diseases Unit, Royal Liverpool University Hospital, Liverpool; 6Roslin Institute, haematologica | 2021; 106(9)

University of Edinburgh, Edinburgh; 7Intensive Care Unit, Royal Infirmary Edinburgh, Edinburgh; 8National Heart and Lung Institute, Imperial College London, London and 9Respiratory Medicine, Alder Hey Children’s Hospital NHS Foundation Trust, Liverpool, UK Correspondence: CHENG-HOCK TOH - toh@liverpool.ac.uk doi:10.3324/haematol.2021.278492 Received: February 5, 2021. Accepted: April 1, 2021. Pre-published: April 8, 2021. Disclosures: all authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported. Contributions: RJS, JT, TD, CD and STA collected and interpreted the clinical data and performed data analysis. RJS, STA and SL performed statistical analysis. JA, STA, RJS and MD measured the levels of circulating histones. RJS, STA, TD, GW and CHT wrote, edited and reviewed the manuscript and figures. PJMO, JKB, LT and MS edited and reviewed the manuscript. STA, GW and CHT designed and supervised the work. STA and CHT had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. STA, GW, CHT, PJMO, JKB, and MS obtained funding for this work. Acknowledgments: this work uses data provided by patients and collected by the NHS as part of their care and support #DataSavesLives. We are extremely grateful to the 2,648 frontline NHS clinical and research staff and volunteer medical students, who collected this data in challenging circumstances; and the generosity of the participants and their families for their individual contributions in these difficult times. We also acknowledge the support of Jeremy J Farrar and Nahoko Shindo. Funding: this work was funded by University of Liverpool COVID-19 strategic funding, the British Heart Foundation [PG/16/65/32313], Bayer AG (Germany) and the Royal Liverpool & Broadgreen University Hospitals NHS Trust. It was funded in whole, or in part, by the Wellcome Trust [205228/Z/16/Z]. This research is supported by grants from: the National Institute for Health Research (NIHR) [award CO-CIN-01]; the Medical Research Council [grant MC_PC_19059] and by the NIHR Health Protection Research Unit (HPRU) in Emerging and Zoonotic Infections at University of Liverpool, in partnership with Public Health England (PHE), in collaboration with the Liverpool School of Tropical Medicine and the University of Oxford [award 200907]; NIHR HPRU in Respiratory Infections at Imperial College London with Public Health England (PHE) [award 200927]; the Wellcome Trust and Department for International Development [215091/Z/18/Z]; the Bill and Melinda Gates Foundation [OPP1209135]; the Liverpool Experimental Cancer Medicine Centre (Grant Reference: C18616/A25153); the NIHR Biomedical Research Centre at Imperial College London [IS-BRC1215-20013]; the EU Platform foR European Preparedness Against (Re-) emerging Epidemics (PREPARE) [FP7 project 602525] and NIHR Clinical Research Network for providing infrastructure support for this research. PJMO is supported by an NIHR Senior Investigator Award [award 201385]. The views expressed are those of the authors and not necessarily those of the DHSC, DID, NIHR, MRC, Wellcome Trust or PHE. The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Data-sharing: this work uses data provided by patients and collected by the NHS as part of their care and support #DataSavesLives. The CO-CIN data was collated by ISARIC4C Investigators. ISARIC4C welcomes applications for data and material access through our Independent Data and Material Access Committee (https://isaric4c.net). 2497

Letters to the Editor

References 1. Huang C, Wang Y, Li X, et al. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. Lancet. 2020; 395(10223):497-506. 2. Abrams ST, Su D, Sahraoui Y, et al. Assembly of alternative prothrombinase by extracellular histones initiates and disseminates intravascular coagulation. Blood. 2021;137(1):103-114. 3. Cheng Z, Abrams ST, Alhamdi Y, et al. Circulating histones are major mediators of multiple organ dysfunction syndrome in acute critical illnesses. Crit Care Med. 2019;47(8):e677-e684. 4. WHO Coronavirus Disease (COVID-19) Dashboard. 2020 [cited 18th September 2020]; Available from: https://covid19.who.int/ 5. Zhou F, Yu T, Du R, et al. Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study. Lancet. 2020;395(10229):1054-1062. 6. Wichmann D, Sperhake JP, Lütgehetmann M, et al. Autopsy findings and venous thromboembolism in patients with COVID-19. Ann Intern Med. 2020;173(4):268-277. 7. Alhamdi Y, Abrams ST, Lane S, Wang G, Toh CH. Histone-associated thrombocytopenia in patients who are critically ill. JAMA. 2016;315(8):817-819. 8. Abrams ST, Zhang N, Manson J, et al. Circulating histones are medi-

2498

ators of trauma-associated lung injury. Am J Respir Crit Care Med. 2013;187(2):160-169. 9. Abrams ST, Zhang N, Dart C, et al. Human CRP Defends against the toxicity of circulating histones. J Immunol. 2013;191(5):24952502. 10. Xu J, Zhang X, Pelayo R, et al. Extracellular histones are major mediators of death in sepsis. Nat Med. 2009;15(11):1318-1321. 11. Docherty AB, Harrison EM, Green CA, et al. Features of 20,133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation Protocol: prospective observational cohort study. BMJ. 2020;369:m1985. 12. Alhamdi Y, Abrams ST, Cheng Z, et al. Circulating histones are major mediators of cardiac injury in patients with sepsis. Crit Care Med. 2015;43(10):2094-2103. 13. Tang N, Li D, Wang X, Sun Z. Abnormal coagulation parameters are associated with poor prognosis in patients with novel coronavirus pneumonia. J Thromb Haemost. 2020;18(4):844-847. 14. Klok FA, Kruip MJHA, van der Meer NJM, et al. Incidence of thrombotic complications in critically ill ICU patients with COVID-19. Thromb Res. 2020;191:145-147. 15. Middleton EA, He XY, Denorme F, et al. Neutrophil extracellular traps contribute to immunothrombosis in COVID-19 acute respiratory distress syndrome. Blood. 2020;136(10):1169-1179.

haematologica | 2021; 106(9)

Letters to the Editor

Heritability and association with distinct genetic loci of erythropoietin levels in the general population Erythropoietin (Epo)-regulated red blood cell (RBC) homeostasis is crucial for oxygen delivery in vertebrates, and Epo over- or under-production causes erythrocytosis or anemia, respectively. Plasma Epo levels are usually very low and characterized by circadian fluctuations and strong induction upon acute exposure to low oxygen conditions such as inspiratory hypoxia or blood loss. Considering the high prevalence of various forms of anemia worldwide as well as the fact that the majority of erythrocytosis cases are of unknown origin, surprisingly little is known about the genetic determinants of circulating Epo levels. While a number of genome-wide association studies (GWAS) could link RBC traits such as RBC size, hemoglobin (Hb) content and hematocrit levels to single-nucleotide polymorphisms (SNP), only few studies analyzed circulating Epo levels. Some specific studies focused on aging populations with gradually decreasing renal function,1 Epo levels after kidney transplantation,2 or the (lack of) sex-specific differences in chronically anemic patients.3 However, none of these initial analyses have been performed in a population-based cohort which would have the advantage of a high external validity of the findings. A first GWAS with circulating Epo levels was performed in a large (6,984 participants) non-familybased Dutch cohort.4 Epo values were obtained from 6,777 participants and 2,691 were used for GWAS which identified a locus comprising the HBS1L and MYB genes as most likely targets, but no replication cohort was available for validation by an independent study. We analyzed RBC traits in the Swiss Kidney Project on Genes in Hypertension (SKIPOGH) cohort, including 1,109 adult participants from the general population with well characterized physiological parameters.5 Participants' characteristics by sex, including RBC traits, are shown in the Online Supplementary Table S1. An important aspect of the SKIPOGH cohort is its familybased design (the average family size in the SKIPOGH cohort is four) which allows for the analysis of heritability, i.e., the genetic component of a given phenotypic trait. As shown in Table 1, a high heritability of RBC traits was observed in the SKIPOGH cohort. Adding a sibling component of variance had little impact on the heritability estimates, suggesting the absence of significant dominance variance and shared environmental components across siblings. These initial results motivated us to analyze Epo blood

plasma levels which, according to our knowledge, had not been studied in such a family-based cohort before. Epo was determined in heparin-treated plasma of a total of 1,066 (96%) participants. For 1,020 samples both duplicate measurements (replicate correlation, r=0.998) and phenotypic data were obtained. Following the elimination of 14 extreme outliers, the remaining 1,006 (91%) unadjusted Epo values (Figure 1A) corresponded to the reported reference values of 2.8-17.9 IU/L (based on 2,506 samples).4 Over a normal Hb range of 135-175 g/L and 120-155 g/L for men and women, respectively, the range of Epo was 4-24 IU/L without any significant difference between sex (Online Supplementary Table S1). A quadratic fit with Hb levels was found (Figure 1B), which may be explained by anemia-responsive Epo at low Hb levels and hormone-responsive Hb at high Epo levels. As shown in Table 2, Epo levels were significantly heritable, without significant sibling or marital components of variance, suggesting the absence of significant dominance variance and shared environmental variance. We next perfomed a GWAS of 2.5x106 genotyped SNP and an additional 4.0x106 of imputed SNP with the mean of duplicate Epo measurements of 872 (79%) individuals, corrected for age, sex, center and familiarity. Figure 1C shows the level of significance of the association between each of the 6.5x106 markers and the normalized Epo levels. No signal reached P<5x10-8, the commonly used level of genome-wide significance (Online Supplementary Table S2). However, a few SNP fell into the suggestive significance zone (P<10-5) and the top hit, lying on chromosome 15, reached P=1.05x10-7 at rs413451. As shown in Figure 1D, the SNP identified on chromosome 15 are located within a linkage disequilibrium (LD) block comprising the last exons of mitogen-activated protein kinase kinase 5 (MAP2K5), the whole SKI family transcriptional corepressor 1 (SKOR1) gene and the 5' upstream and promoter regions of protein inhibitor of activated STAT1 (PIAS1). PIAS1 is a SUMO E3 ligase affecting STAT1 and NFκB pathways. The PIAS1 locus has previously been associated with body mass index (BMI) and related phenotypes (weight, waist circumference, obesity, predicted visceral adipose tissue), as well as smoking-related phenotypes (initiation age, smoking status) and age at menarche.6-8 According to the MR-Base PHEWAS database in the UK Biobank cohort several RBC-related phenotypes were also associated with the MAP2K5-SKOR1-PIAS1 locus, further suggesting that it could be directly associated with Epo levels. Figure 1E shows the significance of the association for the SNP present at the previously associated locus

Table 1. Heritability estimates of red blood cell indices in the SKIPOGH cohort.

Model 1 h2 ± SEM Hemoglobin Hematocrit RBC count MCV MCH MCHC RDW

0.40 ± 0.05 0.37 ± 0.06 0.50 ± 0.05 0.68 ± 0.04 0.63 ± 0.05 0.60 ± 0.06 0.38 ± 0.07

l 0.52 0.57 0.67 0.59 0.52 0.94 0.20

P -7

<1.0x10 0.001 <1.0x10-6 <1.0x10-7 <1.0x10-5 <1.0x10-5 <0.001

Model 2 h2 ± SEM

0.37 ± 0.06 0.33 ± 0.07 0.47 ± 0.06 0.68 ± 0.05 0.61 ± 0.06 0.46 ± 0.06 0.36 ± 0.08

0.52 0.57 0.66 0.59 0.51 0.93 0.20

<1.0x10-7 <1.0x10-7 <1.0x10-7 <1.0x10-7 <1.0x10-5 <1.0x10-7 <0.001

Models are adjusted for age, sex and center. Model 1, no sibship component of variance; model 2, including a sibship component of variance (which captures dominance genetic variance and shared environmental components between siblings). l, power transformation: (l = 0) and (l = 1) correspond to log and no transformations, respectively. SKIPOGH: Swiss Kidney Project on Genes in Hypertension; RBC: red blood cell; MCV: mean corpuscular volume; MCH: mean cell hemoglobin; MCHC: mean corpuscular hemoglobin concentration; RDW: red blood cell distribution width; h2: heritability; SEM: standard error of the mean.

haematologica | 2021; 106(9)

2499

Letters to the Editor

Figure 1. Genome-wide association studies of erythropoietin plasma levels in the SKIPOGH cohort. (A) Histogram distribution of the mean values of duplicate erythropoietin (Epo) measurements in the blood plasma of 1,006 participants. (B) Correlation between Epo and hemoglobin (Hb) levels in the SKIPOGH cohort. (C) Manhattan plot showing the –log10 (P-value) of the association between Epo plasma levels and 6.4x106 single-nucleotide polymorphisms (SNP) in 872 samples following correction for sex, age and center. The blue line shows the indicative suggestive threshold of P<10–5. Markers are ranked by chromosome and positions. The green dot on chromosome 6 shows the top SNP at the HBS1L-MYB locus, and the green dot on chromosome 15 shows the top SNP at the MAP2K5-SKOR1-PIAS1 locus. (D) Region around the top SNP rs413451 in the MAP2K5-SKOR1-PIAS1 locus. (E) Region around the top SNP rs9402685 in the HBS1L-MYB intergenic locus which is the reference SNP found in our study. SKIPOGH: Swiss Kidney Project on Genes in Hypertension.

Table 2. Heritability of plasma erythropoietin levels in the SKIPOGH cohort.

Variance components P* - M - S P* - M P* - S P* P* - M - S P* - M P* - S P*

Adjustment

Heritability ± SD

Age, sex, center Age, sex, center Age, sex, center Age, sex, center Fully adjusted Fully adjusted Fully adjusted Fully adjusted

0.40 ± 0.07 0.41 ± 0.07 0.39 ± 0.07 0.40 ± 0.06 0.51 ± 0.08 0.52 ± 0.07 0.48 ± 0.07 0.49 ± 0.06

<0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001

Shown is the heritability of ln(Epo) ± standard deviation. Narrow sense heritability was estimated from family data using the ASSOC program in the Statistical Analysis in Genetic Epidemiology software package (Case Western Reserve University). SKIPOGH: Swiss Kidney Project on Genes in Hypertension; Epo: erythropoietin; SD: standard deviation; P: polygenic; M: marital; S: sibling. Fully adjusted: age, sex, center, current smoker (yes/no), hemoglobin level, eGFR (ckd-epi formula). *Only the polygenic component of the variance was significantly different from 0 in all models.

HBS1L-MYB. Interestingly, the top common SNP of our study (rs9402685) and the associated HBS1L-MYB locus showed a robust association with Epo levels (P=1.46 x10-4), confirming the previously published results.4 The SNP rs1617640 of the EPO locus itself has been reported to be associated with low Epo serum levels in predialysis chronic kidney disease patients,9 but neither this SNP nor any other SNP of the EPO locus (lowest P=0.17 at rs7789679) were associated with Epo levels in our study. While the recently reported SNP rs1130864 of the C-reactive protein (CRP) locus did not associate with Epo levels in our study (P=0.69), it associated with altered Epo levels in dried neonatal blood spots.10 In the same study, EPO SNP heritability was approximately zero.

2500

However, at this age blood volume and hematocrit are different from the adult stage, the liver-to-kidney switch of Epo synthesis is still ongoing, and the method for Epo determination was much less precise, altogether explaining the discrepancy. The result obtained with our top SNP of the HBS1L-MYB locus (rs9402685) was meta-analyzed with results publicly available from Beverborg et al.4 The combined P-value reached 1.78x10-23, which was more significant than in any of the two individual studies (1.46x10-4 and 1.09x10-20, respectively). The intergenic locus between the HBS1L (GTP-binding elongation factor) and MYB (myeloblastosis oncogene) genes had previously been reported to be associated with deregulated HbF in a

haematologica | 2021; 106(9)

Letters to the Editor

Chinese β-thalassemia anemia population, a favorable genetic environment for the selection of otherwise erythrocytosis-causing mutations.11 Several erythropoietic transcription factors have been shown to be prevented from binding to the mutant locus, resulting in lowered Myb gene activation and increased HbF synthesis.12 A link to increased Epo levels, suggesting secondary (Epodependent) rather than primary (Epo-independent) erythrocytosis, has not been made in these original reports. GWAS performed in a UK Biobank cohort and a Japanese population showed significant associations between the HBS1L-MYB locus and RBC-related phenotypes.13,14 Together with our direct replication of the association with circulating Epo levels in a Swiss cohort, these results further confirm the implication of this locus in erythropoiesis, maybe both upstream as well as downstream of Epo. However, it is currently not known whether the HBS1L-MYB locus also contributes to the heritable genetic determinants triggering Epo levels. A gene score and a pathway analysis, run with the PASCAL algorithm based on our GWAS results, failed to show any significant pathway after applying multiple testing corrections. For a candidate-based approach, we selected 33 genes known to influence EPO gene expression. The association gene scores from PASCAL could be retrieved from 30 of these 33 genes. Bonferroni correction applied to the number of genes observed led to a significance threshold of 1.67x10-3. The OS9 gene was significantly associated with a gene score association P-value of 1.47x10-3 (Online Supplementary Table S3). OS-9 is known to interact with both HIF-1α and HIF prolyl-4-hydroxylases, promoting HIF-1α degradation. Interestingly, a OS9 gene variant has previously been reported to be associated with erythrocytosis in a single patient.15 The top SNP of the MAP2K5-SKOR1-PIAS1 locus (rs413451) was subjected to a phenome-wide association study (PHEWAS) using the MR-Base database of the UK Biobank cohort. SNP rs413451 was most significantly associated with BMI-related phenotypes. Interestingly, Hb concentration (P=6.35x10-6), reticulocyte count (P=7.28x10-6), hematocrit (P=1.16x10-4), reticulocyte fraction of RBC (P=2.04x10-4) and RBC count (P= 2.23x10-4) were also highly associated with rs413451. Circulatory Epo levels were not available in the UK Biobank. However, the GWAS Atlas database showed a preponderance of BMI-related phenotypes for most significant studies in the database of published GWAS. In summary, our study revealed the heritability of circulating Epo levels, validated a previously published association with the HBS1L-MYB locus, and identified an association with the MAP2K5-SKOR1-PIAS1 locus. From the list of candidate Epo-regulatory genes, OS9 showed the highest association with circulating Epo levels. However, the two latter associations require replication, and the functional implication of all three loci in Epo regulation needs to be further investigated. Regarding the idiopathic nature of the majority of erythrocytosis cases, we suggest that especially in patients with high Epo levels, indicative of secondary erythrocytosis, these loci should be considered for further investigation. Tanguy Corre,1,2,3 Belen Ponte,4 Edward Pivin,1 Menno Pruijm,5 Daniel Ackermann,6 Georg Ehret,7 Katharina Spanaus,8 Murielle Bochud,1,2 and Roland H. Wenger2,9 1 Center for Primary Care and Public Health (Unisanté), University of Lausanne, Lausanne; 2National Center of Competence in Research haematologica | 2021; 106(9)

“Kidney.CH”, Zurich; 3Department of Computational Biology, University of Lausanne, Lausanne; 4Nephrology Service, Department Medicine, Geneva University Hospital, Geneva; 5Nephrology Service, University Hospital of Lausanne and University of Lausanne, Lausanne; 6Department of Nephrology and Hypertension, Inselspital, Bern and University Hospital, University of Bern, Bern; 7Cardiology, Department of Medicine, Geneva University Hospital, Geneva; 8 Institute of Clinical Chemistry, University Hospital of Zurich, Zurich and 9Institute of Physiology, University of Zurich, Zurich, Switzerland Correspondence: ROLAND H. WENGER - roland.wenger@access.uzh.ch doi:10.3324/haematol.2021.278389 Received: January 17, 2021. Accepted: February 26, 2021. Pre-published: April 8, 2021. Disclosures: no conflicts of interest to disclose. Contributions: BP, MP, DA, and GE provided materials; KS measured the samples; TC, EP and MB performed data analysis; TC, MB and RHW wrote the manuscript; MB and RHW supervised the study. Acknowledgments: we wish to thank P. Spielmann for expert technical help. Funding: this project was supported by the NCCR "Kidney.CH".

References 1. Ble A, Fink JC, Woodman RC, et al. Renal function, erythropoietin, and anemia of older persons: the inchianti study. Arch Intern Med. 2005;165(19):2222-2227. 2. Goch J, Birgegard G, Wikstrom B, Backman U, Wadstrom J, Danielson BG. Serum erythropoietin and erythropoiesis during six years after kidney transplantation. Nephron. 1996;74(4):687-693. 3. Jelkmann W, Wiedemann G. Lack of sex dependence of the serum level of immunoreactive erythropoietin in chronic anemia. Klin Wochenschr. 1989;67(23):1218. 4. Grote Beverborg N, Verweij N, Klip IT, et al. Erythropoietin in the general population: reference ranges and clinical, biochemical and genetic correlates. PLoS One. 2015;10(4):e0125215. 5. Moulin F, Ponte B, Pruijm M, et al. A population-based approach to assess the heritability and distribution of renal handling of electrolytes. Kidney Int. 2017;92(6):1536-1543. 6. Speliotes EK, Willer CJ, Berndt SI, et al. Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index. Nat Genet. 2010;42(11):937-948. 7. Karlsson Linner R, Biroli P, Kong E, et al. Genome-wide association analyses of risk tolerance and risky behaviors in over 1 million individuals identify hundreds of loci and shared genetic influences. Nat Genet. 2019;51(2):245-257. 8. Perry JR, Day F, Elks CE, et al. Parent-of-origin-specific allelic associations among 106 genomic loci for age at menarche. Nature. 2014; 514(7520):92-97. 9. Yugavathy N, Huri HZ, Kun LS, et al. Clinical and genetic markers of erythropoietin deficiency anemia in chronic kidney disease (predialysis) patients. Biomark Med. 2020;14(12):1099-1108. 10. Wang Y, Nudel R, Benros ME, et al. Genome-wide association study identifies 16 genomic regions associated with circulating cytokines at birth. PLoS Genet. 2020;16(11):e1009163. 11. Stadhouders R, Aktuna S, Thongjuea S, et al. Hbs1l-myb intergenic variants modulate fetal hemoglobin via long-range myb enhancers. J Clin Invest. 2014;124(4):1699-1710. 12. Suzuki M, Yamazaki H, Mukai HY, et al. Disruption of the hbs1lmyb locus causes hereditary persistence of fetal hemoglobin in a mouse model. Mol Cell Biol. 2013;33(8):1687-1695. 13. Astle WJ, Elding H, Jiang T, et al. The allelic landscape of human blood cell trait variation and links to common complex disease. Cell. 2016;167(5):1415-1429.e1419. 14. Kanai M, Akiyama M, Takahashi A, et al. Genetic analysis of quantitative traits in the Japanese population links cell types to complex human diseases. Nat Genet. 2018;50(3):390-400. 15. Camps C, Petousi N, Bento C, et al. Gene panel sequencing improves the diagnostic work-up of patients with idiopathic erythrocytosis and identifies new mutations. Haematologica. 2016; 101(11):1306-1318.

2501

Letters to the Editor

Prognostic impact of soluble CD163 in patients with diffuse large B-cell lymphoma We assessed the clinical utility of soluble CD163 (sCD163) in patients with diffuse large B-cell lymphoma (DLBCL), by measuring sCD163 levels prior to, during and after treatment in two independent cohorts. Our results demonstrate that pre-treatment sCD163 levels decrease in response to therapy and, if elevated, predict an unfavorable outcome. The findings suggest that sCD163 represents a useful and easily assessable biomarker for therapeutic monitoring in DLBCL. Although the combination of rituximab with chemotherapy has revolutionized treatment in DLBCL,1 approximately 20-30% of patients relapse with a dismal survival.2 International Prognostic Index-based classifications are used to predict outcomes,3 but the composition of the tumor microenvironment has also been recognized to have prognostic impact on survival.4 Macrophages infiltrating into the tumor microenvironment are usually polarized as tumor-associated macrophages (TAM), which strongly express CD163,5 and CD163 expression in the tumor microenvironment is regarded as a marker of TAM. Elevated levels of CD163 ectodomain, a sCD163, have been detected in serum, and associated with adverse outcome in lymphoid malignancies.6,7 While TAM content predicts survival in DLBCL4,8 the clinical relevance of sCD163 is unknown. We examined the clinical utility of sCD163 in two independent cohorts of patients with DLBCL to gain fur-

ther understanding of the biological role of macrophages during the clinical course of DLBCL. A prospective clinical trial cohort and a population-based cohort were used to reach generalizable results. The trial cohort included patients <65 years with highrisk DLBCL treated with dose-dense immunochemotherapy in the Nordic NLG-LBC-05 phase II trial.9 Available samples included 119 pre-treatment and 94 paired midtreatment samples. Samples from five healthy volunteers formed a control group. The trial was registered at www.ClinicalTrials.gov as NCT01325194. All patients gave written informed consent to the study. The Institutional Review Boards, National Medical Agencies, and Ethics Committees in Finland, Norway, Denmark, and Sweden approved the protocol and sampling. The population-based cohort was obtained from the Swedish biobank U-CAN10 and included 125 pre-treatment samples collected between 2010 and 2016. Available paired samples included 30 mid-treatment samples, 71 post-treatment samples from patients in complete remission, and 11 samples taken during primary progressive or relapsing disease. A majority of the patients (93%) were treated with rituximab plus cyclophosphamide, doxorubicin, vincristine, prednisone (R-CHOP)-based therapy. Complete remission was defined in clinical routine. All patients gave written informed biobank consent. The study was approved by the Regional Board of the Ethical Committee in Uppsala, Sweden. In the trial cohort, sCD163 was measured using a Quantikine enzyme-linked immunosorbent assay

Figure 1. sCD163 levels in two cohorts and correlation between pre-treatment sCD163 levels, CD163+ tumor-associated macrophages and CD163 mRNA levels in the trial cohort. (A) Comparison of sCD163 levels in pre-treatment and mid-treatment samples (after three courses) in the trial cohort, and in healthy volunteers. (B) sCD163 levels in pre-treatment samples compared to those in paired mid-treatment samples and paired post-treatment samples from patients in complete remission, and at primary progressive or relapsing disease in the population-based cohort. For graphical reasons the two extreme outliers with pre-treatment sCD163 levels over 11,000 ng/mL are not shown in the figure, but are included in the statistical analyses. (C) Correlation of pre-treatment sCD163 levels and CD163+ tumor-associated macrophages (TAM) in the tumor tissue in the trial cohort. (D) Correlation of pre-treatment sCD163 levels and CD163 gene expression levels from the matching tumor tissue in the trial cohort.

2502

haematologica | 2021; 106(9)

Letters to the Editor

(ELISA; R&D Systems, Minneapolis, MN, USA). In the population-based cohort, levels were measured as previously described.6 Measurements were performed independently in different laboratories. The median pre-treatment sCD163 values in the respective cohorts were defined a priori as a cutoff for testing prognostic implications. In the trial cohort, CD163 mRNA levels in 54 tumor

samples were measured with NanoString nCounter (Nanostring Technologies, Seattle, WA, USA).11 Proportions of CD163+ cells in 41 tumor samples were analyzed by multiplex immunohistochemistry11 (Online Supplementary Figure S1A, B). Blood monocyte counts were available for 103 patients. Statistical analyses were performed with IBM SPSS Statistics v.25.0 (IBM, Armonk, NY, USA) and STATA/IC

Table 1. The characteristics of all patients and patients divided by sCD163 pre-treatment levels above and below the median.

All

Total Median sCD163 (ng/mL) Median age Age <60 years 60–65 years >65 years Gender Male Female ECOG PS 0-1 2-3 Missing Stage 1-2 3-4 aaIPI score 0-1 2 3 Missing LDH ≤ ULN > ULN Missing Subtype DLBCL NOS GCB non-GCB ND Other Response CR PR PD NA

Trial cohort sCD163 sCD163 ≥ median, < median, n (%) n (%)

119 (100) 59 (50) 1160 (370-3621) 789 56 (21-65) 57 (21-65)

60 (50) 1707 55 (22-64)

All

Population-based cohort sCD163 sCD163 ≥ median, < median, n (%) n (%)

125 (100) 62 (50) 2950 (870-30000) 2050 67 (26-87) 64 (26-85)

63 (50) 4610 69 (30-87)

78 (65) 41 (35) 0 (0)

37 (63) 22 (37) 0 (0)

41 (68) 19 (32) 0 (0)

0.519

36 (29) 20 (16) 69 (55)

23 (37) 9 (15) 30 (48)

13 (21) 11 (17) 39 (62)

0.126

74 (62) 45 (38)

31 (53) 28 (47)

43 (72) 17 (28)

0.031

70 (56) 55 (44)

36 (58) 26 (42)

34 (54) 29 (46)

0.644

83 (70) 36 (30) 0 (0)

42 (71) 17 (29) 0 (0)

41 (68) 19 (32) 0 (0)

0.735

112 (90) 10 (8) 3 (2)

58 (94) 2 (3) 2 (3)

54 (86) 8 (13) 1 (2)

0.095

7 (6) 112 (94)

5 (9) 54 (91)

2 (3) 58 (97)

0.233

57 (46) 68 (54)

39 (63) 23 (37)

18 (29) 45 (71)

<0.001

9 (7) 71 (60) 39 (33) 0 (0)

4 (7) 36 (61) 19 (32) 0 (0)

5 (9) 35 (58) 20 (33) 0 (0)

0.842a

72 (58) 28 (22) 7 (6) 18 (14)

42 (68) 10 (16) 2 (3) 8 (13)

30 (48) 18 (29) 5 (8) 10 (16)

0.042b

10 (8) 109 (92) 0 (0)

6 (10) 53 (90) 0 (0)

4 (7) 56 (93) 0 (0)

0.491

73 (59) 49 (39) 3 (2)

43 (69) 17 (28) 2 (3)

30 (48) 32 (51) 1 (1)

0.009

48 (40) 37 (31) 16 (14) 18 (15)

32 (54) 13 (22) 7 (12) 7 (12)

16 (27) 24 (40) 9 (15) 11 (18)

0.004c

50 (40) 37 (30) 31 (25) 7 (5)

27 (43) 16 (26) 16 (26) 3 (5)

23 (37) 21 (33) 15 (24) 4 (6)

0.321c

89 (75) 21 (18) 4 (3) 5 (4)

42 (72) 13 (22) 1 (2) 3 (5)

47 (78) 8 (13) 3 (5) 2 (3)

0.315

107 (86) 9 (7) 6 (5) 3 (2)

57 (92) 5 (8) 0 (0) 0 (0)

50 (79) 4 (6) 6 (10) 3 (5)

0.013

The patients’ characteristics are shown, according to sCD163 pre-treatment levels, in the trial cohort and in a population-based cohort. ECOG PS: Eastern Cooperative Oncology Group performance status; aaIPI: age-adjusted International Prognostic Index; LDH: lactate dehydrogenase; ULN: upper limit of normal; DLBCL NOS: diffuse large B-cell lymphoma not otherwise specified; GCB: germinal center B-cell like; ND: not determined; CR: complete response; PR: partial response; PD: progressive disease; NA: not available. a Comparison between aaIPI score 2 and 3. bComparison between aaIPI score 0-1 and 2. cComparison between GCB and non-GCB.

haematologica | 2021; 106(9)

2503

Letters to the Editor

Figure 2. Association of survival outcomes with pre-treatment sCD163 levels. (A) Overall survival (OS) and (B) progression-free survival (PFS) according to pre-treatment sCD163 level (< vs. ≥ median, 1,160 ng/mL) in the trial cohort. (C) OS and (D) PFS according to pre-treatment sCD163 level (< vs. ≥ median, 2,950 mg/mL) in the population-based cohort.

12.1 (StataCorp LP, Texas, TX, USA). The Wilcoxon signed ranks test was used to evaluate changes in sCD163 levels during treatment. The Mann-Whitney Utest was used to compare sCD163 levels between patient and control groups. Spearman rank analysis was used in correlation analyses. The χ2 test and the Fisher-FreemanHalton test were used to evaluate differences in frequency of prognostic factors. The Kaplan-Meier method was used to estimate differences in outcome between the subgroups. The degree of significance was calculated using a log-rank test. Cox regression analysis (with 95% confidence intervals [95% CI]) was used and adjusted for gender and the variables in the International Prognostic Index, and in the trial cohort also for molecular subtype. P-values <0.05 were considered statistically significant. All statistical tests were two-tailed. The baseline characteristics of the patients in the trial cohort are presented in Table 1. The median follow-up for the patients alive was 61 months (range, 40-87). In total, 16 (13%) patients died and 18 (15%) relapsed during follow-up. Five-year overall survival and progressionfree survival rates were 87% and 83%, respectively. Pretreatment sCD163 levels were higher in these patients than in healthy volunteers (median 1,160 ng/mL [range, 370-3,621] vs. 437 ng/mL [range, 220-518]; P<0.001), and higher in the subgroup with non-germinal center B-cell like (non-GCB) lymphoma than in the GCB subgroup (Table 1, Online Supplementary Figure S1C), declining in response to therapy (median 975 ng/mL [range, 2992504

1,923], P<0.001) (Figure 1A, Online Supplementary Figure S1E, G). Pre-treatment sCD163 levels correlated with CD163+ TAM (Figure 1C) and with CD163 mRNA levels (Figure 1D), whereas no correlation was found with blood monocyte counts (ρ=0.0, P=1.00). High pre-treatment sCD163 levels (above the median) translated into poor outcome (Figure 2A, B). Relative risks of death and progression were, respectively, 3.4-fold (95% CI: 1.1210.51, P=0.031) and 2.7-fold (95% CI: 1.05-6.94, P=0.04) higher. In Cox regression analysis including International Prognostic Index factors, gender and molecular subtype, sCD163 remained the only significant prognostic factor for progression-free survival (hazard ratio [HR]=4.40, 95% CI: 1.09-17.83; P=0.038) (Online Supplementary Table S1). A similar trend was seen for poor overall survival (HR=5.08, 95% CI: 0.98-26.39; P=0.053) (Online Supplementary Table S1). The mean pre-treatment sCD163 levels in patients stratified by their later response to therapy are presented in Online Supplementary Figure S1I. The baseline characteristics of the population-based cohort are presented in Table 1. During a median followup of 40 months (range, 0-94), 42 (34%) patients died and 29 (23%) relapsed. The estimated 5-year overall and progression-free survival rates were 65% and 57%, respectively. The median pre-treatment sCD163 level was 2,950 ng/mL (range, 870-30,000 ng/mL). All cases later developing progressive disease had a diagnostic sCD163 value above the median (Table 1). Pre-treatment haematologica | 2021; 106(9)

Letters to the Editor

levels did not differ according to subtype (Online Supplementary Figure S1D). In the 71 patients with paired pre-treatment and post-treatment samples in complete response, sCD163 levels declined significantly (median 2,510 ng/mL [range 870-30,000] to 2,120 ng/mL [range 670-5,000], P=0.018) (Figure 1B, Online Supplementary Figure S1F). In 30 patients with paired pre- and mid-treatment samples, levels also declined, although not statistically significantly (Figure 1B, Online Supplementary Figure S1F, H). sCD163 levels in 11 patients with primary progressive or relapsing disease did not differ from the paired pre-treatment levels (Figure 1B). The distribution of pretreatment sCD163 levels in groups of patients stratified by their later response to treatment are presented in Online Supplementary Figure S1J. The outcomes in the subgroup with pre-treatment sCD163 levels above the median were worse than those with levels below the median (Figure 2C, D). The relative risks of death and progression were, respectively, 2.2-fold (95% CI: 0.99-4.94; P=0.052) and 2.2-fold (95% CI: 1.054.48; P=0.037) higher. In Cox regression analysis, sCD163 remained an independent prognostic factor for poor progression-free survival (HR=2.16, 95% CI: 1.054.48; P=0.037) (Online Supplementary Table S2) together with age, poor performance status, elevated lactate dehydrogenase concentration and male gender. A similar trend was seen for poor overall survival (HR=2.21, 95% CI: 0.99-4.94, P=0.052) (Online Supplementary Table S2). Taken together, we measured sCD163 levels in two independent DLBCL cohorts. Pre-treatment sCD163 levels correlated with CD163+ TAM and CD163 mRNA levels in the lymphoma tissue, while no correlation with monocyte counts was seen, suggesting that circulating sCD163 predominantly arises from the lymphoma tissue and that the elevated levels reflect host response to an aggressive lymphoma presentation. Pre-treatment sCD163 levels were elevated compared to those in healthy controls, and high levels were associated with unfavorable outcomes. We observed a decline in sCD163 levels in response to therapy, which in turn suggests that sCD163 could be used as a disease response biomarker in DLBCL. Similar observations have been previously made in chronic lymphocytic leukemia and multiple myeloma.6,7 The few samples at relapse prevented us from drawing firm conclusions, but the levels seemed in line with diagnostic values. The two different ELISA methods implemented in the cohorts have been compared in the past and their results showed a strong correlation (r2=0.97), but a systematic bias due to different calibration levels.12 This likely explains the difference in sCD163 levels between the two subsets, with higher levels observed in the population-based cohort even though the trial cohort had a larger number of patients with advanced disease. Another contributing factor may be the larger number of patients >60 years in the population-based cohort. An advantage of sCD163 as a potential biomarker is its stability in plasma, simplifying sample collection and handling.12 While absolute levels might differ between individuals for reasons other than tumor burden, levels could also be used as a patient-specific measure of response, indicated by declining levels in patients achieving complete remission. Indeed, the intraindividual biological variation in sCD163 is low, supporting the use of sCD163 for monitoring.12 While several prognostic factors are already used in clinical routine, disease monitoring tools in DLBCL are less common. Interim fluorodeoxyglucose positron emission tomography/computed tomography13 and down-modulation of circulating haematologica | 2021; 106(9)

tumor DNA14 are useful and promising for disease monitoring, but the impact of host-related factors such as the tumor microenvironment should not be ignored. Our results show that sCD163 is an indicator of biologically aggressive DLBCL. The findings were similar in two independent cohorts despite differences in clinical variables, implying that the prognostic impact of sCD163 is not limited to a particular population of patients. Therefore, these results suggest that sCD163 represents a useful and easily assessable biomarker for therapeutic monitoring of patients with DLBCL. Heli Vajavaara,1,2,3* Frida Ekeblad,4* Harald Holte,5 Judit Jørgensen,6 Suvi-Katri Leivonen,1,2,3 Mattias Berglund,4 Peter Kamper,6 Holger J. Møller,7,8 Francesco d’Amore,6,8 Daniel Molin,4 Gunilla Enblad,4 Maja Ludvigsen,6,8 Ingrid Glimelius4,9# and Sirpa Leppä1,2,3# *HV and FE contributed equally as co-first authors. # IG and SL contributed equally as co-senior authors. 1 Research Program Unit, Applied Tumor Genomics, Faculty of Medicine, University of Helsinki, Helsinki, Finland; 2Department of Oncology, Helsinki University Hospital Comprehensive Cancer Center, Helsinki, Finland; 3iCAN Digital Precision Cancer Medicine Flagship, Helsinki, Finland; 4Experimental and Clinical Oncology, Department of Immunology, Genetics and Pathology, Uppsala University, Uppsala, Sweden; 5Department of Oncology and KG Jebsen center for B-cell malignancies, Oslo University Hospital, Oslo, Norway; 6 Department of Hematology, Aarhus University Hospital, Aarhus, Denmark; 7Department of Clinical Biochemistry, Aarhus University Hospital, Aarhus, Denmark; 8Department of Clinical Medicine, Aarhus University, Aarhus, Denmark and 9Department of Medicine, division of clinical epidemiology, Karolinska Institutet, Solna, Sweden Correspondence: SIRPA LEPPA - sirpa.leppa@helsinki.fi INGRID GLIMELIUS - ingrid.glimelius@igp.uu.se doi:10.3324/haematol.2020.278182 Received: December 15, 2020. Accepted: March 17, 2021. Pre-published: March 25, 2021. Disclosures: HV has received honoraria from Roche (not related to this study). SL has received honoraria and research funding from Celgene/BMS, Cho Pharma USA, Incyte, GILEAD, Novartis, Roche, Takeda, Bayer and Janssen-Cilag (not related to this study). IG has received honoraria from Janssen-Cilag (not related to this study). DM received honoraria from Roche, Merck, Bristol-Myers Squibb, and Takeda (not related to this study). Other authors declare no conflicts of interest pertinent to the topic of this manuscript. Contributions: HV and FE designed and conceived the study, analyzed the data, and wrote the manuscript; JJ, HH, MB, DM, GE, IG and SL provided samples and clinical data; S-KL analyzed the data; PK, FdA, ML, DM and GE designed the study; HJM performed laboratory analyses; IG and SL designed and supervised the study and wrote the manuscript. All authors read, critically reviewed and approved the manuscript. Acknowledgments: the authors would like to thank Anne Aarnio and Marika Tuukkanen for technical assistance and Sara Ekberg for statistical support. Funding: the study was supported by grants from the Academy of Finland (to SL), Finnish Cancer Foundation (to SL), Juselius Foundation (to SL), Ida Montin Foundation (to HV), Finnish Society for Oncology (to HV), University of Helsinki (to SL), Helsinki University Hospital (to SL), Swedish Cancer Society (19 0123 Pj 01 H and 19 0109 SCIA) (to IG), Swedish Society of Medicine (to IG) and Lions Research Cancer Fund, Uppsala Sweden (to IG). 2505

Letters to the Editor

References 1. Salles G, Barrett M, Foà R, et al. Rituximab in B-cell hematologic malignancies: a review of 20 years of clinical experience. Adv Ther. 2017;34(10):2232-2273. 2. Harrysson S, Eloranta S, Ekberg S, et al. Incidence of relapsed/refractory diffuse large B-cell lymphoma (DLBCL) including CNS relapse in a population-based cohort of 4205 patients in Sweden. Blood Cancer J. 2021; 11(1):9. 3. Ziepert M, Hasenclever D, Kuhnt E, et al. Standard International Prognostic Index remains a valid predictor of outcome for patients with aggressive CD20+ B-cell lymphoma in the rituximab era. J Clin Oncol. 2010;28(14):2373-2380. 4. Kridel R, Steidl C, Gascoyne RD. Tumor-associated macrophages in diffuse large B-cell lymphoma. Haematologica. 2015;100(2):143-145. 5. Mosser DM. The many faces of macrophage activation. J Leukoc Biol. 2003;73(2):209-212. 6. Andersen MN, Abildgaard N, Maniecki MB, Møller HJ, Andersen NF. Monocyte/macrophage-derived soluble CD163: a novel biomarker in multiple myeloma. Eur J Haematol. 2014;93(1):41-47. 7. Nederby L, Roug AS, Knudsen SS, et al. Soluble CD163 as a prognostic biomarker in B-cell chronic lymphocytic leukemia. Leuk

2506

Lymphoma. 2015;56(11):3219-3221. 8. Wada N, Zaki MAA, Hori Y, et al. Tumour-associated macrophages in diffuse large B-cell lymphoma: a study of the Osaka Lymphoma Study Group. Histopathology. 2012;60(2):313-319. 9. Leppä S, Jørgensen J, Tierens A, et al. Patients with high-risk DLBCL benefit from dose-dense immunochemotherapy combined with early systemic CNS prophylaxis. Blood Adv. 2020;4(9):1906-1915. 10. Glimelius B, Melin B, Enblad G, et al. U-CAN: a prospective longitudinal collection of biomaterials and clinical information from adult cancer patients in Sweden. Acta Oncol. 2018;57(2):187-194. 11. Autio M, Leivonen S, Brück O, et al. Immune cell constitution in the tumor microenvironment predicts the outcome in diffuse large B-cell lymphoma. Haematologica. 2021; 106(3):718-729. 12. Møller HJ. Soluble CD163. Scand J Clin Lab Invest. 2012;72(1):1-13. 13. Sun N, Zhao J, Qiao W, Wang T. Predictive value of interim PET/CT in DLBCL treated with R-CHOP: meta-analysis. Biomed Res Int. 2015;2015:648572. 14. Kurtz DM, Scherer F, Jin MC, et al. Circulating tumor DNA measurements as early outcome predictors in diffuse large B-cell lymphoma. J Clin Oncol. 2018;36(28):2845-2853.

haematologica | 2021; 106(9)

Letters to the Editor

Dual intracellular targeting by ruxolitinib and the Mcl-1 inhibitor S63845 in interleukin-6-dependent myeloma cells blocks in vivo tumor growth Multiple myeloma remains an incurable malignancy with most patients experiencing relapse despite the introduction of novel therapies. While the first monoclonal antibodies have been approved for the treatment of myeloma, small molecule inhibitors of signaling pathways are still investigational. Although the concept of Janus kinase (JAK)/signal transducer and activator of transcription (STAT)3 inhibition in myeloma has shown promising results in preclinical studies, the efficacy of

JAK inhibitors as single agents seems to be limited.1 Ruxolitinib is a potent JAK1/2 inhibitor and approved for the treatment of patients with myeloproliferative disease and for graft-versus-host disease.2 While it has activity as a single agent in multiple myeloma, the combination with the myeloid cell leukemia (Mcl)-1 protein inhibitor S63845 resulted in superior survival in a preclinical in vivo model. The results obtained in the INA-6 xenograft model strongly support evaluation of the combination of JAK and Mcl-1 inhibition in humans. The JAK/STAT3 pathway is activated by cytokines of the gp130 family including interleukin (IL)-6 as the most prominent member with an established pathophysiological role in multiple myeloma.3,4 Ruxolitinib phosphate

Figure 1. Effects of ruxolitinib on malignant plasma cell growth and STAT3 phosphorylation in vitro and in vivo. (A) Inhibition of INA-6 growth in vitro by ruxolitinib is dose-dependent. Cells were cultured in the presence of 2.5 ng/mL interleukin-6 (IL-6) for 3 days and absorbance was measured in an MTS-based colorimetric assay as described elsewhere.8 The mean values of ten independent experiments, each performed in triplicate or quadruplicate, are shown. Error bars, standard deviation. The concentration at 50% inhibition was calculated with CalcuSyn software (Biosoft, UK). (B) Inhibitory effect of ruxolitinib on IL-6-stimulated proliferation of primary plasma cells from the peripheral blood of a patient with plasma cell leukemia. 3H-thymidine uptake was measured as described previously.5 (C) Ruxolitinib dose-dependently inhibits IL-6-induced STAT3 phosphorylation in INA-6 cells, as demonstrated by western blot analysis. INA-6 cells were starved of IL-6 and serum for 4 hours (h), treated with different concentrations of ruxolitinib for 2 h, and then stimulated with 10 ng/mL IL-6 (Gibco®/Life Technologies, Darmstadt, Germany) for 15 min. Control cells did not receive IL-6. Cropped blots are shown. (D) Induction of apoptosis by ruxolitinib as shown by annexin V-FITC/7-AAD staining (Beckman-Coulter) and flow cytometric analysis (FC500). Cells were cultured in IL-6 and different concentrations of ruxolitinib for 48 h and 72 h. Control cells (Ctrl.) did not receive IL-6 or ruxolitinib. (E) Inhibition of STAT3 phosphorylation in vivo. A single oral dose of ruxolitinib (60 mg/kg) was given to tumor-bearing mice (at day 27 or day 33 after cell inoculation). One control animal received vehicle (0.5 % w/v methylcellulose, day 33), one engrafted mouse remained untreated. Tumors were explanted 2 h after drug administration. One part of the cells was stimulated ex vivo with IL-6 (10 ng/mL) for 10 min (+), the other part remained unstimulated (-). Cell lysates were prepared for sodium dodecylsulfate polyacrylamide gel electrophoresis and western blot analysis. Cropped blots of cell lysates from the two control animals and two ruxolitinib-treated mice are shown.

haematologica | 2021; 106(9)

2507

Letters to the Editor

Figure 2. Effects of ruxolitinib, S63845, and their combination on inhibition of plasma cell growth in vitro and in vivo. (A) Inhibition of INA-6.Tu1 growth in vitro by ruxolitinib, S63845, or their combination. Cell growth was measured by an MTS-based colorimetric assay. Drugs were added at the indicated concentrations. The mean values of a representative experiment, performed in quadruplicate, are shown. Error bars, standard deviation. *Significant difference of the effect of the combination from the effect of either single drug (P<0.05, unpaired two-tailed t-test). The drug combination indices (CI) for experimental values at a constant ratio were calculated with the method of Chou and Talalay with CalcuSyn v2.0 software (Biosoft, UK). CI<1, synergistic effect; CI=1, additive effect; CI>1, antagonistic effect. Fa: affected fraction. (B) Treatment scheme of SCID mice with the combination of ruxolitinib and S63845. Treatment started at day 1 after intraperitoneal (IP) cell inoculation and continued for 10 consecutive days. Ruxolitinib was administered orally (PO) twice daily, with the time between two doses being approximately 6 h. S63845 was injected intravenously (IV) on days 1, 4, 7, and 10. Treatment with ruxolitinib or S63845 as single agents was performed accordingly with vehicle always used as a substitute for the second drug. (C) Survival of SCID mice treated with the combination of ruxolitinib and S63845 (red line) was superior (100% alive) to that of animals treated with ruxolitinib alone (green line; 43% alive; P=0.0325) or S63845 as a single agent (purple line; 50% alive; P=0.0514). The control group (black line) received vehicle (0% alive; P≤0.0001 against all other groups). There was no significant difference between the ruxolitinib- and the S63845-treated group (P=0.4768). P-values were calculated using the log-rank (Mantel-Cox) test: P<0.05 is considered significant. (D) Soluble interleukin-6 (IL-6) receptor levels in the serum of mice at the day of sacrifice. Animals with undetectable levels had no visible tumors and survived until the experiment was terminated.

salt (INC424; formerly INCB018424) was supplied by Novartis Pharma (Basel, Switzerland) and Incyte Corp. (Wilmington, DE, USA). Among a number of human myeloma cell lines, the IL-6-dependent INA-6 (established in our laboratory and described in detail elsewhere5) was chosen because cytokine pathways after gp130 stimulation are well characterized and the line is sufficiently sensitive to growth inhibition by ruxolitinib (Figure 1A and Table 1). A similar high sensitivity to ruxolitinib in the nanomolar range was observed for growth inhibition of IL-6-stimulated primary plasma cell leukemia cells (Figure 1B). In INA-6, the JAK inhibitor specifically abrogated IL-6-stimulated STAT3 phosphorylation while the MAPK pathway, which is constitutively activated by an N-RAS mutation,5 was not inhibited (Figure 1C). Concomitantly with signaling inhibition, ruxolitinib induced apoptosis in INA-6 cells in a dose-dependent manner (Figure 1D). These findings are consistent with the essential role of STAT3 for the survival of INA-6 cells6 and other plasma cells.7 The INA-6 xenograft model also seemed to be particularly suitable for evaluating ruxolitinib given the high in vivo activity of gp130 monoclonal antibodies.8 As pharmacodynamic studies on tumor-bearing mice demonstrate, the constitutive as well as (ex vivo) IL-6-stimulated STAT3 activation observed in tumors of untreated or vehicle-treated control mice were inhibited in vivo by one single oral dose of ruxolitinib (60 mg/kg) (Figure 1E). Other signaling pathways activated in INA-6 cells in vitro and in vivo, such as the MAPK pathway constitutively 2508

Table 1. IC50 values of ruxolitinib in myeloma cell lines.

Cell line JAK driven HEL (JAK2 V617F) IL-6 dependent INA-6 INA-6.Tu1 B9 Autonomous growth EJM JJN3 JK-6 L363 MM1.S NCI-H929 RPMI8226 U266

IC50 (µM) 0.8 0.15 0.85 0.6 2.67 >8* 4.19 >8* >8* >8* >8* >8*

Cell growth was measured by an MTS-based colorimetric assay and half maximal inhibitory concentration (IC50) values were calculated with CalcuSyn v2.0 software (Biosoft, UK). The erythroleukemia line HEL carrying the activating JAK2 V617F mutation served as a control. With the exception of the murine B9 hybridoma, all cell lines were of human origin. B9 was a kind gift from L. A. Aarden (Central Laboratory Blood Transfusion Service, Amsterdam, the Netherlands); MM1.S was kindly provided by Yu Tzu Tai (Dana Farber Cancer Institute, Boston, MA, USA); INA6, INA-6.Tu1 and JK-6 were established as described elsewhere.5,16 All other cell lines were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ), Braunschweig, Germany. *Highest concentration evaluated.

haematologica | 2021; 106(9)

Letters to the Editor

activated by an N-RAS mutation5 and the phosphorylation of the S6 ribosomal protein downstream of PI-3 kinase/AKT were not inhibited (Online Supplementary Figure S1). For the in vivo studies, a subline of INA-6 was used.5,8 In general, 25x106 INA-6.Tu1 cells were injected intraperitoneally into approximately 8-week-old female SCID/beige (C.B.-17.Cg-Prkdcscid Lystbg/Crl) mice (Charles River, Sulzfeld, Germany). All animal experiments were performed in strict adherence to German laws for animal welfare and were approved by the governmental authorities of Schleswig-Holstein. Animals were kept under specified pathogen-free conditions with free access to food and water in a light-dark cycle of 12 hours. Blocking one single survival pathway may not be sufficient to eradicate myeloma cells in their tumor environment.1 The choice of the anti-apoptotic Mcl-1 protein as a second target is based on the knowledge that Mcl-1 is a critical survival factor for myeloma cells and is upregulated by IL-6 produced in the bone marrow microenvironment in a STAT3-dependent manner.9-11 An additional pathway leading to Mcl-1 upregulation, involving phosphatase of regenerating liver (PRL)-3, has recently been identified.12 S63845, provided by Novartis, is a potent and selective BH3-mimetic with higher affinity for human than for murine Mcl-1.13 Ruxolitinib and S63845 were used in combination and the effects in vitro and in animal studies compared with those of the single agents. INA-6.Tu1 cell growth in vitro was dose-dependently inhibited by both drugs with a significantly greater effect in combination at higher concentrations (Figure 2A). For the in vivo study, INC424 was freshly formulated in 0.5% w/v methylcellulose (SigmaAldrich, M0430) in sterile water every 3 to 4 days. S63845 was freshly dissolved in 2% D-α-tocopherol polyethylene glycol 1000 succinate (vitamin E-TPGS) (Sigma-Aldrich) in 0.9% sodium chloride solution shortly before every application. SCID/beige mice were inoculated with INA-6.Tu1 cells as described above and treated for 10 consecutive days starting 1 day after injection of the cells (Figure 2B). Ruxolitinib was administered by oral gavage (60 mg/kg body weight) twice daily with a 6 h interval between the two doses. S63845 was injected intravenously at the dose of 25 mg/kg on days 1, 4, 7 and 10 according to the scheduling described previously.13 Mice were monitored regularly for signs of tumor growth. The survival time was defined as the time between cell inoculation and the day of sacrifice, which occurred before tumor burden caused paraplegia, cachexia, or any other signs of suffering. Animals without any signs of tumors were sacrificed at the end of the experiment on day 98 (Figure 2B). Treatment was well tolerated in all groups, as indicated by no body weight losses during the first 20 days (Online Supplementary Figure S2). The Kaplan-Meier survival analysis (Figure 2C) shows that all mice of the control group (n=8) developed overt plasmacytomas and had to be sacrificed before day 40. The median survival time in this group was 23 days. A significant delay in tumor growth was observed in four out of seven mice treated with ruxolitinib, while three mice did not show any signs of tumors until the end of the experiment on day 98, resulting in a significantly prolonged median survival time of 56 days (P<0.0001 by the log-rank test). Treatment of mice with the Mcl-1 inhibitor (n=8) prevented tumor growth in 50% of the animals and significantly prolonged the median survival time compared to that of the control group (P<0.0001). In mice treated with single agents, tumor growth seen in some of the animals was not caused by the development of drug haematologica | 2021; 106(9)

resistance, as the sensitivity to both ruxolitinib and S63845 was retained in explanted tumors (Online Supplementary Figure S3). Remarkably, none of the mice treated with the combination (n=6) showed any signs of disease; at day 98 the experiment was terminated and animals were found to be tumor-free. The combination therapy was significantly superior to treatment with ruxolitinib alone, as determined by the log-rank test (P=0.0325). In INA-6-bearing mice, human soluble IL-6 receptors (sIL-6R) accumulate in the blood representing a tool for the detection of minimal residual disease.5 sIL-6R levels were measured in the serum taken from all mice at the time of sacrifice (Figure 2D). Mice with overt plasmacytomas, i.e., all mice of the control group and four mice each in the ruxolitinib and in the S63845 treatment groups, had measurable sIL-6R levels of up to 180 ng/mL (by enzyme-linked immunosorbent assay; Diaclone, Besançon, France). sIL-6R was not detected in any of the mice with long-term survival. These results strongly indicate that these mice were indeed free of INA-6 tumors. Ruxolitinib is currently in early clinical evaluation for patients with relapsed/refractory multiple myeloma in combination with steroids, immunomodulatory drugs and proteasome inhibitors. Likewise, clinical trials with the highly selective Mcl-1 inhibitor S64315 (MIK665), a molecule resembling S63845, as well as other Mcl-1 inhibitors are underway.14 Inhibitors of JAK1/2 and Bcl-2 family proteins are synergistic in myeloid malignancies and are currently being evaluated.15 In myeloma, simultaneously targeting JAK/STAT3 and Mcl-1 may either disturb one single signaling pathway or, more likely, block more than one pathway to efficiently control myeloma cell growth in vivo. The use of ruxolitinib with an Mcl-1 inhibitor in clinical studies is warranted. Renate Burger, Anna Otte, Jan Brdon, Matthias Peipp and Martin Gramatzki Division of Stem Cell Transplantation and Immunotherapy, Department of Medicine II, University Medical Center SchleswigHolstein and University of Kiel, Kiel, Germany Correspondence: RENATE BURGER - r.burger@med2.uni-kiel.de doi:10.3324/haematol.2020.276865 Received: November 30, 2020. Accepted: April 13, 2021. Pre-published: April 22, 2021. Disclosures: no conflicts of interest to disclose. Contributions: RB designed and conducted experiments, analyzed data, and wrote the manuscript; AO was responsible for the design and institutional approval of the animal experiments; JB conducted the animal experiments; MP helped with the animal experiments, reviewed the manuscript and discussed the results; MG supervised the research project, revised the manuscript and discussed the results. Acknowledgments: the authors thank Kathrin Richter, Tanja Ahrens, Anna Böttiger, and the staff of the animal facility in Kiel for excellent technical assistance. Our special thanks go to Thomas Radimerski from Novartis, Basel, Switzerland, who supported us with the provision of the compounds and very helpful discussions.

References 1. Mughal TI, Girnius S, Rosen ST, et al. Emerging therapeutic paradigms to target the dysregulated JAK/STAT pathways in hematological malignancies. Leuk Lymphoma. 2014;55(9):1968-1979. 2. Quintás-Cardama A, Vaddi K, Liu P, et al. Preclinical characteriza-

2509

Letters to the Editor

tion of the selective JAK1/2 inhibitor INCB018424: therapeutic implications for the treatment of myeloproliferative neoplasms. Blood. 2010;115(15):3109-3117. 3. Klein B, Zhang XG, Lu ZY, Bataille R. Interleukin-6 in human multiple myeloma. Blood. 1995;85(4):863-872. 4. Kishimoto T, Akira S, Narazaki M, Taga T. Interleukin-6 family of cytokines and gp130. Blood. 1995;86(4):1243-1254. 5. Burger R, Guenther A, Bakker F, et al. Gp130 and ras mediated signaling in human plasma cell line INA-6: a cytokine-regulated tumor model for plasmacytoma. Hematol J. 2001;2(1):42-53. 6. Brocke-Heidrich K, Kretzschmar AK, Pfeifer G, et al. Interleukin-6dependent gene expression profiles in multiple myeloma INA-6 cells reveal a Bcl-2 family-independent survival pathway closely associated with Stat3 activation. Blood. 2004;103(1):242-251. 7. Catlett-Falcone R, Landowski TH, Oshiro MM, et al. Constitutive activation of Stat3 signaling confers resistance to apoptosis in human U266 myeloma cells. Immunity. 1999;10(1):105-115. 8. Burger R, Günther A, Klausz K, et al. Due to interleukin-6 type cytokine redundancy only glycoprotein 130 receptor blockade efficiently inhibits myeloma growth. Haematologica. 2017;102(2):381390. 9. Puthier D, Bataille R, Amiot M. IL-6 up-regulates Mcl-1 in human

2510

myeloma cells through JAK/STAT rather than Ras/MAP kinase pathway. Eur J Immunol. 1999;29(12):3945-3950. 10. Slomp A, Peperzak V. Role and regulation of pro-survival BCL-2 proteins in multiple myeloma. Front Oncol. 2018;8:533. 11. Gupta V, Matulis SM, Conage-Pough JE, et al. Bone marrow microenvironment-derived signals induce Mcl-1 dependence in multiple myeloma. Blood. 2017;129(14):1969-1979. 12. Abdollahi P, Vandsemb EN, Hjort MA, et al. Src family kinases are regulated in multiple myeloma cells by phosphatase regenerating liver-3. Mol Cancer Res. 2016;15(1):69-77. 13. 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. 14. Wei AH, Roberts AW, Spencer A, et al. Targeting Mcl-1 in hematologic malignancies: rationale and progress. Blood Rev. 2020;44:100672. 15. Karjalainen R, Pemovska T, Popa M, et al. JAK1/2 and BCL2 inhibitors synergize to counteract bone marrow stromal cellinduced protection of AML. Blood. 2017;130(6):789-802. 16. Meister S, Schubert U, Neubert K, et al. Extensive immunoglobulin production sensitizes myeloma cells for proteasome inhibition. Cancer Res. 2007;67(4):1783-92.

haematologica | 2021; 106(9)

Letters to the Editor

Clinical diagnostic value of telomere length measurement in inherited bone marrow failure syndromes Bone marrow failure (BMF) is characterized by a hypocellular marrow and encompasses a diverse group of inherited and acquired disorders. Inherited bone marrow failure syndromes (IBMFS) occur in approximately 5%– 30% of patients with BMF in pediatric cohorts and consist of more than 25 defined disease entities, including dyskeratosis congenita (DC), Fanconi anemia (FA), Diamond–Blackfan anemia (DBA), and Shwachman– Diamond syndrome (SDS).1 IBMFS are a heterogeneous group of disorders in which BMF is usually associated with physical abnormalities. The diagnosis of IBMFS previously relied on the recognition of characteristic clinical features. Recent diagnostic advances using next-generation sequencing have revealed that some patients initially diagnosed with idiopathic aplastic anemia (AA) had cryptic presentations of IBMFS.2 This issue is important as a more accurate diagnosis may improve treatment outcomes. Telomeres are the end segments of chromosomes: they are composed of long DNA repeats and a protein complex, and are essential for genome integrity. Germline mutations in genes involved in telomere biology can result in significantly short telomere length (TL) in

peripheral blood lymphocytes in patients with DC.3 Although there is a consensus on the usefulness of TL for screening for DC, but not for other IBMFS, several investigators have demonstrated that TL is excessively short in patients with AA4 and non-DC IBMFS,5 including FA, SDS, and DBA. To assess the diagnostic value of TL, we measured TL in 133 patients with BMF and compared it to that in patients with DC, non-DC IBMFS, and AA. We retrospectively studied 133 patients (68 male and 65 female) with BMF in Japan between 2013 and 2018. We collected peripheral blood samples at diagnosis from all patients, measured TL from peripheral blood lymphocytes, and performed targeted sequencing analysis covering 184 genes associated with IBMFS (Online Supplementary Table S1), as described in our previous studies.2,4 TL was measured by flow-fluorescence in situ hybridization (flow-FISH) using a Telomere PNA Kit (Dako Cytomation, Glostrup, Denmark) according to the manufacturer’s instructions. We calculated the ageadjusted relative TL in terms of the standard deviation (SD) from 71 normal, age-matched, healthy controls (median age, 29 years; range, 1-47 years) as previously described.4 In 112 of 133 (84%) patients, paroxysmal nocturnal hemoglobinuria-type granulocytes and red blood cells were also evaluated by flow cytometry.4 As thresholds for determining the presence of minor paroxysmal nocturnal hemoglobinuria clones, we defined >0.020% and >0.037% for CD11b+ CD55-

Table 1. Clinical characteristics and laboratory findings of patients with bone marrow failure.

All patients (N = 133)

DC (n = 11)

Non-DC IBMFS (n = 15)

AA (n = 107)

Age, years, median (range) 7 (0-22) 7 (1-19) 6 (0-15) 7 (0-22) Gender, n (%) Male 68 (51) 5 (45) 7 (47) 56 (52) Female 65 (49) 6 (55) 8 (53) 51 (48) Cytopenia, n (%) Unilineage cytopenia 4 (3) 1 (9) 3 (20) 0 (0) Bicytopenia 24 (18) 3 (27) 3 (20) 18(17) Pancytopenia 105 (79) 7 (64) 9 (60) 89 (83) Severity, n (%)* Moderate 67 (52) 7 (70) 10 (83) 50 (47) Severe 35 (27) 2 (20) 2 (17) 31 (29) Very severe 27 (21) 1 (10) 0 (0) 26 (24) WBC, ×109/L, median (range) 2.8 (0.3-12.8) 2.7 (1.7-5.3) 3.1 (1.7-7.8) 2.8 (0.3-12.8) ANC, ×109/L, median (range) 0.6 (0.0-5.4) 0.8 (0.6-1.7) 0.9 (0.2-5.4) 0.5 (0-5.2) ALC, ×109/L, median (range) 1.8 (0.1-8.2) 1.3 (0.5-4.3) 2.1 (1.2-5.9) 1.8 (0.1-8.2) Hb, g/dL, median (range) 8.0 (2.7-14.3) 8.4 (3.1-14.3) 6.6 (3.7-11.3) 8.1 (2.7-14.0) 9 Platelets, ×10 /L, median (range) 2.5 (0.2-40.9) 2.5 (0.3-7.5) 3.6 (0.2-40.9) 2.3 (0.2-38.0) ARC, ‰, median (range) 11.5 (0.0-57.0) 15.5 (4.0-32.0) 12.7 (1.2-39.1) 10.0 (0.0-57.0) Minor PNH clones, n (%) Positive 42 (32) 3 (27) 3 (20) 36 (34) Negative 70 (53) 5 (46) 10 (67) 55 (51) Not done 21 (15) 3 (27) 2 (13) 16 (15) Very short TL, < -2.19 SD, n (%) 31 (23) 10 (91) 4 (25) 17(16) Relatively short TL, < -1.71 SD, n (%) 44 (33) 10 (91) 9 (60) 25 (23) TL, SD, median (range) -0.96 (-5.73 to +4.00) -3.50 (-5.73 to +0.83) -1.89 (-4.74 to +2.05) -0.84 (-4.27 to +4.00)

P 0.675 0.851

<0.001

0.024

0.374 0.011 0.065 0.307 0.121 0.106 0.632

<0.001 <0.001 <0.001

DC: dyskeratosis congenital; IBMFS: inherited bone marrow failure syndromes; AA: aplastic anemia; WBC: white blood cell count; ANC: absolute neutrophil count; ALC: absolute lymphocyte count; Hb: hemoglobin; ARC: absolute reticulocyte count; PNH: paroxysmal nocturnal hemoglobinuria; TL: telomere length; SD: standard deviation; *excluding patients with unilineage cytopenia.

haematologica | 2021; 106(9)

2511

Letters to the Editor

Figure 1. Diagnostic flowchart and profiles of patients with bone marrow failure. (A) Diagnostic flowchart for inherited bone marrow failure syndromes (IBMFS) and aplastic anemia (AA). The diagnoses were based on clinical criteria, syndrome-specific laboratory tests, and genetic analysis using targeted sequencing. (B) Clinical and genetic profiles of 133 patients with bone marrow failure (BMF). Each column indicates one patient. DC: dyskeratosis congenita; FA: Fanconi anemia; DBA: Diamond-Blackfan anemia; SDS: Shwachman-Diamond syndrome; AA: aplastic anemia; IBMFS: inherited bone marrow failure syndromes; PNH: paroxysmal nocturnal hemoglobinuria; TL: telomere length.

2512

haematologica | 2021; 106(9)

Letters to the Editor

CD59+ granulocytes and glycophorin A+ CD55– CD59– erythrocytes, respectively. We diagnosed patients on the basis of a diagnostic flowchart (Figure 1A) developed using published diagnostic criteria for specific IBMFS and acquired AA.6,7 The severity of cytopenia was determined according to the Camitta severity criteria for AA.8 We divided the 133 patients into three groups: those with DC, those with non-DC IBMFS, and those with AA. All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan).9 Written informed consent was obtained from patients or their legal guardians. This study was approved by the ethics committee of the Nagoya University Graduate School of Medicine.

Table 1 shows the clinical characteristics of patients included in this study. The median age at diagnosis of the total cohort was 7 years (range, 0-22 years). Of the 133 patients, 105, 24, and 4 were diagnosed with pancytopenia, bicytopenia, and unilineage cytopenia (3 anemia and 1 thrombocytopenia), respectively. In patients with pancytopenia or bicytopenia, severity was assessed as very severe, severe, and moderate in 27, 35, and 67 patients, respectively. The median TL in all 133 patients was –0.96 SD (range, −5.73 to +4.00 SD). Using targeted sequencing, in 24 patients (18%) we detected 35 pathogenic variants (5 nonsense, 13 missense, 5 frameshift, 7 splice site, and 5 deletions) of known causative IBMFS genes, including TINF2 (n=6), TERT (n=3), FANCA (n=6),

haematologica | 2021; 106(9)

Figure 2. Comparison of peripheral blood lymphocyte telomere length in patients with bone marrow failure. (A) Comparison of standard deviations (SD) in telomere length (TL) in patients with dyskeratosis congenita (DC), non-DC inherited bone marrow failure syndromes (IBMFS), and aplastic anemia (AA). Kruskall-Wallis and Holm tests were used to investigate the relationships among the three groups. P-values < 0.05 were considered statistically significant. (B, C) The cut-off values for TL were set according to the optimal combination of sensitivity and false positive rate (1-specificity) derived from receiver operating characteristic curves, which determined <−2.19 SD (very short TL) as the optimal TL threshold for evaluating DC patients, and <−1.71 SD (relatively short TL) for evaluating IBMFS patients, including non-DC patients. AUC, area under the curve.

2513

Letters to the Editor

FANCG (n=3), RPL5 (n=2), RPS19 (n=1), RPS17 (n=1), SBDS (n=1), and BLM (n=1). Homozygous mutations were found in three patients (2 in FANCG and 1 in FANCA), compound heterozygous mutations in four patients (2 in FANCA and 1 each in FANCG and SBDS), hemizygous mutations in three patients in FANCA, and heterozygous mutations in 14 patients (6 in TINF2, 3 in TERT, 2 in RPL5, and 1 each in RPS17, RPS19, and BLM). Each patient’s genetic variants are shown in Online Supplementary Table S2. Out of the 133 patients and following the diagnostic flowchart (Figure 1A), 11 were diagnosed with DC (8%), 15 with non-DC IBMFS (11%), and 107 with AA (81%). The schematic representation of the results of gene analysis and the clinical features of IBMFS are shown in Figure 1B. Of the 11 patients with DC, nine were genetically diagnosed (6 with mutations in TINF2 and 3 with mutations in TERT), and those without diagnostic genetic mutations were diagnosed on the basis of clinical criteria. The 15 non-DC IBMFS cases consisted of nine FA, four DBA, one SDS, and one Bloom syndrome. All of these diagnoses were confirmed by the presence of germline mutations in IBMFS-related genes. Physical anomalies were observed in 11 of 15 (73%) patients. The individual clinical features and genetic results of the patients with IBMFS are shown in Online Supplementary Table S2. We compared the clinical characteristics of patients with DC, non-DC IBMFS, and AA (Table 1). The median age and gender distribution did not show significant differences among the three groups. Severe or very severe cytopenia was significantly more frequent (P=0.024) in AA cases (57/107, 53%) than in DC (3/10, 30%) and non-DC IBMFS cases (2/12, 17%). The median TL in patients with DC, non-DC IBMFS, and AA were −3.50 SD (range, −5.73 to +0.83 SD), −1.89 SD (range, −4.74 to +2.05 SD), and −0.84 SD (range, −4.27 to +4.00 SD), respectively (Figure 2A). Patients with DC had significantly shorter TL compared to those with non-DC IBMFS (P=0.031) and AA (P<0.001). Furthermore, patients with non-DC IBMFS tended to have shorter TL than those with AA (P=0.096). To validate the efficacy of TL measurement in diagnosing DC and IBMFS, receiver operating characteristic curves identified two cut-off values with the optimum sensitivity and false positive rate (1-specificity) combination, <−2.19 SD (for patients with DC) (Figure 2B) and <−1.71 SD (for patients with IBMFS) (Figure 2C), defined as “very short TL” and “relatively short TL,” respectively. For the diagnosis of patients with IBMFS, the TL cut-off value at −1.71 SD (relatively short TL) yielded a relatively high negative predictive value (0.921; 95% confidence interval [95% CI]: 0.873-0.958) and a moderately positive predictive value (0.432; 95% CI: 0.333-0.505). Of the total cohort, 44 patients (33%) were classified as having “relatively short TL”, which was significantly more frequent (P<0.001) in DC (10/11, 91%) and non-DC IBMFS (9/15, 60%) than in AA (25/107, 23%). Although germline mutations in TL maintenance genes are known to cause very short TL in patients with DC,3 several studies found that patients with AA also had shorter TL than healthy individuals.4 Furthermore, several cases of short TL in patients with non-DC IBMFS have been reported (Online Supplementary Table S3).3,5,10-15 Alter et al.5 previously reported that very short telomeres (<1st percentile of normal) were observed in five of 78 (6%) patients with non-DC IBMFS, although little overlap was seen in the distribution of TL between DC and non-DC IBMFS cases.

2514

In this study we measured TL by standard flow-FISH in a cohort of BMF patients comprehensively and genetically evaluated by next-generation sequencing. We defined TL < −1.71 SD of normal as a new criterion of “relatively short TL”; the proportions of patients who met this criterion were significantly higher in DC (91%) and non-DC cases (60%) than in AA cases (23%). These results suggest that TL measurement is useful as a screening test for DC and as a clinical diagnostic tool for non-DC IBMFS patients needing comprehensive genetic analysis. One limitation of this study is its small sample size: we found that 73% (n=26) of IBMFS cases and 23% (n=107) of AA cases in this cohort had “relatively short TL.” A power calculation to check “relatively short TL” effectiveness in diagnosing IBMFS concluded that the power of 0.998 was high enough to support the assumption of a sufficient number of cases in this study. A second limitation of the present study is the small number of non-DC cases (n=15), which was insufficient to discuss the significance of TL measurements in each IBMFS subtype. However, “relatively short TL” (<−1.71 SD) was observed in six of nine FA cases, two of four DBA cases, and one case of SDS (Figure 1B). The median TL were −1.84 SD (range, −4.74 to +2.05 SD), −0.89 SD (range, −2.83 to +1.21 SD), and −1.99 for FA, DBA, and SDS cases, respectively (Online Supplementary Table S2). These results support those in previous case reports demonstrating relatively short TL in FA, DBA, and SDS (Online Supplementary Table S3). Nevertheless, future studies in larger cohorts of patients are warranted. This study confirms that a relatively short TL was present in a significant proportion of patients with DC and non-DC IBMFS, indicating the clinical diagnostic value of TL measurement in identifying patients who need further testing, particularly comprehensive genetic analysis. Shunsuke Miwata,1 Atsushi Narita,1 Yusuke Okuno,1,2 Kyogo Suzuki,1 Motoharu Hamada,1 Taro Yoshida,1 Masayuki Imaya,1 Ayako Yamamori,1 Manabu Wakamatsu,1 Kotaro Narita,1 Hironobu Kitazawa,1 Daisuke Ichikawa,1 Rieko Taniguchi,1 Nozomu Kawashima,1 Eri Nishikawa,1 Nobuhiro Nishio,1,2 Seiji Kojima,1 Hideki Muramatsu1 and Yoshiyuki Takahashi1 1 Department of Pediatrics, Nagoya University Graduate School of Medicine and 2Center for Advanced Medicine and Clinical Research, Nagoya University Hospital, Nagoya, Japan Correspondence: HIDEKI MURAMATSU - hideki-muramatsu@med.nagoya-u.ac.jp YOSHIYUKI TAKAHASHI - ytakaha@med.nagoya-u.ac.jp doi:10.3324/haematol.2021.278334 Received: January 22, 2021. Accepted: April 13, 2021. Pre-published: April 22, 2021. Disclosures: no conflicts of interest to disclose. Contributions: SM, AN and HM performed laboratory analyses, gathered clinical information, designed and conducted the research, analyzed data, and wrote the paper; MI, AY, MW, KN, HK, DI, RT and YO performed laboratory analyses; MH and YT gathered clinical information; KS, NK, EN, NN and SK conducted the research; YT directed the research and wrote the paper. Acknowledgments: the authors acknowledge all the clinicians, patients, and their families involved in this study. The authors thank Ms. Yoshie Miura and Ms. Hiroko Ono for their valuable assistance.

haematologica | 2021; 106(9)

Letters to the Editor

References 1. Chhabra P, Bhatia P, Singh M, et al. Pediatric bone marrow failure: clinical, hematological and targeted next generation sequencing data. Blood Cells Mol Dis. 2021;87:102510. 2. Muramatsu H, Okuno Y, Yoshida K, et al. Clinical utility of nextgeneration sequencing for inherited bone marrow failure syndromes. Genet Med. 2017;19(7):796-802. 3. Du HY, Pumbo E, Ivanovich J, et al. TERC and TERT gene mutations in patients with bone marrow failure and the significance of telomere length measurements. Blood. 2009;113(2):309-316. 4. 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. 5. Alter BP, Giri N, Savage SA, Rosenberg PS. Telomere length in inherited bone marrow failure syndromes. Haematologica. 2015;100(1):49-54. 6. Marsh JCW, Ball SE, Cavenagh J, et al. Guidelines for the diagnosis and management of aplastic anaemia. Br J Haematol. 2009;147(1):43-70. 7. Shimamura A, Alter BP. Pathophysiology and management of inherited bone marrow failure syndromes. Blood Rev. 2010;24(3):101122.

haematologica | 2021; 106(9)

8. Camitta BM, Storb R, Thomas ED. Aplastic anemia (second of two parts): pathogenesis, diagnosis, treatment, and prognosis. N Engl J Med. 1982;306(12):712-718. 9. Kanda Y. Investigation of the freely available easy-to-use software “EZR” for medical statistics. Bone Marrow Transplant. 2013;48(3):452-458. 10. Ball SE, Gibson FM, Rizzo S, Tooze JA, Marsh JCW, Gordon-Smith EC. Progressive telomere shortening in aplastic anemia. Blood. 1998;91(10):3582-3592. 11. Hanson H, Mathew CG, Docherty Z, Mackie Ogilvie C. Telomere shortening in Fanconi anaemia demonstrated by a direct FISH approach. Cytogen Cell Genet. 2001;93(3-4):203-206. 12. Thornley I, Dror Y, Sung L, Wynn RF, Freedman MH. Abnormal telomere shortening in leucocytes of children with ShwachmanDiamond syndrome. Br J Haematol. 2002;117(1):189-192. 13. Li X, Leteurtre F, Rocha V, et al. Abnormal telomere metabolism in Fanconi’s anaemia correlates with genomic instability and the probability of developing severe aplastic anaemia. Br J Haematol. 2003; 120(5):836-845. 14. Pavesi E, Avondo F, Aspesi A, et al. Analysis of telomeres in peripheral blood cells from patients with bone marrow failure. Pediatr Blood Cancer. 2009;53(3):411-416. 15. Ong SY, Li ST, Wong GC, Ho AYL, Nagarajan C, Ngeow J. Delayed diagnosis of Shwachman Diamond syndrome with short telomeres and a review of cases in Asia. Leuk Res Rep. 2018;9:54-57.

2515

Letters to the Editor

Synergistic interaction between HDAC and MCL-1 inhibitors through downregulation of BCL-XL in multiple myeloma The development of novel therapies is the most important catalyst for the advancement in the treatment of patients with multiple myeloma (MM). Presently, a

number of new therapies including immunotherapeutic drugs, cellular therapies and BH3 mimetics are introduced into clinical practice. This steadily increasing number of effective treatment options makes it practically impossible to compare all available regimens and treatment concepts with each other in different settings. Hence, it becomes increasingly important to obtain a

Figure 1. Legend on following page.

2516

haematologica | 2021; 106(9)

Letters to the Editor

Figure 1. Concurrent MCL-1 and HDAC inhibition synergistically kills multiple myeloma cells in vitro via apoptosis induction. Cell viability of MM1.S cells 96 hours (h) after treatment with S63845 (A and D), venetoclax (B and E), A-1331952 (C and F) alone or in combination with either panobinostat (A to C) or ricolinostat (D to F). Results are presented relative to 0.1% dimethyl sulfoxide (DMSO) control. The combination index was calculated and stated as a range. Combination index values of <0.8, 0.8–1.2, and >1.2 were interpreted as synergistic, additive, and antagonistic drug activity, respectively. Apoptosis induction in MM.1S, U266 and KMS-12-BM cells was assessed by 7AAD/Annexin V staining 72 h after treatment in the absence (G to I) or presence of MSCT+ stromal cells (J to L). (M) Cytochrome c release assay was performed 24 h after treatment initiation at the indicated concentrations. One representative experiment of two is shown. (N) Assessment of cleaved caspase 3 and cleaved PARP via flow cytometry was performed 48 h or 72 h post treatment induction, respectively. Error bars indicate standard deviation of the mean (SDM) of triplicate experiments. Differences between groups were calculated with one-way ANOVA, corrected for multiple comparison with Bonferroni-Holm correction with ****P<0.0001, **P<0.001 and *P<0.05.

Figure 2. Legend on following page.

haematologica | 2021; 106(9)

2517

Letters to the Editor

Figure 2. BCL-XL overexpression sequesters BIM and BAK and impairs S63845 mediated apoptosis induction. (A) Immunoblot analysis of the indicated BCL-2 family members was performed 24 hours (h) post-treatment initiation. One representative experiment of three independent biological replicates is shown. (B) BAK activation in MM.1S was determined by staining with antibodies against its active form. (C) MM.1S cells transduced with pcW57.1 EGFP (left panel) or pcW57.1 BCL-XL (right panel) were treated for 24 h with 0.5 μg/mL doxycycline to induce protein overexpression and afterwards exposed to the indicated treatments. Apoptotic cells were assessed 24 h post- treatment induction. Results indicate the mean +/- standard deviation of the mean (SDM) of three independent experiments. Differences between groups were calculated with one-way ANOVA, corrected for multiple comparison with Bonferroni-Holm correction with ****P<0.0001, **P<0.001 and *P<0.05. (D and G) Co-immunoprecipitation experiments in MM.1S cells transduced with either pcW57.1-EGFP (right panels) or pcW57.1-BCL-XL (left panels) were performed after 24 h pretreatment with 0.5 μg/mL doxycycline to induce protein overexpression and subsequent drug exposure for 24 h.

deeper understanding of the mechanism of activity of individual drugs to enable the optimal selection of combination partners and treatment sequences in clinical practice.1 In MM, the anti-apoptotic BH3 family member MCL1 was shown to act as a master regulator of cell survival and resistance to therapy.2,3 Accordingly, several MCL-1 inhibitors, such as S63845, S64315, AMG176 and AZD5991 are currently under evaluation in clinical trials.4 However, elevated expression of either BCL-XL or BCL-2 may affect the activity of MCL-1 inhibitors (MCL-1i).5 This suggests that the simultaneous targeting of multiple anti-apoptotic proteins might significantly enhance the activity of BH3 mimetics and overcome intrinsic as well as acquired drug resistance. In this context, a deregulation of BH3 protein family members by HDAC inhibitors (HDACi) was reported in MM6 making these compounds attractive combination partners for BH3 mimetics. Here, we aimed to evaluate synergistic or additive combination approaches for selected BH3 mimetics. We assessed the activity of the pan-HDACi panobinostat or HDAC6i ricolinostat in combination with inhibitors targeting either BCL-2 (venetoclax), BCL-XL (A-1331952) or MCL-1 (S63845) in a panel of MM cell lines (MM.1S, KMS-12-BM, MOLP-8, U266, SKMM-1, RPMI-8226, OPM-2, NCI-H929). Interestingly, in six of eight MM cell lines (KMS-12-BM, MM.1S, U266, MOLP-8, NCIH929, OPM-2) a synergistic or additive effect was observed when combining S63845+HDACi (Figure 1A to D; Online Supplementary Figure S1A to H). The combination of either venetoclax or A-1331952 with HDACi led to synergistic or additive activity in three and four cell lines, respectively (Figure 1B and E; Online Supplementary Figure S1I to N). The observed synergism was confirmed with alternative MCL-1 inhibitors (AZD5991, AMG-701) (data not shown) and translated into a significant increase in apoptosis in MM.1S, U266 and KMS-12-BM monoculture experiments using non-lethal concentrations of S63845, panobinostat and ricolinostat (Figure 1G to I). Similar effects were observed in co-culture experiments with MSCT+ stromal cells (Figure 1J to L). Additional validation experiments confirmed the observed augmentation of the apoptotic signaling cascade including an enhanced release of cytochrome c (Figure 1M), cleavage of caspase 3 and PARP (Figure 1N; Online Supplementary Figure S1O and P) in all cell lines analyzed. No cell cycle alterations were observed upon single agent or combinational therapy (Online Supplementary Figure S2C to E). The combination of S63845+HDACi proved to be particularly pronounced in the BCL-XL (co)-dependent MM cells MM1.S and U266,3,7 which otherwise did either not respond at all or only barley responded to single-agent MCL-1 inhibition.8 Moreover, BCL-XL is not only a major driver of intrinsic but also acquired MCL-1 inhibitor resistance,9 as well as dual MCL1/BCL2 inhibition.10 Hence, concurrent BCL-XL inhibition seems to 2518

optimally augment the efficacy of MCL-1 inhibitors, but prior clinical trials aiming to directly inhibit BCL-XL failed due to untoward toxicity.11,12 Based on these results we evaluated whether deregulation of pro- or anti-apoptotic BCL-2 family proteins by HDAC inhibitors explains the observed synergism. Single-agent treatment with S63845 monotherapy led to the accumulation of MCL-1 and BCL-XL in MM.1S (Figure 2A) and U266 (Online Supplementary Figure S2A), but not in KMS-12-BM cells (Online Supplementary Figure S2B). Conversely, combined MCL-1+HDAC inhibition led to the downregulation of BCL-XL and MCL-1 protein levels in all tested cell lines prior to the onset of apoptosis (Figure 2B; Online Supplementary Figure S2A and B). In addition, a significant increase in BAK activation in KMS-12-BM cells (Online Supplementary Figure S1R) was noted. In MM.1S and U266 cell lines BAK is already fully activated by S63845 alone and the combination of MCL-1i and HDACi did not augment it any further (Figure 2B; Online Supplementary Figure S1M to Q), suggesting that active BAK is kept under control by alternative anti-apoptotic family members – most likely BCL-XL. In order to test our assumption that BCL-XL inhibits apoptosis induction by S63845, we transduced MM cell lines with the Tet-on pcW57.1 vector harboring either full length BCL-XL or EGFP (control) and assessed apoptotic cells via flow cytometry. In MM.1S-BCL-XL cells, apoptosis significantly decreased upon treatment with S63845+HDACi (left panel of Figure 2C) compared to MM.1S-EGFP cells (right panel of Figure 2C). Similar findings were obtained in U266 and KMS-12-BM cells (Online Supplementary Figure 2F and G). In order to explore this rescue mechanism further, we examined the binding kinetics of BAK and BIM to BCL-XL via coimmunoprecipitations. Upon doxycycline-mediated induction of BCL-XL or EGFP protein, we treated the cells for 24 hours (h) with S63845 alone or in combination with either panobinostat or ricolinostat. Strikingly, BCL-XL overexpression in S63845+HDACi-treated cells resulted in an increased association of BCL-XL and BAK as well as BIM in all investigated cell lines (left panel of Figure 2D to G; Online Supplementary Figures 2H and I and S3A to D). On the contrary, in EGFP expressing cells, BIM and/or BAK binding to BCL-XL strongly decreased upon combined S63845+HDACi exposure as compared to S63845 treatment alone (right panel of Figure 2D to G; Online Supplementary Figures S2H and I and S3A to D). Noteworthy, we also observed cell line-specific and combination-specific effects such as an exclusive impact of ricolinostat or panobinostat on BIM-BCL-XL, but not BAK-BCL-XL binding, in MM.1S-EGFP and U266-EGFP cells, respectively (right panel of Figure 2F and G; right panel of Online Supplementary Figure 2H and I). Furthermore, in KMS-12-BM-EGFP cells BIM was not associated with BCL-XL (right panel of Online Supplementary Figure S3C), as BIM is rather sequestered by BCL-2 (data not shown) in line with the MCL-1/BCLhaematologica | 2021; 106(9)

Letters to the Editor

Figure 3. Ricolinostat promotes the activity of S63845 independent of HDAC6 inhibition. (A) MM.1S cells were either treated with panobinostat or ricolinostat for 24 hours (h), then whole-cell lysates were blotted for the indicated proteins. (B) Single-cell clones of MM.1S cells harboring a Tet-on miR-E vector expressing either a short hairpin RNA (shRNA) targeting Renilla (control) or an shRNA targeting HDAC6 were exposed to 0.3 μg/mL doxycycline for 48 h and whole-cell lysates were blotted for the indicated proteins. Western blots are representative for three independent experiments. (C and D) Cells were pretreated with doxycycline for 48 h and viability was assessed after an additional 48-h treatment with increasing concentrations of S63845 as indicated in the Figure. Results show the mean +/- standard deviation (SD) of three independent experiments performed in triplicates. (E and J) CD138 purified plasma cells sorted from patients with multiple myeloma (MM) were exposed to S63845, panobinostat or ricolinostat the respective combinations for 20 h (E and H) or 10 h (I and J). (E and H) Apoptotic cells were determined via Annexin V/7AAD positive staining. (I and J) Whole-cell lysates of primary MM cells were blotted for the indicated proteins. Short-term exposure of primary patient samples was chosen to avoid spontaneous cell death. (K) Proposed model of the underlying mechanism of the observed synergism between MCL-1i and HDACi in MM. In S63845 treated cells BCL-XL is capable to sequester BAK, hence inhibiting the apoptotic signaling cascade resulting in diminished MM cell death. In combination with HDACi, BCL-XL protein is downregulated and thereby BAK, released by MCL-1i, can be activated, can oligomerize and can insert into the MOMP which in turn releases cytochrome c and activates initiator as well as effector caspases, leading to PARP cleavage and cell death.

haematologica | 2021; 106(9)

2519

Letters to the Editor

2 dependency of this t(11;14) carrying cell line.3 In conclusion, these results strengthen a model where BCL-XL is capable of sequestering BAK and BIM released by S63845, thus prohibiting the onset of the apoptotic signaling cascade. We next aimed to confirm the impact of panobinostat and ricolinostat on histone and tubulin acetylation as mechanism of synergy with MCL-1 inhibitors. Panobinostat is expected to strongly increase histone acetylation, while ricolinostat, which is selectively targeting HDAC6,13,14 should increase acetylated α-tubulin without altering acetylated histone levels.15 We treated MM.1S, U266 and KMS-12-BM cells with both HDACi and assessed acetylated and total protein expression levels of α-tubulin and histone-3 after 24 h. Panobinostat strongly increased acetyl-histone-3 in all three cell lines (Figure 3A; Online Supplementary Figure S3E and F). Contrary to our expectations, ricolinostat treatment likewise led to a strong increase in acetyl-histone-3 besides elevating acetyl-α-tubulin levels (Figure 3A; Online Supplementary Figure S3E and E), indicating that ricolinostat has off-target effects on additional HDAC family members. Accordingly, we aimed to investigate whether the synergism of ricolinostat and S63845 is facilitated via HDAC6 inhibition or via the epigenetic off-target effect. To this end, we performed viability assays with MM.1S and U266 cells transduced with a Tet-on miR-E plasmid harboring a control short hairpin RNA (shRNA) (Renilla), or HDAC6 shRNA. HDAC6 knockdown was confirmed 48 h after induction with 0.3 μg/mL doxycycline (Figure 3B; Online Supplementary Figure S3G). However, S63845 significantly decreased cell viability regardless of the presence of HDAC6 knockdown (Figure 3C and D; Online Supplementary Figure S3H to J). These results suggest that the synergism between S63845 and ricolinostat is due to the unspecific epigenetic effect of ricolinostat. In order to validate our findings in primary MM cells we treated CD138-selected MM cells ex vivo for 20 h with single-agent HDACi and S63845 as well as the corresponding combinations before evaluating apoptosis induction. This demonstrated an increase of apoptotic cells upon combination treatment with S63845+HDACi in three of four samples tested, whereas the magnitude was highly variable (Figure 3E to H). Unfortunately, we were unable to collect sufficient cell material to establish a link between BCL-2 family dependencies and combination activity. However, we investigated whether the synergism was accompanied by a downregulation of BCL-XL in MM patient samples ex vivo. For this purpose, we treated patient samples for 10 h with S63845 alone or in combination with HDACi and determined BCL-XL protein expression. In both analyzed patient samples, downregulation of BCL-XL protein expression was observed in the MM cell lines (approximately 30% vs. S63845 single-agent treatment) upon S63845 combination with either panobinostat or ricolinostat, respectively. (Figure 3I to J). This suggests that the combination of MCL-1 inhibitors with HDACi is capable to tackle both, MCL-1 and BCL-XL, in MM patient cells. However, our findings need to be confirmed in enlarged patient cohorts and advanced in vivo models (i.e., carrying humanized MCL1)4 to better evaluate the clinical potential of our results as well as to define patient stratification markers. In conclusion, our findings support a model where BCL-XL sequesters BAK/BIM released in response to MCL-1 inhibition, particularly in tumor clones with 2520

baseline BCL-XL functionality such as MM.1S and U266 cells. By combining MCL-1i with HDACi, BCL-XL protein is downregulated, leading to unrestrained BAK activation and initiation of the apoptotic signaling cascade (Figure 3K). Previous efforts to pharmacologically target BCL-XL failed due to its role in megakaryopoiesis.12,11 Hence, our findings point towards an alternative opportunity to indirectly tackle BCL-XL/MCL-1 co-dependent MM cells by combining MCL-i1 and HDACi and highlight the importance of exploring various options of apoptosis induction for designing new treatment concepts for clinical evaluation Anja Schneller, Niklas Zojer,* Arnold Bolomsky* and Heinz Ludwig* *

NZ, AB and HL contributed equally as co-senior authors

Department of Medicine I, Wilhelminen Cancer Research Institute, Clinic Ottakring, Vienna, Austria Correspondence: HEINZ LUDWIG - heinz.ludwig@extern.gesundheitsverbund.at doi:10.3324/haematol.2020.277152 Received: November 30, 2020. Accepted: April 20, 2021. Pre-published: April 29, 2021. Disclosures: no conflicts of interest to disclose. Contributions: AS designed and performed experiments and analyzed results; AB encouraged the investigation, designed experiments and provided intellectual input and expertise; NZ provided supervision and expertise; HL provided supervision and secured funding; AS and AB wrote the manuscript. Acknowledgments: the authors would like to thank Waltraud Scherbler and Martin Schreder for providing primary patient material and Anna Walzl for excellent assistance. Funding: this study was funded by the Wilhelminen Cancer Research Institute, the Austrian Forum Against Cancer and the Austrian Academy of Science (# 25542). AS is the recipient of a DOC Fellowship of the Austrian Academy of Sciences at the Wilhelminen Cancer Research Institute.

References 1. Rajkumar SV. Multiple Myeloma: 2016 update on diagnosis, riskstratification and management. Am J Hematol. 2016;91(7):719734. 2. Robillard N, Gomez P, Moreau P, Gouill S Le, Harousseau J, Amiot M. Mcl-1 is overexpressed in multiple myeloma and associated with relapse and shorter survival. Leukemia. 2005;19(7):1248-1252. 3. Gong J, Khong T, Segal D, et al. Hierarchy for targeting prosurvival BCL2 family proteins in multiple myeloma : pivotal role of MCL1. Blood. 2016;128(14):1834-1844. 4. Bolomsky A, Vogler M, Köse MC, et al. MCL - 1 inhibitors , fast - lane development of a new class of anti - cancer agents. J Hematol Oncol. 2020;13(1):173. 5. Hockings C, Alsop AE, Dewson G, et al. Mcl-1 and Bcl-x L sequestration of Bak confers differential resistance to BH3-only proteins. Cell Death Differ. 2018;25(4):721-734. 6. Ramakrishnan VG, Miller KC, Macon EP, et al. Histone deacetylase inhibition in combination with MEK or BCL-2 inhibition in multiple myeloma. Haematologica. 2019;104(10):2061-2074. 7. C Touzeau, J Ryan, J Guerriero, et al. BH3 profiling identifies heterogeneous dependency on Bcl-2 family members in multiple myeloma and predicts sensitivity to BH3 mimetics. Leukemia. 2016;30(3):761-764. 8. 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. 9. Patricia Gomez-Bougie, Sophie Maïga, Benoit Tessoulin, et al. Respective place of venetoclax and MCL1 BH3 mimetics in mul-

haematologica | 2021; 106(9)

Letters to the Editor

tiple myeloma treatment. Blood. 2018;132(25):2656-2669. 10. Seiller C, Maiga S, Touzeau C, et al. Dual targeting of BCL2 and MCL1 rescues myeloma cells resistant to BCL2 and MCL1 inhibitors associated with the formation of BAX / BAK heterocomplexes. Cell Death Dis. 2020;11(5):316. 11. Wilson WH, Connor OAO, Czuczman MS, et al. Navitoclax , a targeted high-affi nity inhibitor of BCL-2 , in lymphoid malignancies: a phase 1 dose-escalation study of safety , pharmacokinetics , pharmacodynamics, and antitumour activity. Lancet Oncol. 2010; 11(12):1149-1159. 12. Mason KD, Carpinelli MR, Fletcher JI, et al. Programmed anu-

haematologica | 2021; 106(9)

clear cell death delimits platelet life span. Cell. 2007;128(6):11731186. 13. Santo L, Hideshima T, Kung AL, et al. Preclinical activity , pharmacodynamic, and pharmacokinetic properties of a selective HDAC6 inhibitor, ACY-1215 , in combination with bortezomib in multiple myeloma. Blood. 2019;119(11):2579-2590. 14. Carew JS, Espitia CM, Zhao W, et al. Rational cotargeting of HDAC6 and BET proteins yields synergistic antimyeloma activity. Blood Adv. 2019;3(8):1318-1329. 15. Hubbert C, Guardiola A, Shao R. HDAC6 is a microtubule-associated deacetylase. Nature. 2002;417(6887):455-458.

2521

Letters to the Editor

Myeloma natural killer cells are exhausted and have impaired regulation of activation Multiple myeloma (MM) is an immunotherapy responsive disease. Treatment strategies including immunemodulatory drugs lenalidomide and pomalidomide, bi-specific t-cell engagers (BiTE), and antibodies targeting myeloma surface proteins SLAMF7 (elotuzumab) or CD38 (daratumumab and isatuximab) and chimeric antigen receptor T cells have been effective.1-4 Currently, myeloma-targeting antibodies against CD38 and SLAMF7 mediate their effect in part, via natural killer (NK) cells as key effectors.1,2 However, NK cells from myeloma patients have decreased functional responses to myeloma in vitro.5 Despite this, myeloma targeting antibodies that are reliant on NK-cell mediated cytotoxicity have been successful in treating MM patients.2 In order to understand this further, we explored NKcell differentiation and function in newly diagnosed MM patients (NDMM) and, for the first time, gene expression profiles of NK-cell subsets from refractory relapsed MM (RRMM) patients. These analyses revealed that underlying NK-cell intrinsic properties explain this myeloma patient NK-cell dysfunction. We also characterized whether NK-cell dysfunction was rescued following induction therapy with lenalidomide and dexamethasone and post-autologous stem cell transplantation (ASCT) to understand whether myeloma-targeting antibodies such as elotuzumab could be used at these time points. We compared peripheral blood and bone marrow NK cells from a NDMM patient cohort consecutively treated in the context of a prospective phase II clinical LITVACC trial (clinicaltrials gov. Identifier: ACTRN12613000344796)6 or RRMM patient cohort from the RevLite trial (clinicaltrials gov. Identifier: NCT00482261)7 to healthy donor (HD) NK cells (for details see Online Supplementary Figure S1). We first confirmed that RRMM and NDMM patients have a higher percentage of terminally differentiated CD57+ NK cells compared to HD both in the peripheral blood and bone marrow (Online Supplementary Figure S1C and D). These RRMM NK cells are dysfunctional. In contrast, HD CD56dimCD16+KIR+CD57+ NK cells are highly cytotoxic and secrete increased levels of interferon-γ (IFN-γ) in response to contact with targets.8 In order to explore reasons for this difference, principal component analysis of RNA sequencing data showed RRMM patient NK-cell gene expression profile (GEP) was distinct from HD NK cells (Figure 1A; Online Supplementary Figure S2A). Differential GEP analysis revealed numerous genes either down- or up-regulated in patient or HD CD57+ NK cells (Online Supplementary Figure S2B). When CD57+ NK cells from myeloma patients and HD were compared, we revealed differentially expressed genes (DEG) (n=133 and 533 DEG respectively), where 97 DEG were common to both RRMM patients and HD (Online Supplementary Figure S2C). Of the 36 DEG unique to patient CD57+ NK cells, 13 were up-regulated and 23 were down-regulated (Online Supplementary Figure S3C). When NK cell-specific genes were examined, we found decreased expression of genes associated with CD16 cleavage such as ADAM17 in RRMM patient NK cells, increased expression of genes associated with cytotoxicity and activation such as PRF1, GZMB, NCR1, NCR2, and increased expression of novel immune checkpoint genes, CISH and TIGIT (Figure 1B; Online Supplementary Figure S2E). Cytokine-inducible SH2-containing protein (CIS, encod2522

ed by CISH) is a critical negative regulator of IL-15 signaling and inhibits cytotoxicity against tumor cells.9 Gene set enrichment analysis (GSEA) revealed genes related to NK-cell activation pathways were significantly up-regulated in RRMM patient NK cells compared to HD NK cells, suggesting that NK cells from patients are constitutively more activated (Figure 1C). This finding was also true when comparing NK-cell activation pathways between RRMM patient and HD CD57– NK cells or CD57+ NK cells (Figure 1C and D). However, genes related to pathways regulating NK-cell activation (IL23A, IL23R, GAS6, IL18, IL15, AXL, FLT3LG, TICAM1 and PLDN) were downregulated in CD57+ NK cells from RRMM patients, suggesting dysregulation of patient NK cell activation (Figure 1C and E). GSEA enrichment plots highlight significantly increased MM patient NK-cell activation, yet co-existing suppression of positive regulation of these activation pathways (Figure 1F). ADAM17 transcript levels also correlated negatively with NK-cell activation in RRMM patients as compared to HD (Online Supplementary Figure S2D). Taken together, these data indicate MM patient CD57+ NK cells are constitutively more activated than their normal donor counterparts. However, they lack expression of key regulators of NKcell activation and have increased levels of the NK-cell immune checkpoint molecules CIS and TIGIT, suggesting an ‘exhausted’ state. We next explored whether NK-cell chronic activation and low levels of ADAM17 observed in the GEP data in Figure 1 would affect the capacity of NK cells to respond via CD16 or SLAMF7 mediated signaling. In order to do this, peripheral blood mononuclear cells (PBMC) from NDMM and RRMM patients and HD were co-cultured with OPM2 myeloma targets and the anti-human SLAMF7 antibody, elotuzumab. In this context, activated NK cells were expected to down-regulate CD16 due to cleavage by ADAM17,10 and this would be evident by a reduction in the percentage of CD56dimCD16+ NK cells. Only HD NK cells significantly decreased the percentage of CD56dimCD16+ NK cells in response to elotuzumab (Figure 2A and B, left panel), which was inhibited in the presence of an ADAM17 inhibitor (Figure 2C). Whilst there was a trend to decreased CD56dimD16+ NK cells in NDMM patients, this did not reach significance. We observed a similar result for terminally differentiated CD56dimCD57+ NK cells. HD NK cells were responsive to activation via OPM2 cells and elotuzumab and significantly reduced the percentage of CD56dimCD57+ NK cells (Figure 2B, right panel). In the same conditions, untreated NDMM NK cells showed a trend for decreased percentage of CD56dimCD57+CD16+ NK cells (P=0.051), whereas RRMM NK cells were relatively unresponsive. No difference was observed in the percentage of CD56dimCD16+ NK cells in RRMM patients in the presence of ADAM17 inhibitor (Figure 2C). Prior studies demonstrated no loss of NK cells in PBMC treated with elotuzumab at higher concentrations than used in our assays,11 suggesting fratricide was unlikely to occur. Our data supports this as the SLAMF7 levels on NK cells between MM patients and HD were similar (Online Supplementary Figure S3A). Subsequently, NK cell subsets were examined for degranulation (CD107a+) in the presence of OPM2 cells and elotuzumab. Of all NK cell subsets, only the CD56dimCD16– NK cells degranulated at significantly higher levels in HD compared to both groups of MM patients (Figure 2C; Online Supplementary Figure S3B). A similar trend was also observed for the HD versus myeloma patient CD57+ NK cells, but this did not reach significance. These results suggest that low levels of ADAM17 may lead to constihaematologica | 2021; 106(9)

Letters to the Editor

Figure 1. Gene expression analysis of natural killer (NK) cell subsets from refractory relapsed multiple myeloma patient and donor peripheral blood mononuclear cells reveals increased activation but loss of regulatory pathways in myeloma patient CD57+ NK cells. Refractory relapsed multiple myeloma (RRMM) patient and healthy donor (HD) NK cells were FACS-sorted to CD57+ and CD57– subsets, RNA extracted and RNA sequencing performed using the SMART-seq v4 low input RNA kit (Takara Bio USA) and sequenced on the NextSeq 550 sequencing system (Illumina, USA). The 36 samples, each containing on average 14,496,483 reads, were aligned using seqliner v0.7.1 to hg19 reference genome and quantified using Htseq v0.6.1 software. Normalization and differential expression analysis was performed with Limma-Voom in R v3.3.3 on a total of 20,850 genes. (A) Overarching differences in HD and myeloma NK cell subset GEP are depicted in two-dimensional principal component analysis (PCA) of patient or HD in four groups (n=6 per group). (B) Normalized log2 counts-per-million (cpm) transcript levels of B3GAT1 (CD57), ADAM17, PRF1 (perforin), GZMB (granzyme B), FCGR3A (CD16), SLAMF7, KIR3DL2 and KIR2DL1. Protein products are indicated in parentheses. Statistical analysis performed using Student’s t-test *P<0.05, ** P<0.01, ***P<0.001 and ****P<0.0001. (C) Schema showing directionality of GSEA comparisons performed between the four NK-cell groups (upper panel) and bubble chart of GSEA analysis NES and FDR scores when compared to curated NK-related gene sets from MSigDB (lower panel). Red arrows indicate analyses depicted in heatmaps and running enrichment score (ES) analysis. GSEA heatmaps for all replicates for (D) patient CD57+ vs. HD CD57+ cells in NK-cell activation pathways in GO, and (E) patient CD57+ vs. patient CD57– cells in GO: positive regulation of NK-cell activation pathway. (F) Running enrichment score (ES) analysis of panels (D) and (E).

haematologica | 2021; 106(9)

2523

Letters to the Editor

Figure 2. CD56+ natural killer cells from multiple myeloma patients are hypo-responsive to elotuzumab-labeled myeloma cells. Peripheral blood mononuclear cells (PBMC) from healthy donors (n=9), and newly diagnosed mulptiple myeloma (NDMM) patients (n=10) or refractory relapsed MM (RRMM) patients (n=10) at baseline (pre-treatment) were cultured with OPM2 target cells in the presence of 10 μg/mL elotuzumab (elo) or human IgG1 (iso) isotype control. Shown in (A) histogram overlay of changes in CD16 expression on CD56dimCD16+ subset of natural killer (NK) cells. (B) Percentage distribution of NK-cell subsets (left panel) or percentage of CD16+ on CD56dimCD57+ NK cells (right panel) in healthy donor (HD), newly diagnosed MM and RRMM patient PBMC after treatment under the same conditions as above. (C) Percentage distribution of CD56dimCD16+ subset of NK cells in HD and RRMM patient PBMC after treatment under the same conditions as above in the presence or absence of ADAM17 inhibitor (n=5 per group) (D) Collated data for HD, NDMM and refractory relapsed (RR) MM patients (n=9-10 per group) showing CD107a degranulation by different NK-cell subsets. Data are pooled from four independent experiments. *P<0.05, Oneway ANOVA with Bonferroni post-hoc test.

tutive activation of NK cells via CD16, causing NK-cell exhaustion in MM patients. We then investigated whether NDMM patient NK-cell cytotoxicity recovered post-induction treatment or postASCT and if they can be targeted with monoclonal antibody therapy. In order to reveal myeloma patient NK cell killing potential, we investigated their cytotoxicity against the MHC class I negative erythro-leukemia cell line, K562 (Figure 3A), their antibody-dependent cellular cytotoxicity (ADCC) capacity against OPM2 myeloma cells with elotuzumab (Figure 3B), or an isotype control (Figure 3C). After induction treatment or ASCT, NK cells from newly diagnosed MM patients killed K562 cells at equivalent levels to HD NK cells (Figure 3A). In contrast, 2524

NDMM patient NK cells were significantly less efficient at myeloma cell ADCC than HD NK cells, requiring higher numbers of NK effectors to achieve target lysis (Figure 3B); this reduced ADCC function was present after induction therapy, and after ASCT (Figure 3B). Finally, in the presence of an isotype control, myeloma patient NK cells were significantly less efficient than HD NK cells at killing myeloma targets (Figure 3C). Taken together, this data suggests whilst myeloma patient NK cells have cytotoxic potential, they are unable to effectively kill myeloma targets. There was no difference in CD16+ NK cells from preand post-induction treatment (Online Supplementary Figure S3C); however, there were less mature CD57+ NK haematologica | 2021; 106(9)

Letters to the Editor

cells post-ASCT (Online Supplementary Figure S3C). NKcell CD107a degranulation (although lower) was not significantly different in NDMM patients compared to HD at the post-induction or ASCT timepoints (Online Supplementary Figure S3D). These findings reveal an apparent separation between NK-cell degranulation and

effective cytotoxicity against myeloma cells (but not against K562). This was previously observed in a model system where phospholipase γ2 signaling was impaired,12 and also when adhesion was impaired between effector and target cells.10 No apparent differences were observed in the level of cytokine TNF and chemokines (CCL3,

Figure 3. Natural killer cells from newly diagnosed multiple myeloma patients show significantly lower myeloma antibody-dependent cellular cytotoxicity response post-induction therapy and post-autologous stem cell transplant. A standard 4-hour chromium release assay was used to assess natural killer (NK) cell function, adapted from Hsu et al.5 Peripheral blood mononuclear cells (PBMC) from healthy donors (HD) (n=8) and newly diagnosed multiple myeloma (NDMM) patients at the time points end of induction (EOI) and post-autologous stem cell transplant (post-ASCT), n=10 per group, were co-cultured with K562 target cells to determine NK cell natural cytotoxicity levels (A), or with OPM2 myeloma target cells and 10 μg/mL elotuzumab (Elo) (B) or human IgG1 isotype (iso) control (C) to determine antibody-dependent cellular cytotoxicity (ADCC) capacity. Cytotoxicity was assessed by chromium (51Cr) release assays and the data displayed as percentage of target cell lysis (A to C, left panels). Each line represents a non-linear regression curve for HD (blue line), or NDMM at EOI (green line) and post-ASCT (orange line) time points at the indicated effector:target (E:T) cells ratios (normalized for the percentage of NK cells). Inserted bar graphs on the right for (A), (B) and (C) show the NK E:T ratio required to achieve 40% target lysis (A, K562), 20% target lysis (B, OPM2 with elotuzumab) and 10% target lysis (C, OPM2 with isotype control) target lysis, extrapolated from the non-linear regression curves on the left. Each symbol represents an individual patient or HD. Data are pooled from five independent experiments. **P<0.01, * P<0.05, Student’s t-test.

haematologica | 2021; 106(9)

2525

Letters to the Editor

CCL2 and CCL5) secreted by NK cells in the same co-culture conditions (Online Supplementary Figure S3E). A recent study also demonstrated that continual lenalidomide treatment of MM patients did not improve NK-cell function with a lower ADCC response and decreased reactivity against K562 target cells.13 These observations are similar to our findings suggesting that lenalidomide treatment alone is insufficient to rescue MM patient NK cell function in vivo. In contrast, in vitro lenalidomide-treated HD NK cells up-regulate genes for IL2/STAT5, mTORC1 and TNF signalling pathway suggesting activation (data not shown). In summary, our results showed that NK cells in MM patients are chronically stimulated with an increase in terminally differentiated NK cells and loss of regulation of activation. This scenario is plausible considering the bone marrow is a site of myeloma disease as well as NK-cell development and maturation. Thus repetitive stimulation by the myeloma cells would impact NK-cell maturation. In addition, we showed lenalidomide and dexamethasone combination treatment did not repair this intrinsic NK-cell defect in MM. In order to address this issue, future combination immunotherapy approaches could use a tumor targeting antibody (e.g., Daratumumab, anti-CD38) with agonistic anti-CD137 mAb14 or anti-TIGIT15 to rescue NK-cell dysfunction in MM. Criselle D’Souza,1,2* Simon P. Keam,1,2,3 Han Xian Aw Yeang,1 Michael Neeson,1 Kelden Richardson,1 Andy K. Hsu,1 Rachael Canfield,1 Natalie Bezman,4° Michael Robbins,5° Hang Quach,6,7 David S. Ritchie,3,7,8 Simon J. Harrison,3,8 Joseph A. Trapani,1,2 H. Miles Prince,1,2,8 Paul A. Beavis,1,2 Phillip K. Darcy1,2 and Paul J. Neeson1,2* 1 Cancer Immunology Program, Peter MacCallum Cancer Center, Melbourne, Victoria, Australia; 2Sir Peter MacCallum Department of Oncology, University of Melbourne, Victoria, Australia; 3 Tumor Suppression and Cancer Sex Disparity Laboratory, Sir Peter MacCallum Cancer Center, Melbourne, Victoria, Australia; 4Oncology Discovery Research, Bristol-Myers Squibb, Redwood City, CA, USA; 5 Translational Medicine, Bristol-Myers Squibb, Cambridge, MA, USA; 6Department of Hematology, St Vincent’s Hospital, Melbourne, Victoria, Australia; 7Faculty of Medicine, University of Melbourne, Melbourne, Victoria, Australia and 8Clinical Hematology, Sir Peter MacCallum Cancer Center and Royal Melbourne Hospital, Melbourne, Victoria, Australia °NB current address: Arsenal Bio, San Francisco, CA, USA °MR current address: io904 LLC, Jacksonville Beach, FL, USA Correspondence: PAUL J. NEESON - paul.neeson@petermac.org CRISELLE D’SOUZA - criselle.dsouza@petermac.org doi:10.3324/haematol.2020.277525 Received: December 3, 2020. Accepted: April 29, 2021. Pre-published: May 20, 2021. Disclosures: PJN received research funding from BMS for this project. Contributions: CD, AKH, HXAY, MN, RC performed experi-

2526

ments; CD, AKH, SPK, MN, RC, KR analyzed experiment data; CD and PJN interpreted, analyzed and wrote the manuscript; PJN conceived and supervised the project; HQ, DSR, SJH, and HMP conducted the clinical trials; NB, MR, JAT, HMP, PAB, PKD helped interpret data. All authors reviewed, edited and approved the manuscript. Acknowledgments: we thank Dr Jessica Li, Dr Deborah Meyran and Dr Minyu Wang for critical review of the manuscript. Funding: funding for this study was provided by Cure Cancer Australia, Australian NHMRC program grant No. 113 2373 and Bristol-Myers Squibb.

References 1. de Weers M, Tai YT, van der Veer MS, et al. Daratumumab, a novel therapeutic human CD38 monoclonal antibody, induces killing of multiple myeloma and other hematological tumors. J Immunol. 2011;186(3):1840-1848. 2. Dimopoulos MA, Lonial S, White D, et al. Elotuzumab plus lenalidomide/dexamethasone for relapsed or refractory multiple myeloma: ELOQUENT-2 follow-up and post-hoc analyses on progression-free survival and tumour growth. Br J Haematol. 2017;178(6):896-905. 3. Hipp S, Tai YT, Blanset D, et al. A novel BCMA/CD3 bispecific T-cell engager for the treatment of multiple myeloma induces selective lysis in vitro and in vivo. Leukemia. 2017;31(8):1743-1751. 4. Mikkilineni L, Kochenderfer JN. Chimeric antigen receptor T-cell therapies for multiple myeloma. Blood. 2017;130(24):2594-2602. 5. Hsu AK, Quach H, Tai T, et al. The immunostimulatory effect of lenalidomide on NK-cell function is profoundly inhibited by concurrent dexamethasone therapy. Blood. 2011;117(5):1605-1613. 6. Khot A SR, Stokes K, et al. Low dose lenalidomide induction followed by autologous transplantation in untreated patients with myeloma is associated with adequate collection of haematopoietic and dendritic cell precursors and high response rates. Cytotherapy. 2013;15(Suppl 4):S6-7. 7. Quach H, Fernyhough L, Henderson R, et al. Upfront lower dose lenalidomide is less toxic and does not compromise efficacy for vulnerable patients with relapsed refractory multiple myeloma: final analysis of the phase II RevLite study. Br J Haematol. 2017;177(3):441-448. 8. Lopez-Verges S, Milush JM, Pandey S, et al. CD57 defines a functionally distinct population of mature NK cells in the human CD56dimCD16+ NK-cell subset. Blood. 2010;116(19):3865-3874. 9. Delconte RB, Kolesnik TB, Dagley LF, et al. CIS is a potent checkpoint in NK cell-mediated tumor immunity. Nat Immunol. 2016;17(7):816-824. 10. Romee R, Foley B, Lenvik T, et al. NK cell CD16 surface expression and function is regulated by a disintegrin and metalloprotease-17 (ADAM17). Blood. 2013;121(18):3599-3608. 11. Pazina T, James AM, MacFarlane AWt, et al. The anti-SLAMF7 antibody elotuzumab mediates NK cell activation through both CD16dependent and -independent mechanisms. Oncoimmunology. 2017;6(9):e1339853. 12. Caraux A, Kim N, Bell SE, et al. Phospholipase C-gamma2 is essential for NK cell cytotoxicity and innate immunity to malignant and virally infected cells. Blood. 2006;107(3):994-1002. 13. Besson L, Charrier E, Karlin L, et al. One-year follow-up of natural killer cell activity in multiple myeloma patients treated with adjuvant lenalidomide therapy. Front Immunol. 2018;9:704. 14. Ochoa MC, Perez-Ruiz E, Minute L, et al. Daratumumab in combination with urelumab to potentiate anti-myeloma activity in lymphocyte-deficient mice reconstituted with human NK cells. Oncoimmunology. 2019;8(7):1599636. 15. Zhang Q, Bi J, Zheng X, et al. Blockade of the checkpoint receptor TIGIT prevents NK cell exhaustion and elicits potent anti-tumor immunity. Nat Immunol. 2018;19(7):723-732.

haematologica | 2021; 106(9)

Letters to the Editor

Targeting BRD4 in acute myeloid leukemia with partial tandem duplication of the MLL gene The lysine methyltransferase 2a (KMT2A) (which is also known and hereafter referred to as mixed-lineage leukemia [MLL], trithorax [Drosophila] homolog gene) plays a pivotal role in embryogenesis and hematopoiesis. Recurrent, balanced translocations involving the MLL gene [t(v;11q23)] are heterogeneous, and more than 75 different fusion partners have been described as important drivers in acute myeloid leukemia (AML) leukemogenesis.1,2 Beside chromosomal aberrations, a unique gene rearrangement in MLL known as partial tandem duplication (PTD) can be found in approximately 5-11% of cytogenetically normal AML (CN-AML) patients. This mutation is associated with poor prognosis.3-8 Although both t(v;11q23) and the MLL-PTD result in increased HOMEOBOX (HOX) gene expression in leukemic blasts, t(v;11q23) have been shown to be genetically and functionally distinct from the MLL-PTD.9 While almost all t(v;11q23) lose their C-terminal transactivation and methyltransferase domains, these C-terminal domains are retained in the MLL-PTD.9 The transcription factor BRD4 is a member of the bromodomain and extra terminal (BET) family of proteins. Aberrant BRD4 binding and gene activation has been shown to be important for t(v;11q23)-mediated leukemogenesis.10 JQ1 is one of the best-characterized, small molecule bromodomain inhibitors.11 However, the potential use of JQ1 in MLL-PTD AML has not been extensively studied yet. Therefore, we examined whether MLL-PTD AML blasts are sensitive to JQ1 treatment and if BRD4 inhibition results in an altered binding of the transcription factor to DNA.12 First, we tested whether JQ1 treatment has an impact on cell proliferation and survival in a MLL-PTD+ AML cell line (i.e., EOL-1) and in a MLL wild-type cell line (i.e., K562).13 Both cell lines were treated with JQ1 or dimethyl sulfoxide (DMSO) vehicle control for 24 hours (h) at different concentrations and cell growth was assessed by WST-1 assay. We found a significant decrease in cell proliferation in EOL1 cells (IC50 = 321 nM) but not in K562 cells (Figure 1A). Concomitantly, we found a significant and dose dependent increase in the number of apoptotic EOL-1 cells (Figure 1B) but not in the K562 cells (Online Supplementary Figure S1A). We also analyzed the effect of JQ1 treatment on primary blast cell growth from three AML patients that harbor a MLL-PTD compared to normal hematopoietic stem and progenitor cells (HSPC; CD34+ cord blood) controls. We found JQ1 treatment significantly reduced blast cell growth assessed by a decrease in the number of colony-forming cells (CFC) in JQ1-treated MLL-PTD AML samples, with no significant decreases of CFC in normal HSPC (Figure 1C) or MLL wild-type (wt) primary patient samples (Online Supplementary Figure S1B). Next, we tested the effect of JQ1 in a murine AML mouse model. For these experiments we used our well established MllPTD/WT Flt3ITD/WT double knockin AML mouse model14,15 that develops lethal CN-AML with ~100% penetrance. In secondary bone marrow transplantation, it leads to death within 6 to 12 weeks.14,15 Of note, MLL-PTD is predominantly found in CN-AML in humans. First, we wanted to determine whether JQ1 also induced apoptosis in the MllPTD/WT Flt3ITD/WT mouse AML blasts, similar to what we observed in human AML blasts. We found that JQ1 induced a significant increase in apoptosis assessed by Annexin V staining in the haematologica | 2021; 106(9)

MllPTD/WT Flt3ITD/WT blasts (Figure 1D) with essentially no toxicity to normal murine bone marrow cells (Online Supplementary Figure S1C). Based on these promising in vitro results, we then wanted to test, whether targeting BRD4 in vivo would result in prolonged survival of mice with Mll-PTD+ leukemia. In order to test the antileukemic activity of JQ1 in a murine AML model, we used our previously established MllPTD/WT Flt3ITD/WT mouse model.14-16 We observed a significant increase in survival of JQ1 treated MllPTD/WT Flt3ITD/WT mice compared to mice treated with vehicle control (Figure 1E). Moreover, the mice that eventually succumbed to disease and were treated with JQ1 had significant lower spleen weight, indicating a lower leukemic burden (Figure 1F). Interestingly, we also found that the leukemic bone marrow cells from JQ1-treated mice had a significant lower engraftment potential after re-transplantation than cells from mice treated with vehicle control (Online Supplementary Figure S1D). These data suggested that JQ1 might also have an impact on leukemic cell self-renewal and consequently leukemia stem cells (LSC) in Mll-PTD AML, however further experiments are needed to fully address effects on LSC by JQ1 in Mll-PTD leukemia. After identifying the ability of JQ1 to decrease MLL-PTD+ AML blast growth in vitro and in vivo, we wanted to determine whether alterations in BRD4 binding accounts for MLL-PTD blast sensitivity to BRD4-inhibition. BRD4 has been shown to be a positive regulator of gene transcription and aberrant BRD4-binding in cancer induces alterations in gene expression. Thus, we hypothesize that MLL-PTD leukemogenesis is driven by dysregulation of gene expression patterns resulting from aberrant BRD4 binding. Furthermore, we wanted to determine whether normal BRD4 binding could be restored by treatment with JQ1. In order to address this question, we performed total RNA sequencing (RNA-seq) on primary MLL-PTD AML blasts and normal HSPC treated with JQ1 or vehicle control (n=3 for each group, pooled) before and after JQ1 treatment and analyzed as previously described.18 In addition, we performed chromatin immunoprecipitation sequencing (ChIP-seq) using a BRD4 antibody. Cells from three healthy cord blood donors (CB) and primary leukemic cells from three patients with MLL-PTD were treated with either JQ1 (10 nM) or vehicle (DMSO) for 24 hours (h) and ChIP was performed as previously described.18 The 75-basepair sequence reads were generated using an llumina sequencing platform (NextSeq 500) and then mapped to the human reference genome (GRCh37/hg19) using the BWA algorithm with default settings.19 Aligned reads were normalized and genomic regions with local enrichments against corresponding input sample, peaks, were defined using MACS algorithm with a cutoff P-value of 1e-7.20 Consensus peaks were defined by merging overlapping peak coordinates and peak scores were calculated. The resulting matrix was annotated with gene information by calculating distances from RefSeq gene starts and ends to the center of the consensus peak regions and applying an annotation cutoff of 5 kb. Similar to what has previously been described for t(v;11q23)-AML that BRD4 has an aberrant binding profile, we found three times more and stronger genomic interactions of BRD4 in the MLL-PTD patient sample compared to normal HSPC, and only 22% of the BRD4 peaks in MLL-PTD overlapped with normal HSPC in ChIP-seq (Figure 2A). When we treated the MLL-PTD sample with JQ1, both the number of peaks and their intensity decreased, whereas JQ1 treatment did not affect BRD4 binding in HSPC (Figure 2B). Next, by integrating the RNA-seq and ChIP2527

Letters to the Editor

seq data, we wanted to determine which genes had alterations in BRD4 binding, leading to mRNA changes in the AML patient cells compared to normal CD34+ cells. For these analyses, the data obtained from RNA-seq counts, and ChIP-seq using BRD4, were merged by gene id. Our strategy to detect different patterns of changes in peak

scores and gene expression consisted in the conversion of the values into quartiles in way to define MLL-PTD specific response to JQ1 treatment. Furthermore, we also wanted to determine the alterations that were reversed by BRD4-inhibiton. First, we identified BRD4 binding sites that were close by to a

Figure 1. Targeting BRD4/Brd4 using the inhibitor JQ1 has an effect on MLL-partial tandem duplication (MLL-PTD)/Mll-PTD acute myeloid leukemia cells. (A) WST-1 assay on EOL-1 and K562 cells treated with the indicated concentrations of JQ1 for 48 hours (h). (B) EOL-1 cells were treated for 24 h with the indicated concentration of JQ1. Cells were then assessed for apoptosis using Annexin V+ staining and flow cytometry at 24 h post treatment; *P<0.05, ***P<0.001. (C) Colony forming unit assays were performed on three CD34+ cord blood and three MLL-partial tandem duplication (MLL-PTD) acute myeloid leukemia (AML) patients’ samples, cells were plated in triplicates, normalized results are shown.13 Cells were treated with JQ1 at a concentration of 9 nM or 12.5 nM or with vehicle control (dimethyl sulfoxide [DMSO]); **P<0.01. (D) Primary murine MllPTD/WT Flt3ITD/WT blasts were treated for 24 h with the indicated concentration of JQ1. Cells were then assessed for apoptosis using Annexin V+ staining and flow cytometry at 24 h post treatment; *P<0.05, **P<0.01. (E) MllPTD/WT Flt3ITD/WT blasts were transplanted into sub-lethally irradiated BoyJ mice. Starting at 2 weeks post transplantation, mice were treated with 50 mg/kg body weight of JQ1 or vehicle control for 6 days each week for the entire duration of the study. Mice that died early without any signs of leukemia were excluded. The experiment was stopped after 120 days post treatment initiation. Treatment with JQ1 prolonged survival compared to controls (P=0.001) and (F) also led to a reduced weight of the spleen.

2528

haematologica | 2021; 106(9)

Letters to the Editor

Figure 2. Aberrant BRD4 binding in MLL-partial tandem duplication cells drives a distinct gene expression profile that can be restored by JQ1 treatment. (A) Venn diagram showing overlap of BRD4 binding sites between CD34+ selected cord blood (CB) samples primary samples from MLL-partial tandem duplication (MLL-PTD) acute myeloid leukemia (AML) patients. (B) Scatter plots showing changes in BRD4 binding peak scores upon JQ1 treatment in healthy CB and MLLPTD AML samples. (C) Schematic overview over the bioinformatic approach used to identify the genes deregulated by BRD4 in MLL-PTD cells. A peak and gene pair was classified as MLL-PTD specific positive response when peak score and gene expression were in first quartile (0-25%) for CB samples, peak score and gene expression were in the top quartile (75-100%) for MLL-PTD dimethyl sulfoxide (DMSO) sample, and a decreased expression was observed in both peak score and gene expression for MLL-PTD sample upon JQ1 treatment (25% or more). Similarly, we identified potential oncosuppressor genes downregulated by MLL-PTD but restored by JQ1 treatment. Genes whose expressions were positively or negatively correlated with changes in BRD4 binding with at least 25% change under JQ1 treatment are shown in the Online Supplementary Table S1.(D) Heat map of genes that were found to be deregulated in MLL-PTD cells and which also showed a response to JQ1 treatment. (E) Example of one of 130 genes that fulfilled all criteria. ADAMDEC1 expression is increased in MLL-PTD AML cells by a binding of BRD4, and is downregulated upon JQ1 treatment.

haematologica | 2021; 106(9)

2529

Letters to the Editor

Figure 3. MLL-partial tandem duplication drives the aberrant expression profile through BRD4. (A) Cell lysate of EOL-1 cells, either treated with vehicle (dimethyl sulfoxide [DMSO]) or with the indicated concentration of JQ1, were subjected to a chromatin immunoprecipitation (ChIP) assay. Antibodies against BRD4 or POL1RA (control) or an unspecific immunoglobulin G (IgG) were used for the pulldown. DNA of ADAMDEC1 and SLAMF8 were quantified using quantitative realtime polymerase chain reaction (qRT-PCR). The enriched binding of BRD4 to both genes, i.e., ADAMDEC1 and SLAMF8, was decreased after treatment of JQ1; **P<0.01, ***P<0.001. (B) Relative expression of ADAMDEC1 and SLAMF8 relative to GAPDH. qRT-PCR was performed on CD34+ selected cord blood (CB) cell samples (n=3, pooled) and three primary cell samples from MLL-partial tandem duplication (MLL-PTD) acute myeloid leukemia (AML) patients treated with JQ1 at 50 nM for 24 hours. Treatment with JQ1 leads to significantly lower expression of the genes in MLL-PTD patients’ cells but not in CB cells; ns = not significant, *P<0.05, **P<0.01, ***P<0.001. (C) Normalized expression of BRD4 downstream targets, BCL2, CDK6 and MYC, relative to GAPDH in EOL1 cells. Cells treated with JQ1 for 48 hours showed significantly reduced expression of the genes; *P<0.05, **P<0.01, ***P<0.001. (D) Western Blot analysis of EOL1 cells, which were previously treated with JQ1 or vehicle (DMSO) as control, validated the downregulation of these genes. Similar results were found when cells were transfected with a short hairpin RNA (shRNA) against MLL-PTD (shMLL-PTD) but not when transfected with a scramble control. Data of the densitometry are shown. RNA, cDNA, real-time PCR, ChIP and western blots were performed using previously published methods.13,16-18

2530

haematologica | 2021; 106(9)

Letters to the Editor

gene’s transcription start site (TSS), specifically bound in MLL-PTD cells but not in HSPC and could be suppressed by at least 25% through JQ1 treatment. Then, the genes close by such BRD4 binding sites and presenting the same pattern in their expression were classified as positive response to treatment, whereas the genes that present the opposite trend in their expression were classified as negative response to treatment (Figure 2C). As result, we identified genes whose expression was positively or negatively regulated by changes in BRD4 binding specific in the MLL-PTD AML sample and could be reversed by JQ1 treatment (Figure 2D). We identified 92 genes which were significantly upregulated in MLL-PTD AML cells through BRD4 binding and which were downregulated by BRD4 inhibition (Online Supplementary Table S1). As an example, Figure 2E shows the expression of ADAMDEC1 which is increased in MLL-PTD AML cells and correlates with binding of BRD4. On the other hand, 38 genes were downregulated by BRD4 in the MLL-PTD AML cells and its inhibition led to a re-expression (Online Supplementary Table S1). We also performed ingenuity pathway analysis (IPA) from the ChIP-RNA integration and identified pathways that are affected by BRD4-mediated transcriptional changes (Online Supplementary Table S2). These data suggest that the distinct gene expression profile of MLL-PTD positive AML, is at least partly driven by the transcription factor BRD4, similar to AML cells which harbor a t(v;11q23). In addition, we showed that this aberrant expression can be restored after JQ1 treatment. Because we compared only cells from one patient (treated vs. untreated) with HSPC from three pooled CB samples, next we wanted to validate the direct binding of BRD4 to two genes that were upregulated in MLL-PTD AML cells, i.e., ADAMDEC1 and SLAMF8 (Figure 3A). These genes were also shown to have a functional role in MLL-PTD cells since knockdown of either of these genes resulted in decreased leukemic cell growth (Online Supplementary Figure S1E). Using a chromatin immunoprecipitation followed by quantitative real-time polymerase chain reaction (qRT-PCR), we show that BRD4 binds to ADAMDEC1 and SLAMF8, and this binding can be repressed by JQ1 treatment. Moreover, both genes are upregulated in MLL-PTD AML patients cells compared to CB cells (Figure 3B). For both genes, JQ1 treatment led to decreased expression. Similar results were found when MLL-PTD was knocked down (Online Supplementary Figure S1F and G). We also found similar results in our in vivo mouse model. Mice treated with JQ1 had lower expression of several potential oncogenes, including Adamdec1 and Slamf8 (Online Supplementary Figure S1H to J). Finally, we showed that JQ1 treatment of MLL-PTD cells also results in the decreased expression of BCL2, CDK6, and MYC, well-established downstream targets of BRD4 (Figure 3C).1,2 We found that knocking down the MLL-PTD fusion gene using a short hairpin RNA (shRNA) also resulted in downregulation of these proteins regulated by BRD4, similar to treatment with JQ1 (Figure 3D). Previously, it has been shown that fusion proteins from t(v;11q23) can initiate aberrant gene expression profiles by recruiting BRD4.1 Our data suggest that in MLL-PTD cells utilize a similar mechanism and could account for the distinct gene expression profile. Taken together, our data shows that targeting BRD4 with the small molecule JQ1 reduces cell proliferation of MLL-PTD cells in vitro and induces apoptosis. In line, we found that JQ1 treatment decreased leukemic burden in vivo and improved survival in vivo. We show for the first time to our knowledge, that aberrant BRD4 binding in MLL-PTD cells results in a distinct deregulation of genes. haematologica | 2021; 106(9)

By integrating the RNA-seq and ChIP-seq analysis, we identified targets relevant to the MLL-PTD subgroup of AML patients and validated in additional samples from MLL-PTD+ AML primary patient blasts and our unique AML mouse model. This novel group of genes might be associated with leukemogenesis of MLL-PTD+ CN-AML and also with the poor prognosis of this subgroup. Importantly, we were able to reverse the aberrant gene expression patterns by treatment with JQ1. Therefore, targeting BRD4 might be an effective and promising treatment option for patients harboring a MLL-PTD to improve their outcome. Marius Bill,1,2* Chinmayee Goda,1* Felice Pepe,1 Hatice Gulcin Ozer,3 Betina McNeil,1 Xiaoli Zhang,3 Malith Karunasiri,1 Rohan Kulkarni,1 Sonu Kalyan,1 Dimitrios Papaioannou,1,4 Gregory Ferenchak,1 Ramiro Garzon,1,4 James E. Bradner,5 Guido Marcucci,6 Michael A. Caligiuri,6 and Adrienne M. Dorrance1,4 *MB and CG contributed equallly as co-first authors. 1 The Ohio State University, Comprehensive Cancer Center, Columbus, OH, USA; 2Medizinische Klinik und Poliklinik I, Universitätsklinikum Carl Gustav Carus Dresden, Dresden, Germany; 3The Ohio State University, Department of Biomedical Informatics, Columbus, OH, USA; 4Division of Hematology, Department of Internal Medicine, The Ohio State University, Columbus, OH, USA; 5Dana-Faber Cancer Institute, Boston, MA, USA and 6City of Hope Comprehensive Cancer Center, Duarte, CA, USA Correspondence: ADRIENNE DORRANCE adrienne.dorrance@osumc.edu doi:10.3324/haematol.2020.271627 Received: September 8, 2020. Accepted: April 30, 2021. Pre-published: May 13, 2021. Disclosures: JEB is a shareholder and executive of Novartis AG and provided JQ1 for the studies. All other authors declare no conflicts of interest. Contributions: AMD designed the study; MB, CG, FP, BM, MK, RK, DP and GF performed the experiments; MB, CG, FP, BM, MK, RK, SK, DP, GF, RG, JB, GM, MAC and AMD contributed to the data interpretation; MB, CG, FP and AMD wrote the manuscript; HGO and XZ performed bioinformatics and statistical analyses. All authors reviewed the manuscript. Acknowledgments: the authors would like to thank the patients who consented to participate and the families who supported them; to Donna Bucci, Christopher Manring and the Leukemia Tissue Bank at The Ohio State University Comprehensive Cancer Center, Columbus, OH, for sample processing and storage services.

References 1. Yue Zhang, Aili Chen, Xiao-Mei Yan and Gang Huang. Disordered epigenetic regulation in MLL-related leukemia. Int J Hematol. 2012;96(4):428-437. 2. Winters AC, Bernt KM. MLL-rearranged leukemias-an update on science and clinical approaches. Front Pediatr. 2017;5:4. 3. Ballabio E, Milne TA. Molecular and epigenetic mechanisms of MLL in human leukemogenesis. Cancers (Basel). 2012;4(3):904-944. 4. Mrózek K, Marcucci G, Paschka P, Whitman SP, Bloomfield CD. Clinical relevance of mutations and gene-expression changes in adult acute myeloid leukemia with normal cytogenetics: are we ready for a prognostically prioritized molecular classification? Blood. 2007; 109(2):431-448. 5. Whitman SP, Liu S, Vukosavljevic T, et al. The MLL partial tandem duplication: evidence for recessive gain-of-function in acute myeloid leukemia identifies a novel patient subgroup for molecular-targeted therapy. Blood. 2005;106(1):345-352.

2531

Letters to the Editor

6. Patel JP, Gönen M, Figueroa ME, et al. Prognostic relevance of integrated genetic profiling in acute myeloid leukemia. N Engl J Med. 2012;366(12):1079-1089. 7. Whitman SP, Ruppert AS, Marcucci G, et al. Long-term disease-free survivors with cytogenetically normal acute myeloid leukemia and MLL partial tandem duplication: a Cancer and Leukemia Group B study. Blood. 2007;109(12):5164-5167. 8. Schnittger S, Kinkelin U, Schoch C, et al. Screening for MLL tandem duplication in 387 unselected patients with AML identify a prognostically unfavorable subset of AML. Leukemia. 2000;14(5):796-804. 9. Dorrance AM, Liu S, Yuan W, et al. Mll partial tandem duplication induces aberrant Hox expression in vivo via specific epigenetic alterations. J Clin Invest. 2006;116(10):2707-2716. 10. Abedin SM, Boddy CS, Munshi HG. BET inhibitors in the treatment of hematologic malignancies: current insights and future prospects. Onco Targets Ther. 2016;9:5943-5953. 11. Alqahtani A, Choucair K, Ashraf M, et al. Bromodomain and extraterminal motif inhibitors: a review of preclinical and clinical advances in cancer therapy. Future Sci OA. 2019;5(3):FSO372. 12. Dawson MA, Gudgin EJ, Horton SJ, et al. Recurrent mutations, including NPM1c, activate a BRD4-dependent core transcriptional program in acute myeloid leukemia. Leukemia. 2014;28(2):311-320. 13. Bill M, Pathmanathan A, Karunasiri M, et al. EGFL7 antagonizes

2532

NOTCH signaling and represents a novel therapeutic target in acute myeloid leukemia. Clin Cancer Res. 2020;26(3):669-678. 14. 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. 15. Bernot KM, Nemer JS, Santhanam R, et al. Eradicating acute myeloid leukemia in a Mll(PTD/wt):Flt3(ITD/wt) murine model: a path to novel therapeutic approaches for human disease. Blood. 2013; 122(23):3778-3783. 16. Dorrance AM, Neviani P, Ferenchak GJ, et al. Targeting leukemia stem cells in vivo with antagomiR-126 nanoparticles in acute myeloid leukemia. Leukemia. 2015;29(11):2143-2153. 17. Papaioannou D, Shen C, Nicolet D, et al. Prognostic and biological significance of the proangiogenic factor EGFL7 in acute myeloid leukemia. Proc Natl Acad Sci U S A. 2017;114(23):E4641-E4647. 18. Papaioannou D, Petri A, Dovey OM, et al. The long non-coding RNA HOXB-AS3 regulates ribosomal RNA transcription in NPM1-mutated acute myeloid leukemia. Nat Commun. 2019;10(1):5351. 19. Li H, Durbin R. Fast and accurate long-read alignment with BurrowsWheeler transform. Bioinformatics. 2010;26(5):589-595. 20. Zhang Y, Liu T, Meyer CA, et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 2008;9(9):R137.

haematologica | 2021; 106(9)

CASE REPORT Clinical genomic profiling of novel grey zone lymphoma paired lesions with sequential central nervous system involvement in two adolescent patients Grey zone lymphoma (GZL), defined as B-cell lymphoma, unclassifiable, with features intermediate between large B-cell lymphoma (LBCL) and classic Hodgkin lymphoma (cHL) (BCL-U-IND) is a rare diagnos-

tic entity.1-3 Synchronous GZL, LBCL and cHL occurring simultaneously in the same patient, and sequential GZL, LBCL preceding or following a diagnosis of cHL, are even less common.4 We identified two adolescent patients, a 17 year-old male (17M, case #1) and 16 year-old female (16F, case #2), who were diagnosed with stage IV nodular sclerosis cHL (NS-cHL) with primary mediastinal location and subsequent central nervous system (CNS) LBCL. Copy-number alterations were assessed using Affymetrix OncoScan® microarray analysis, and targeted next-gener-

Table 1. Clinicopathological summary of sequential grey zone lymphomas.

Case/ age (yr)/ sex

Presentation (time after initial diagnosis)

Biopsy site

#1/17/M

Large Bone mediastinal and Marrow^ supraclavicular masses with spleen, liver, abdominal and bone lesions

Left internal jugular LN

Multiple supraFrontal and infra-tentorial brain lobe lesions with extensive brain leptomeningeal diseases lesion (6 months)

Diagnosis

LBCL-like synchronous GZL

Morphology

Focal sheets of large lymphoma cells with large round nuclei, smooth nuclear contours, vesicular chromatin, and prominent centrally located nucleoli with eosinophilic cytoplasm cHL, Characteristic mononucleated nodular Hodgkin and binucleated sclerosis Reed-Sternberg cells subtype, in the background of lymphocytes, stage IVA histiocytes, neutrophils, and eosinophils; the nodules separated by thick collagen band LBCL-like Diffuse sheets of large sequential lymphoma cells having open GZL chromatin, prominent centrally located nucleoli and a moderate amount of clear to eosinophilic cytoplasm.

Immunophenotype

Therapy

Outcome (follow-up period)

Large lymphoma cells: Positive: CD19, CD79a, CD45; Negative: CD3, Cytokeratin, TdT, CD30.

HRS cells: Positive: CD30, CD15, Pax-5 (weak); Negative: CD45, CD20, CD79a, LMP-1, EBER and EMA

ABVE-PC

Complete remission

Large POG9917 Arm lymphoma cells: A bridged to Alive with Positive: CD45, MMUD BMT no CD20, CD30, PAX-5 with conditioning evidence CD79a; Negative: and total body of disease CD15, EBER, ALK irradiation (81.7 (450 cGy). months) #2/16/F Large mediastinal Deep cHL, Characteristic mononucleated HRS cells: ABVE-PC with Complete mass with cervical right nodular sclerosis Hodgkin and binucleated Positive: CD30, radiotherapy remission LN, multiple supraclavicular subtype, Reed-Sternberg cells (HRS) CD15, Pax-5 (weak); to the bilateral pulmonary LN stage IVB with focal aggregates Negative: CD45, mediastinal mass and renal nodules in the background of lymphocytes, CD20, EBER and slow-responding histiocytes, neutrophils, eosinophils, areas of disease and plasma cells; the nodules separated by thick collagen band Solitary right Right LBCL-like Diffuse sheets of intermediate Lymphoma cells: ANHL1131 Alive with temporal lobe temporal sequential to large cells with smooth Positive: CD45, Group C1 no evidence brain lesion lobe brain GZL to irregular nuclear contour, CD20, CD30, MUM1; and surgical of disease (7 months) lesion inconspicuous to occasionally Negative: EBER excision (13.5 months) centrally located prominent nucleoli and moderate amount of cytoplasm. Occasional mitotic figures present ^

Outside bone marrow with limited slides reviewed as consultation. ABVE-PC: adriamycin, bleomycin, vincristine sulfate, etoposide phosphate, prednisone, cyclophosphamide; BMT: bone marrow transplant; cHL: classic Hodgkin lymphoma; F: female; GZL: grey zone lymphoma; HRS: Hodgkin and Reed-Sternberg; LBCL: large B-cell lymphoma; LN: lymph node; M: male; MMUD: mismatched unrelated donor; NA: not applicable.

haematologica | 2021; 106(9)

2533

Case Report

Table 2. Tissue-based cancer microarray and next-generation sequencing analysis of sequential grey zone lymphomas.

Case #1, 16M, Paired NS-cHL and CNS LBCL Microarray Cytobands Size (Mbp) 2p16.3-p12 28.2 9p24.3-q34.3 140.9 4q21.22-q21.23 1.1 6p25.3-p11.2 57.0 12p13.33-q24.33 133.6 Xp22.33-q28 155.0 Yp11.31-q11.23 26.1 Gene APC FAT4 NOTCH3 CREBBP APC ORAI1 SETX SOS1 SYNE1 SYNE1 SYNE1 SYNE1

Pos (hg19) chr5:112102044 chr4:126239848 chr19:15272113 chr16:3819314 chr5:112175211 chr12:122064705 chr9:135145055 chr2:39251255 chr6:152651971 chr6:152630998 chr6:152565729 chr6:152464786

Cytobands Size (Mbp) 2p25.3-q37.3 243.0 9p24.3-q34.3 140.9 16p13.13-p11.1 35.1 1p36.11-p35.3 3.0 5p15.33-q23.3 128.0 5q23.3-q31.1 4.0 5q31.1-q35.3 49.2 6p25.3-p23 14.6 6p23-p21.1 28.1 12p12.3-q24.33 117.3 15q11.1-q14 16.9 15q14-q21.2 12.5 15q21.2-q26.3 52.8 16q11.2-q24.3 43.7 19p13.3-p13.3 1.1 21p11.2-q22.3 38.4 22q13.2-q13.33 7.7 Xp22.33-p22.11 22.3 Xp22.11-q23 88.1 Xq23-q28 44.5 Gene TP53 FBXW7 CBL FAT1 KMT2D RELN ERG ZFHX3

Pos (hg19) chr17:7577569 chr4:153249384 chr11:119170426 chr4:187557893 chr12:49423015 chr7:103338388 chr21:39795460 chr16:72830889

Type Gain Gain Loss CN-LOH Gain Gain Loss RefSeq RNA NM_000038.5 NM_024582.4 NM_000435.2 NM_004380.2 NM_000038.5 NM_032790.3 NM_015046.5 NM_005633.3 NM_182961.3 NM_182961.3 NM_182961.3 NM_182961.3 Type Gain Gain Gain CN-LOH Gain Gain CN-LOH CG-LOH CN-LOH Gain Gain Loss Gain CN-LOH Gain Gain Loss Loss CN-LOH CG-LOH RefSeqRNA NM_000546.5 NM_033632.3 NM_005188.3 NM_005245.3 NM_003482.3 NM_005045.3 NM_001136154.1 NM_006885.3

Array Nomenclature arr[hg19] 2p16.3p12(50,889,958-79,060,207)x3 arr[hg19] 9p24.3q34.3(204,737-141,054,761)x3 arr[hg19] 4q21.22q21.23(83,278,777-84,335,477)x1 arr[hg19] 6p25.3p11.2(204,908-57,160,035)x2 hmz arr[hg19] 12p13.33q24.33(189,399-133,818,115)x3 arr[hg19] Xp22.33q28(177,941-155,219,364)x2 arr[hg19] Yp11.31q11.23(2,660,162-28,799,935)x0 Case #1, 16M, Paired NS-cHL and CNS LBCL NGS CDS; Protein; VAF (NS-cHL/LBCL) c.157G>A; p.Gly53Arg; 0.05 c.2282T>G; p.Leu761Trp; 0.37 c.6326G>A; p.Arg2109Gln; 0.48 c.2921C>A; p.Thr974Asn; 0.44/0.47 c.3920T>A; p.Ile1307Lys; 0.56/0.45 c.58G>A; p.Gly20Ser; 0.62/0.29 c.7234A>G; p.Ile2412Val; 0.47/0.28 c.1098T>A; p.Asp366Glu; 0.44/0.48 c.13849A>C; p.Asn4617His; 0.41/0.47 c.17174C>A; p.Thr5725Asn; 0.42/0.51 c.19635G>T; p.Arg6545Ser; 0.44/0.46 c.25091C>T; p.Pro8364Leu; 0.47/0.51 Case #2, 17F, Paired NS-cHL and CNS LBCL Microarray

Interpretation CNS LBCL only, Reported in GZL* CNS LBCL only, Reported in GZL* CNS LBCL only CNS LBCL only* CNS LBCL only* CNS LBCL only CNS LBCL only Interpretation III, CNS LBCL only III, CNS LBCL only III, CNS LBCL only III, Shared, Reported in GZL III, Shared III, Shared III, Shared III, Shared III, Shared III, Shared III, Shared III, Shared

Array Nomenclature Interpretation arr[hg19] 2p25.3q37.3(21,493-243,052,331)x3 CNS LBCL only, Reported in GZL* arr[hg19] 9p24.3q34.3(204,737-141,054,761)x5 CNS LBCL only, Reported in GZL* arr[hg19] 16p13.13p11.1(83,886-35,271,725)x3 CNS LBCL only, Reported in GZL arr[hg19] 1p36.11p35.3(25,194,298-28,160,199)x2 hmz Shared arr[hg19] 5p15.33q23.3(38,138-128,042,790)x3 CNS LBCL only arr[hg19] 5q23.3q31.1(128,063,275-132,042,740)x5 CNS LBCL only arr[hg19] 5q31.1q35.3(131,530,440-180,698,312)x2 hmz CNS LBCL only arr[hg19] 6p25.3p23(204,908-14,823,522)x3 hmz CNS LBCL only* arr[hg19] 6p23p21.1(14,984,113-43,101,670)x2 hmz CNS LBCL only arr[hg19] 12p12.3q24.33(16,480,948-133,818,115)x3 CNS LBCL only* arr[hg19] 15q11.1q14(20,161,371-37,079,572)x3 CNS LBCL only arr[hg19] 15q14q21.2(37,094,935-49,619,400)x1 CNS LBCL only arr[hg19] 15q21.2q26.3(49,643,377-102,397,317)x3 CNS LBCL only arr[hg19] 16q11.2q24.3(46,461,308-90,158,005)x2 hmz CNS LBCL only arr[hg19] 19p13.3(247,231-1,351,916)x3 CNS LBCL only arr[hg19] 21p11.2q22.3(9,648,314-48,097,610)x3 CNS LBCL only arr[hg19] 22q13.2q13.33(43,487,259-51,213,826)x1 CNS LBCL only arr[hg19] Xp22.33p22.11(177,941-22,471,996)x1 CNS LBCL onl y arr[hg19] Xp22.11q23(22,634,971-110,738,270)x2 hmz CNS LBCL only arr[hg19] Xq23q28(110,762,820-155,219,364)x3 hmz CNS LBCL only Case #2, 17F, Paired NS-cHL and CNS LBCL NGS CDS; Protein; VAF (NS-cHL/LBCL) c.712T>C; p.Cys238Arg; 0.51 c.1394G>A; p.Arg465His; 0.45 c.2656G>A; p.Glu886Lys; 0.40 c.3818A>T; p.His1273Leu; 0.45 c.14080G>C; p.Glu4694Gln; 0.50/0.30 c.1055A>G; p.Asn352Ser; 0.51/0.49 c.281T>G; p.Val94Gly; 0.46/0.33 c.5692G>C; p.Gly1898Arg; 0.47/0.95

Interpretation I/II, CNS LBCL only, Reported in GZL I/II, CNS LBCL only III, CNS LBCL only III, CNS LBCL only III, Shared, Reported in GZL III, Shared, Reported in GZL III, Shared III, Shared

*Detected in both CNS LBCL cases. NS-cHL: classic Hodgkin lymphoma, nodular-sclerosing subtype; CNS: central nervous system; LBCL: large B-cell lymphoma; NGS: next-generation sequencing; Pos: genomic coordinate; RefSeq: reference transcript ID; CDS: coding sequence; VAF: variant allele frequency; Mbp: mega basepairs; GZL: grey zone lymphoma; CG-LOH: copy-gain loss of heterozygozity; cHL: CN-LOH, copy-neutral loss of heterozygosity.

2534

haematologica | 2021; 106(9)

Case Report

Figure 1. Representative pathologic findings of sequential grey zone lymphomas (Case #1). Initial cervical lymph node biopsy shows classic Hodgkin lymphoma (upper panel, A to C). (A) Characteristic Hodgkin and Reed–Sternberg (HRS) cells are present in a polymorphous inflammatory background (hematoxylin and eosin stain[H&E]); the HRS cells are negative for EBER (inset A). (B) The neoplastic HRS cells are positive for CD30 (red, membranous) and weekly positive for PAX-5 (brown, nuclear). (C) They are negative for CD20. (D to F) Lesional brain biopsy shows sequential central nevous system large B-cell lymphoma (lower panel). (D) Diffuse sheets of large lymphoma cells shows centrally located prominent nucleoli (H&E); they are negative for EBER (inset D). (E) The lymphoma cells are diffusely positive for CD30 (red, membranous) with strong nuclear PAX-5 (brown) expression, (F) and express strong and homogeneous CD20.

ation sequencing (NGS) using a capture-based 152 gene custom-designed hematologic malignancy panel was performed on paired cHL and CNS LBCL tumors to assess for genomic alterations as previously described.5 These studies were performed under institutional-approved study protocols. We present the clinicopathologic and genomic features of the paired lesions in this previously unreported presentation of pediatric sequential GZL. Case #1, 17M. A 17 year-old male presented with large mediastinal and supraclavicular masses with disseminated spleen, liver, and bone lesions. Left cervical lymph node sampling revealed the classic histology and immunophenotype of NS-cHL (Figure 1A to C, Table 1). A concomitant outside bone marrow sample performed approximately 3 weeks prior revealed LBCL with sizable clusters of large lymphoma cells, consistent with a diagnosis of synchronous GZL. Staging bone marrow was negative for involvement by lymphoma. Complete remission was achieved after initial treatment with ABVE-PC chemotherapy regimen. Six months, after initial diagnosis (2 months post-therapy), several supra- and infra-tentorial brain lesions and extensive leptomeningeal disease appeared. A biopsy of a CNS lesion revealed diffuse sheets of large lymphoma cells having open chromatin, prominent centrally located nucleoli, and a moderate amount of clear to eosinophilic cytoplasm. The lymphoma cells showed diffuse and strong expression for CD45, CD20, PAX-5, CD30, and expression of CD79a, and were negative for CD15, EBER, and ALK (Figure 1D to F, Table 1). A diagnosis of BCL-U-IND consistent with sequential GZL was rendered. He was treated according to POG9917 Arm A as a bridge to bone marrow transplant with a mismatched unrelated donor and received total body irradiation (450 cGy). He was alive at 81.7 months follow-up.

haematologica | 2021; 106(9)

Case #2, 16F. A 16 year-old female presented with a large mediastinal mass with cervical lymphadenopathy and multiple bilateral renal and pulmonary nodules. NScHL was diagnosed from cervical lymph node biopsy; staging bone marrow was negative. She achieved complete remission after ABVE-PC and radiotherapy to the mediastinal mass and other slow-responding areas of disease. Seven months after initial diagnosis (2 months posttherapy), a solitary right temporal lesion was identified. A biopsy revealed essentially similar morphologic and immunophenotypic findings to the CNS lesion of case #1 (Table 1), and a diagnosis of BCL-U-IND, consistent with sequential GZL was rendered. She was treated with ANHL1131 Group C1 and surgical excision and was alive at 13.5 months follow-up. Molecular findings. The microarray and NGS results are summarized in Table 2. In both NS-cHL, near-diploid male or female genomes and no variants of established or potential clinical significance (Tier I/II, Table 2) were detected consistent with “negative” genomic profiles reported in bulk cHL lesions without Reed-Sternberg cell enrichment.6,7 In case #2, a shared 3.0 MB region of copyneutral loss of heterozygosity (LOH) in chromosome 1p36.11-p35.3 was observed that was most likely germline in origin. Both CNS LBCL harbored complex cytogenomic arrays including 2p16.1 and 9p24.1 gains (detected in both cases, Table 2, denoted by *) and 16p13.3 copy-number abnormalities (case #2 only). LOH of chromosome 6p and gain of chromosome 12p were also observed in both CNS LBCL (Table 2, denoted by *). NGS revealed shared NS-cHL/CNS LBCL variants of uncertain significance (VUS, Tier III) in CREBBP p.T974N (case #1) and RELN p.N352S and KMT2D p.E4694Q (case #2). The sequential CNS LBCL in case #1 harbored addi-

2535

Case Report

tional Tier III variants including APC p.G53R, FAT1 p.L761W, and NOTCH3 p.R2109Q. The sequential CNS LBCL in case #2 harbored pathogenic (Tier I/II) TP53 p.C238R and FBXW7 p.R465H missense variants. In this report, we detailed the clinicopathologic and molecular features of two adolescent patients with sequential GZL involving the CNS. Notably, this is the first report describing CNS involvement as a manifestation of sequential GZL, a finding which expands the clinicopathologic spectrum of this rare pediatric disease. Consistent with previous reports, both patients presented with mediastinal NS-cHL and advanced extranodal disease with similar histopathologic and immunophenotypic findings, and developed GZL in a similar chronologic fashion.4,8 The sequential CNS lesions showed differing morphologic and immunohistochemical profiles with strong and diffuse expression of several B-cell markers and CD30, the latter arguing against an extramediastinal primary mediastinal B-cell lymphoma (PMBCL) diagnosis, and the NS-cHL diagnosis preceded the diagnosis of LBCL temporally establishing the sequential GZL diagnosis. Additionally, the findings of synchronous GZL with subsequent development of sequential GZL in the first patient is also exceptional. Furthermore, unlike previous reports, an early evolution (e.g., second lymphoma diagnosis within 1 year) may not necessarily portend a poor clinical outcome4 given the favorable clinical responses in our two patients and a relatively long term follow-up in the first. Recent molecular characterization of GZL supports the classification of two distinct subtypes of GZL: a "thymic" subtype that occurs in the anterior mediastinum and resembles Epstein-Barr virus (EBV)-negative cHL and PMBCL, and a “non-thymic” subtype which occurs outside the thymus and harbors TP53 mutations in a subset of cases.9,10 In our two patients, the CNS location and mutations in TP53 (case #2) and other associated genes (e.g., CREBBP, RELN, and KMT2D) support a “nonthymic” GZL classification. The presence of complex genomic profiles is also consistent with dysregulated TP53 signaling, and both CNS LBCL harbored complex cytogenomic arrays with copy number abnormalities previously reported in GZL11-13 and frequently reported in cHL and PMBCL.14,15 We acknowledge that a thorough investigation of enriched Reed-Sternberg cells from the cHL lesions and specific subsets of lesional cells may yield valuable molecular insights but this was beyond the scope of the current study. In summary, we present the first report of sequential GZL with CNS involvement in two adolescent patients, and the first clinical genomic profiling of such paired lesions. These lesions showed chromosome aberrations identified in GZLs and NGS mutations associated with non-thymic GZL. These findings expand the clinicopathologic and genomic spectrum of this rare pediatric disease. Cagla Yasa-Benkli,1 Andrea N. Marcogliese,1,2 Jennifer E. Agrusa,2 Adekunle M. Adesina,1,2 Howard L. Weiner,3 Kevin E. Fisher1# and Choladda V. Curry1# #

KEF and CVC contributed equally as co-senior authors.

Department of Pathology & Immunology, Baylor College of Medicine and Texas Children’s Hospital; 2Department of Pediatrics, Baylor College of Medicine and Texas Children's Cancer Center and

2536

Division of Pediatric Neurosurgery, Department of Surgery, Baylor College of Medicine and Texas Children's Hospital, Houston, TX, USA Correspondence: CHOLADDA V. CURRY - ccurry@bcm.edu doi:10.3324/haematol.2021.278936 Received: April 8, 2021. Accepted: June 16, 2021. Pre-published: June 24, 2021. Disclosures: no conflicts of interest to disclose. Contributions: CYB researched the literature, wrote the manuscript, and constructed the tables/figures; ANM, JEA, AMA, and HLW assisted with reviewing medical and pathological records of patients involved, as well as manuscript editing; KEF and CVC conceived the study, interpreted the data, provided feedback and supervision. All authors contributed to patient care, manuscript editing, and evaluation.

References 1. Liang X, Greffe B, Cook B, et al. Gray zone lymphomas in pediatric patients. Pediatr Dev Pathol. 2011;14(1):57-63. 2. Oschlies I, Burkhardt B, Salaverria I, et al. Clinical, pathological and genetic features of primary mediastinal large B-cell lymphomas and mediastinal gray zone lymphomas in children. Haematologica. 2011;96(2):262-268. 3. Swerdlow SH, Campo E, Harris NL, et al. (Eds.) WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues, Revised 4th ed.; IARC: Lyon, France, 2017. 4. Aussedat G, Traverse-Glehen A, Stamatoullas A, et al. Composite and sequential lymphoma between classical Hodgkin lymphoma and primary mediastinal lymphoma/diffuse large B-cell lymphoma, a clinico-pathological series of 25 cases. Br J Haematol. 2020;189(2):244-256. 5. Zhou T, Bloomquist MS, Ferguson LS, et al. Pediatric myeloid sarcoma: a single institution clinicopathologic and molecular analysis. Pediatr Hematol Oncol. 2020;37(1):76-89. 6. Li MM, Datto M, Duncavage EJ, et al. Standards and guidelines for the interpretation and reporting of sequence variants in cancer: a joint consensus recommendation of the Association for Molecular Pathology, American Society of Clinical Oncology, and College of American Pathologists. J Mol Diagn. 2017;19(1):4-23. 7. Tiacci E, Döring C, Brune V, et al. Analyzing primary Hodgkin and Reed-Sternberg cells to capture the molecular and cellular pathogenesis of classical Hodgkin lymphoma. Blood. 2012;120(23):46094620. 8. Perwein T, Lackner H, Ebetsberger-Dachs G, et al. Management of children and adolescents with gray zone lymphoma: a case series. Pediatr Blood Cancer. 2020;67(5):e28206. 9. Sarkozy C, Chong L, Takata K, et al. Gene expression profiling of gray zone lymphoma. Blood Adv. 2020;4(11):2523-2535. 10. Sarkozy C, Hung SS, Chavez EA, et al. Mutational landscape of grey zone lymphoma. Blood. 2021;137(13):1765-1776. 11. Quintanilla-Martinez L, de Jong D, de Mascarel A, et al. Gray zones around diffuse large B cell lymphoma. Conclusions based on the workshop of the XIV meeting of the European Association for Hematopathology and the Society of Hematopathology in Bordeaux, France. J Hematop. 2009;2(4):211-236. 12. Sarkozy C, Molina T, Ghesquières H, et al. Mediastinal gray zone lymphoma: clinico-pathological characteristics and outcomes of 99 patients from the Lymphoma Study Association. Haematologica. 2017;102(1):150-159. 13. Wilson WH, Pittaluga S, Nicolae A, et al. A prospective study of mediastinal gray-zone lymphoma. Blood. 2014;124(10):1563-1569. 14. Joos S, Otaño-Joos MI, Ziegler S, et al. Primary mediastinal (thymic) B-cell lymphoma is characterized by gains of chromosomal material including 9p and amplification of the REL gene. Blood. 1996; 87(4):1571-1578. 15. Kimm LR, deLeeuw RJ, Savage KJ, et al. Frequent occurrence of deletions in primary mediastinal B-cell lymphoma. Genes Chromosomes Cancer. 2007;46(12):1090-1097.

haematologica | 2021; 106(9)