Haematologica, Volume 107, Issue 2 by Haematologica

haematologica Journal of the Ferrata Storti Foundation

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

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

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

Statistical Consultant Catherine Klersy (Pavia)

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

Managing Director Antonio Majocchi (Pavia)

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

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

haematologica Journal of the Ferrata Storti Foundation

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

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

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

haematologica Journal of the Ferrata Storti Foundation

Table of Contents Volume 107, Issue 2: February 2022 About the Cover 351

Images from the Haematologica Atlas of Hematologic Cytology: dyserythropoiesis Rosangela Invernizzi

https://doi.org/10.3324/haematol.2021.280385

Landmark Papers in Hematology 352

The platelet aggregometer Carlo L. Balduini

https://doi.org/10.3324/haematol.2021.280198

Editorials 353

Too much and not enough: revisiting maintenance rituximab in indolent lymphomas Sonali M. Smith

https://doi.org/10.3324/haematol.2021.279101

354

An exciting RXRA mutant revives interest in retinoids for acute myeloid leukemia Fang Qiu and Hugues De The

https://doi.org/10.3324/haematol.2021.279152

356

Are clinical pharmacology studies still needed in childhood acute lymphoblastic leukemia? Valentino Conter and Francesco Ceppi https://doi.org/10.3324/haematol.2021.279059

Review Article 358

Integrating genetic and epigenetic factors in chronic myeloid leukemia risk assessment: toward gene expression-based biomarkers Vaidehi Krishnan

https://doi.org/10.3324/haematol.2021.279317

Articles Acute Lymphoblastic Leukemia 371 Comprehensive analysis of dose intensity of acute lymphoblastic leukemia chemotherapy Seth E. Karol et al.

https://doi.org/10.3324/haematol.2021.278411

Hematopoiesis 381 Reversible switching of leukemic cells to a drug-resistant, stem-like subset via IL-4-mediated cross-talk with mesenchymal stroma Hae-Ri Lee et al.

https://doi.org/10.3324/haematol.2020.269944

393

Aging of human hematopoietic stem cells is linked to changes in Cdc42 activity Amanda Amoah et al.

https://doi.org/10.3324/haematol.2020.269670

Acute Myeloid Leukemia 403 APR-246 induces early cell death by ferroptosis in acute myeloid leukemia Rudy Birsen et al.

https://doi.org/10.3324/haematol.2020.259531

Haematologica 2022; vol. 107 no. 2 - February 2022 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

417

RXRA DT448/9PP generates a dominant active variant capable of inducing maturation in acute myeloid leukemia cells Orsola di Martino et al.

https://doi.org/10.3324/haematol.2021.278603

Cell Therapy & Immunotherapy 427 Human invariant natural killer T cells promote tolerance by preferential apoptosis induction of conventional dendritic cells Hannes Schmid et al.

https://doi.org/10.3324/haematol.2020.267583

437

CD38 knockout natural killer cells expressing an affinity optimized CD38 chimeric antigen receptor successfully target acute myeloid leukemia with reduced effector cell fratricide Mark Gurney et al.

https://doi.org/10.3324/haematol.2020.271908

446

Successful gene therapy of Diamond-Blackfan anemia in a mouse model and human CD34+ cord blood hematopoietic stem cells using a clinically applicable lentiviral vector Yang Liu et al.

https://doi.org/10.3324/haematol.2020.269142

Complications in Hematology 457 Long term follow-up of pediatric-onset Evans syndrome: broad immunopathological manifestations and high treatment burden Thomas Pincez et al.

https://doi.org/10.3324/haematol.2020.271106

Iron Metabolism & its Disorders 467 Risk factors for endocrine complications in transfusion-dependent thalassemia patients on chelation therapy with deferasirox: a risk assessment study from a multi-center nation-wide cohort Maddalena Casale et al.

https://doi.org/10.3324/haematol.2020.272419

478

UBA6 and NDFIP1 regulate the degradation of ferroportin Lisa Traeger et al.

https://doi.org/10.3324/haematol.2021.278530

Non-Hodgkin Lymphoma 489 Early detection of T-cell lymphoma with T follicular helper phenotype by RHOA mutation analysis Rachel Dobson et al.

https://doi.org/10.3324/haematol.2020.265991

500

https://doi.org/10.3324/haematol.2020.274803

Myeloid Biology 510 Unique ethnic features of DDX41 mutations in patients with idiopathic cytopenia of undetermined significance, myelodysplastic syndrome, or acute myeloid leukemia Eun-Ji Choi et al.

https://doi.org/10.3324/haematol.2020.270553

Platelet Biology & its Disorders 519 Sequence-specific 2'-O-methoxyethyl antisense oligonucleotides activate human platelets through glycoprotein VI, triggering formation of platelet-leukocyte aggregates Martina H. Lundberg Slingsby et al.

https://doi.org/10.3324/haematol.2020.260059

Haematologica 2022; vol. 107 no. 2 - February 2022 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

Red Cell Biology & its Disorders 532 Sulfated non-anticoagulant heparin derivative modifies intracellular hemoglobin, inhibits cell sickling in vitro, and prolongs survival of sickle cell mice under hypoxia Osheiza Abdulmalik et al.

https://doi.org/10.3324/haematol.2020.272393

Letters to the Editor 541

SARS-CoV-2 infection in aplastic anemia Daniele Avenoso et al.

https://doi.org/10.3324/haematol.2021.279928

544

The insecticides permethrin and chlorpyrifos show limited genotoxicity and no leukemogenic potential in human and murine hematopoietic stem progenitor cells Virginia C. Rodriguez-Cortez et al.

https://doi.org/10.3324/haematol.2021.279047

550

Somatic STAT3 mutations in CD8+ T cells of healthy blood donors carrying human T-cell leukemia virus type 2 Daehong Kim et al.

https://doi.org/10.3324/haematol.2021.279140

555

Incidence and outcome of SARS-CoV-2 infection in patients with monoclonal gammopathy of undetermined significance: a case-control study Nicola Sgherza et al.

https://doi.org/10.3324/haematol.2021.279895

558

Interferon α-induced SAMHD1 regulates human cultured megakaryocyte apoptosis and proplatelet formation Seema Bhatlekar et al.

https://doi.org/10.3324/haematol.2021.279864

562

Detection of ABL1 kinase domain mutations in therapy-naïve BCR-ABL1-positive acute lymphoblastic leukemia Constance Baer et al.

https://doi.org/10.3324/haematol.2021.279807

Haematologica 2022; vol. 107 no. 2 - February 2022 http://www.haematologica.org/

haematologica Journal of the Ferrata Storti Foundation

The origin of a name that reflects Europe’s cultural roots.

Ancient Greek

Scientific Latin

Modern English

haematologicus (adjective) = related to blood haematologica (adjective, plural and neuter, used as a noun) = hematological subjects The oldest hematology journal, publishing the newest research results. 2020 JCR impact factor = 9.94

ABOUT THE COVER Images from the Haematologica Atlas of Hematologic Cytology: dyserythropoiesis Rosangela Invernizzi University of Pavia, Pavia, Italy E-mail: ROSANGELA INVERNIZZI - rosangela.invernizzi@unipv.it doi:10.3324/haematol.2021.280385

he most frequently altered lineage in myelodysplastic syndromes is the erythroid one. Some of the morphological abnormalities of erythroid precursors are displayed in the Figure, showing representative bone marrow smears. In the top image erythroid hyperplasia with megaloblastoid changes is evident; in addition, note a detached nuclear fragment within the cytoplasm of an erythroblast (center) and a very large, late erythroblast with a pyknotic, irregularly shaped nucleus (top left). Giant multinucleated erythroblasts should also be considered dysplastic (bottom image), but variable degrees of dyserythropoiesis are commonly observed in various hematologic and non-hematologic disorders and have low diagnostic power. It is of critical importance that dyserythropoiesis is not assessed in isolation and, if dysplasia is confined to erythroid cells, other causes of erythroid dysplasia should be considered.1 Disclosures No conflicts of interest to disclose

Reference 1. Invernizzi R. Myelodysplastic syndromes. Haematologica. 2020;105(Suppl 1):78-97.

haematologica | 2022; 107(2)

351

LANDMARK PAPERS IN HEMATOLOGY The platelet aggregometer Carlo L. Balduini Ferrata-Storti Foundation, Pavia, Italy E-mail: carlo.balduini@unipv.it doi:10.3324/haematol.2021.280198

TITLE

Aggregation of blood platelets by adenosine diphosphate and its reversal

AUTHORS

Gustav Victor Rudolph Born

JOURNAL

Nature. 1962;194:927-929. PMID: 13871375

ax Schultze identified platelets in 1865 and described them as "small colorless spherules often grouped together".1 A few years later, in 1882, Giulio Bizzzero realized that the ability of platelets to cluster at the site of vascular damage is the basis of the hemostatic process.2 Although these two fathers of platelets noted that platelets tend to aggregate, the mechanisms of this phenomenon remained mysterious for nearly a century because no technique for studying platelet function was available. The turning point occurred in 1962, when Gustav Born, at that time Professor of Pharmacology at the Royal College of Surgeons of England, published the paper ‘Aggregation of blood platelets by adenosine diphosphate and its reversal' in Nature. In this Letter to the Editor, Born described a simple instrument consisting of a photometer that measures the passage of light through a siliconized tube in which platelets are kept in suspension by a rotating magnetic bar. When platelets aggregate, the optical density of the platelet suspension decreases and light transmission increases. In the same article Born reported that ATP and AMP inhibit platelet aggregation and concluded that 'it is conceivable that AMP or some other substance could be used to inhibit or to reverse platelet aggregation in thrombosis'. The idea of antiplatelet agents was born! Being strictly against patenting anything of potential medical value for mankind, Born did not patent his invention that, in a few years, was marketed by more than ten firms and spread to all laboratories

interested in hemostasis. It is thanks to the aggregometer that Weiss & Aledort demonstrated that the intake of aspirin inhibits platelet aggregation, John Vane and his colleagues showed that aspirin is an inhibitor of the prostaglandin-forming cyclooxygenase, and Hamberg & Samuelsson identified the pro-aggregating substance (now known as) thromboxane A2. Summarizing his work in the field of platelet aggregation, Born wrote "It is gratifying that the feedback hypothesis of platelet aggregation turned out to be explanatory of the remarkable effectiveness of antiplatelet drugs of the aspirin type in the prevention of heart attacks and strokes". There is no doubt that the discovery of antiplatelet agents is one of the most remarkable achievements of pharmacological research of the twentieth century. Moreover, it should not be forgotten that the identification of numerous hemorrhagic diseases due to a functional platelet defect is largely due to Born’s aggregometer, which is still in use in most clinical and research laboratories concerned with thrombosis and hemostasis. The great fame derived from the aggregometer must not make us forget the many other scientific merits of Born, including important studies on atherosclerotic plaques, pioneering use of intravital microscopy and direct recording of white blood cell behavior in small vessels. For his achievements Born received honorary MD or DSci degrees from ten European and US universities, and the Centre for Vascular Research at the University of Edinburgh was named after him.

Born's platelet aggregometer. (A) The first aggregometer produced by Born in the workshop of the Royal College of Surgeons in London. (B) Tracings of platelet aggregation from the paper published in Nature in 1962. The addition of ADP to platelet-rich plasma causes a reduction in the latter’s optical density which is proportional to the extent of platelet aggregation. With low doses of ADP the aggregation is reversible, while with higher doses it becomes irreversible (Reproduced from Nature, with permission).

References 1. Schultze, M. Ein heizbarer Objecttisch und seine Verwendungbei Untersuchungen des Blutes. Archiv für mikroscopische Anatomie. 1865;1:1-42. 2. Bizzozero, J. Ueber einen neuen Formbestandtheil des Blutes und dessen Rolle bei der Thrombose und der Blutgerinnung. Archiv für Pathologische Anatomie und Physiologie und für klinische Medicin 1882;90:261-332.

352

haematologica | 2022; 107(2)

EDITORIALS Too much and not enough: revisiting maintenance rituximab in indolent lymphomas Sonali M. Smith Section of Hematology/Oncology, University of Chicago, Chicago, IL, USA E-mail: SONALI M. SMITH - smsmith@medicine.bsd.uchicago.edu doi:10.3324/haematol.2021.279101

espite being slow-growing and chemosensitive diseases, the inevitable outcome of indolent lymphomas is relapse and prevention of relapse has been an important avenue of investigation for decades. The advent of rituximab allowed consideration of a less toxic (compared to chemotherapy) opportunity to maintain remission following induction regimens without the need for prolonged exposure to cytotoxic agents. The common questions have focused on both the schedule and the duration of maintenance therapy, with the goal to improve progressionfree or overall survival; in an ideal world, maintenance strategies might even seek to cure. Rituximab pharmacokinetics and/or impact on B-cell depletion and subsequent recovery provide sufficient rationale for delivering maintenance rituximab with a variety of approaches: one dose every 8 weeks, one dose every 12 weeks, or four weekly doses every 6 months.1,2 Although none of these schedules has been directly compared, small retrospective reviews suggest relative equivalence in terms of efficacy and small differences in terms of toxicity.3 In frontline follicular lymphoma, the PRIMA trial established one dose of maintenance rituximab every 8 weeks based on achieving a trough level of 25 mg/mL in the majority of patients, and this has arguably become the most common schedule.4 It is important to note that there remains a lack of a survival advantage for maintenance rituximab in frontline follicular lymphoma, but 10-year data show impressive and persistent disease control and validate maintenance rituximab for 2 years as an appropriate option to improve progression-free survival in patients with high-tumor burden follicular lymphoma. While the schedule of rituximab maintenance can be justified based on pharmacokinetics, the duration of rituximab maintenance is more empirically derived. Given the relatively favorable toxicity profile even with prolonged administration, studies have ranged from several limited doses to 5 years of treatment to indefinite treatment. In the relapsed setting, a meta-analysis from a decade ago suggested improved overall survival for maintenance rituximab in relapsed/refractory follicular lymphoma, although it is critical to acknowledge that the majority of trials included in this study had involved chemotherapy induction and not chemoimmunotherapy induction.5 Based on improved progression-free survival and a lure of improved overall survival, is more maintenance better? In this issue of Haematologica,6 Rule and colleagues present the final results of the MabCute trial. This prospective, international, randomized phase III trial sought to determine the added benefit of extended rituximab dosing beyond 2 years in responding patients with indolent lymphomas. All patients had relapsed or refractory indolent lymphomas, and could receive any chemoimmunotherapy induction followed by 2 years of rituximab maintenance. Using 2007 response criteria, responding patients at 2 years were randomized to

haematologica | 2022; 107(2)

receive either an additional 2 years of maintenance therapy with subcutaneous rituximab or active observation. With a primary endpoint of progression-free survival, 274 patients were randomized and the median follow-up is 28 months. Adverse events were slightly increased in the extended maintenance arm. During the observation period, the number of progression events was quite low in both arms, and the trial is now closed without a clear signal of improved progression-free survival for the extended maintenance arm; no conclusions can be made regarding survival. The MabCute trial can thus be added to the list of trials showing no advantage from prolonged maintenance with anti-CD20 targeting strategies if disease control, toxicity, and overall survival are considered collectively. As one example, the SAKK 35/03 trial randomized a heterogeneous group of patients with follicular lymphoma to receive either rituximab every other month for four administrations or rituximab every other month for 5 years;7 patients in this trial could have had treatment-naïve, relapsed, or refractory disease and all received induction therapy with rituximab monotherapy and not chemoimmunotherapy. While event-free survival was improved, there was more toxicity and no impact on survival in the prolonged treatment arm. Other key trials have shown no benefit from rituximab treatment at relapse compared to a maintenance approach;8 specifically, the RESORT trial found no difference in time to next treatment and no improvement in overall survival between maintenance and retreatment in low tumor burden indolent lymphomas. Finally, the induction chemotherapy backbone also influences the risk-benefit profile; for example, the GALLIUM trial observed that patients receiving bendamustinebased induction had more toxicity and even increased mortality, particularly during the maintenance component of therapy.9 It is worth noting that the majority of patients in the current trial also received bendamustine-based regimens and no toxicity signal was seen with the extended maintenance, but this may be because of drop-out during the initial maintenance component. The COVID-19 pandemic has forced us to re-evaluate data with an additional critical lens related to treatment-associated B-cell suppression. Early studies during this pandemic showed that patients on immune suppression or those on chemotherapy suffer more severe complications related to SARS-CoV2, and mortality in patients with hematologic malignancies is high.10 Since maintenance rituximab has yet to offer cure or improved overall survival after chemoimmunotherapy induction, it seems more appropriate to identify the minimum number of doses rather than trying to expand or extend treatment. Indeed, only a minority of patients with blood cancers mount a sufficient response to accordingly, recent scientific society vaccines;11 (https://www.hematology.org/covid-19/ash-astct-covid-19-andvaccines) and advocacy guidelines12 highlight that patients receiving B-cell-directed therapies have attenuated or even

353

Editorials

absent responses to vaccination against SARS-CoV2 and cannot abandon social distancing and masking precautions. In the end, there is no advantage from prolonged maintenance in indolent lymphomas, and 2 years should remain the standard duration if maintenance rituximab is offered. The MabCute trial thus supports that extended dosing of rituximab is both too much, and also not enough to offer cure or improve overall survival. Future studies should focus on identifying the minimum number of maintenance doses needed to improve outcomes, particularly in light of a pandemic threat. Disclosures No conflicts of interest to disclose.

References 1. Berinstein NL, Grillo-Lopez AJ, White CA, et al. Association of serum rituximab (IDEC-C2B8) concentration and anti-tumor response in the treatment of recurrent low-grade or follicular non-Hodgkin's lymphoma. Ann Oncol. 1998;9(9):995-1001. 2. Gordan LN, Grow WB, Pusateri A, Douglas V, Mendenhall NP, Lynch JW. Phase II trial of individualized rituximab dosing for patients with CD20-positive lymphoproliferative disorders. J Clin Oncol. 2005;23(6):1096-1102. 3. Nabhan C, Ollberding NJ, Villines D, et al. A systematic review of comparative schedule-related toxicities with maintenance rituximab in follicular and mantle cell lymphomas. Leuk Lymphoma. 2014;55(6):1288-1294.

4. Salles G, Seymour JF, Offner F, et al. Rituximab maintenance for 2 years in patients with high tumour burden follicular lymphoma responding to rituximab plus chemotherapy (PRIMA): a phase 3, randomised controlled trial. Lancet. 2011;377(9759):42-51. 5. Vidal L, Gafter-Gvili A, Salles G, et al. Rituximab maintenance for the treatment of patients with follicular lymphoma: an updated systematic review and meta-analysis of randomized trials. J Natl Cancer Inst. 2011;103(23):1799-1806. 6. Rule S, Barreto W, Briones J, et al. Efficacy and safety assessment of prolonged maintenance with subcutaneous rituximab in patients with relapsed or refractory indolent non-Hodgkin lymphoma: results of the phase III MabCute study. Haematologica. 2022;107(2):500-509. 7. Taverna C, Martinelli G, Hitz F, et al. Rituximab maintenance for a maximum of 5 years after single-agent rituximab induction in follicular lymphoma: results of the randomized controlled phase III trial SAKK 35/03. J Clin Oncol. 2016;34(5):495-500. 8. Kahl BS, Hong F, Williams ME, et al. Rituximab extended schedule or re-treatment trial for low-tumor burden follicular lymphoma: Eastern Cooperative Oncology Group protocol E4402. J Clin Oncol. 2014;32(28):3096-3102. 9. Hiddemann W, Barbui AM, Canales MA, et al. Immunochemotherapy with obinutuzumab or rituximab for previously untreated follicular lymphoma in the GALLIUM study: influence of chemotherapy on efficacy and safety. J Clin Oncol. 2018;36(23):2395-2404. 10. Mato AR, Roeker LE, Lamanna N, et al. Outcomes of COVID-19 in patients with CLL: a multicenter international experience. Blood. 2020;136(10):1134-1143. 11. Herishanu Y, Avivi I, Aharon A, et al. Efficacy of the BNT162b2 mRNA COVID-19 vaccine in patients with chronic lymphocytic leukemia. Blood. 2021;137(23):3165-3173. 12. Proceedings of the COVID-19 Vaccine Panel: a Lymphoma Research Foundation White Paper. https://lymphoma.org/wpcontent/uploads/2021/04/Proceedings-of-the-COVID-19-VaccinePanel_LRF-White-Paper_April-2021.pdf; 2021. (Last accessed May 12, 2021)

An exciting RXRA mutant revives interest in retinoids for acute myeloid leukemia Fang Qiu1 and Hugues De The1-3 1

INSERM UMR 944, CNRS UMR 7212, Université de Paris, IRSL, Hôpital Saint Louis; 2Collège de France, Oncologie Cellulaire et Moléculaire, PSL University, INSERM UMR 1050, CNRS UMR 7241, and 3Department of Hematology, Hôpital Saint Louis (Assistance publique Hôpitaux de Paris) and Paris University, Paris, France E-mail: HUGUES DE THÉ - hugues.dethe@inserm.fr doi:10.3324/haematol.2021.279152

he idea that "one size fits all" is obviously outdated for acute myeloid leukemia (AML) therapy: tomorrow’s treatments will depend on phenotypically or genetically defined subtypes. The most striking example is acute promyelocytic leukemia (APL), driven by the PMLRARA fusion protein. In APL, a subtype that accounts for 5% of cases of AML, a combination of two targeted agents, all trans-retinoic acid (ATRA) and arsenic trioxide, cures over 90% of patients through PML-RARA driver degradation, differentiation and restoration of PML-dependent senescence.1 These clinical successes have spurred attempts to harness the power of retinoids in other cancers. Unfortunately, ATRA treatment alone remains poorly effective in most non-APL AML.2 Retinoid signaling is complex and still incompletely understood.3 ATRA acts primarily through heterodimeric complexes of retinoic acid receptors (RAR) assembled with retinoid X receptors (RXR). These RXR are key heterodimerization partners of many class-II nuclear receptors and may be ligand-dependent transcription factors or

354

silent receptors, allowing sequence-specific DNA recognition.4 Hence, therapeutic targeting of RXR could be a strategy to activate targets under the control of the RXR/RAR transcriptional complex. However, in principle, RXR/RAR signaling cannot be activated by RXR ligands alone, at least in part because co-repressors remain firmly bound to RAR. This may be modulated by other signaling cascades/second messengers, such as cAMP.5 In AML, this simple view has been challenged. RXR ligands (rexinoids, such as bexarotene) may exert some differentiating effects ex vivo and in vivo.5,6 Hematopoietic cells and some AML express endogenous RXRA ligands.7,8 Two recent studies have revived interest in RXRA signaling in AML. The first demonstrated that, in AML driven by KMT2A-MLLT3, rexinoids partially suppressed AML growth and triggered differentiation.8 Moreover, genetic ablation of RXR accelerated AML growth, while concomitant activation of both RXRA and RARA precipitated differentiation or apoptosis. It is hoped that dual activation of these key regulators may harness retinoids more effi-

haematologica | 2022; 107(2)

Editorials

Figure 1. Schematic summary of the effects of constitutively active RXRA DT448/9PP. (A) Under normal circumstances, the transcriptional activity of RXRA heterodimerized with other nuclear receptors (NR), including RARA, remains silent, because of co-repressor binding. Selective agonists activate RXRA/NR-driven transcription, resulting in cellular differentiation and growth arrest. (B) Mutations of residues 488/9 in RXRA allow potent ligand-independent transcriptional activation and drive differentiation.

ciently in AML.2 In the second study, published last year in Haematologica, di Martino et al. report a serendipitously identified activating mutation in RXRA (RXRA DT448/9PP), which potently activates rexinoid/retinoid downstream signaling and suffices to induce terminal differentiation of KMT2A-MLLT3-transformed cells.9 The Cterminal helix 12 or AF-2 helix of RXRA, is a critical determinant of ligand-dependent transcriptional activity through control of co-activator/co-repressor binding. Surprisingly, di Martino et al. demonstrated that RXRA DT448/9PP overexpression resulted in enhanced transcriptional activity leading to multiple features of differentiation, notably loss of colony-forming ability, in KMT2AMLLT3-transformed AML cells. Amazingly, this constitutively active RXRA variant binds co-activators completely independently of ligands. Accordingly, transactivation could not be abrogated or further boosted by selective antagonists of RXR or other nuclear receptors, or their agonists, respectively (Figure 1). These intriguing observations imply that even though rexinoids and retinoids synergize for myeloid differentiation of those AML,8 more profound "unconventional" activation by RXRA can initiate terminal differentiation. This master transcriptional regulatory complex deserves further studies to mechanistically decipher how it can become so potent in the absence of ligands. Issues of partner proteins, post-translational modifications or non-coding RNA, all come to mind. Whatever the molecular mechanism, these observations suggest that the RXRA/RARA axis, when super-activated, has the potential to initiate terminal differentiation of some AML cells. Further studies should determine which AML exhibit this exquisite sensitivity to RXRA signaling. This re-emerging

haematologica | 2022; 107(2)

theme of retinoid sensitivity in non-APL AML2 could be particularly important in the context of combinations of treatment, particularly with decitabine, as encouraging clinical trials have been published recently,10 with more likely to come. Disclosures No conflicts of interest to disclose.

References 1. de Thé H, Pandolfi PP, Chen Z. Acute promyelocytic leukemia: a paradigm for oncoprotein-targeted cure. Cancer Cell. 2017;32(5):552-560. 2. Geoffroy MC, Esnault C, de Thé H. Retinoids in hematology: a timely revival? Blood. 2021;137(18):2429-2437. 3. de Thé H. Differentiation therapy revisited. Nat Rev Cancer. 2018;18(2):117-127. 4. De Bosscher K, Desmet SJ, Clarisse D, Estébanez-Perpiña E, Brunsveld L. Nuclear receptor crosstalk-defining the mechanisms for therapeutic innovation. Nat Rev Endocrinol. 2020;16(7):363-377. 5. Altucci L, Rossin A, Hirsch O. Rexinoid-triggered differentiation and tumor-selective apoptosis of acute myeloid leukemia by protein kinase A-mediated desubordination of retinoid X receptor. Cancer Res. 2005;65(19):8754-8765. 6. Sanchez PV, Glantz ST, Scotland S, Kasner MT, Carroll M. Induced differentiation of acute myeloid leukemia cells by activation of retinoid X and liver X receptors. Leukemia. 2014;28(4):749-760. 7. Niu H, Fujiwara H, di Martino O. Endogenous retinoid X receptor ligands in mouse hematopoietic cells. Sci Signal. 2017;10(503):eaan1011. 8. Di Martino O, Niu H, Hadwiger G. Endogenous and combination retinoids are active in myelomonocytic leukemias. Haematologica. 2021;106(4):1008-1021. 9. Di Martino O, Ferris MA, Hadwiger G, et al. RXRA DT448/9PP generates a dominant active variant capable of inducing maturation in acute myeloid leukemia cells. Haematologica. 2022;107(2):417-426. 10. Lübbert M, Grishina O, Schmoor C. Valproate and retinoic acid in combination with decitabine in elderly nonfit patients with acute myeloid leukemia: results of a multicenter, randomized, 2 x 2, phase II trial. J Clin Oncol. 2020;38(3):257-270.

355

Editorials

Are clinical pharmacology studies still needed in childhood acute lymphoblastic leukemia? Valentino Conter1 and Francesco Ceppi2 1

Pediatric Hemato-Oncology, Fondazione MBBM, University Milano Bicocca, Ospedale San Gerardo, Monza, Italy and 2Pediatric Hematology-Oncology Unit, Division of Pediatrics, Department Woman-Mother-Child, University Hospital of Lausanne & University of Lausanne, Lausanne, Switzerland E-mail: VALENTINO CONTER - valentino.conter@gmail.com doi:10.3324/haematol.2021.279059

n this issue of Haematologica, Karol et al. report a study on dose intensities for all drugs in two consecutive acute lymphoblastic leukemia (ALL) clinical trials at St. Jude Children’s Research Hospital, which differed in their asparaginase formulation and intensity.1 The amount of data is impressive, with more than 500,000 dosing records. The main message of the manuscript is that the lack of benefit from increased asparaginase intensity may be due to the decrease of dose intensity of other drugs, induced by the additional treatment with asparaginase. It is widely recognized that intensity of chemotherapy delivered has an impact on outcome and that drug interactions, which are difficult to assess, can influence anticancer activity and acute and/or late toxicity too. The fast improvement of outcome in childhood ALL in the last three decades of the last century were strictly associated with progressive treatment intensity. Dr. Riehm was the pioneer in this historical process, which was thereafter pursued by all major pediatric oncology groups. In the early 1990s, Sallan summarized the Dana-Faber Cancer Institute (DFCI) experience, largely based on treatment intensification with asparaginase, with the words “More is better!”,2 and Niemeyer (with Riehm and Sallan) suggested that merging the intensive elements of BerlinFrankfurt-Münster (BFM) and DFCI protocols would be a logical program to improve outcomes.3 Various attempts were made in this frame, sometimes successfully, such as in the Children’s Cancer Study Group (CCSG) study with Augmented BFM.4 Most studies did not, however, show any benefit in intensive BFM-oriented protocols, either from additional asparaginase treatment as done in Associazione Italiana Ematologia Oncologia Pediatrica (AIEOP) ALL 9102,5 European Organization for Research and Treatment of Cancer - Children Leukemia Group (EORTC-CLG) 58951,6 Nordic Society of Pediatric Hematology and Oncology (NOPHO) ALL-2008,7 and BFM ALL 90 trials,8 or from the marked intensification in the Children’s Oncology Group (COG) AALL1131 trial with clofarabine, which was interrupted early due to an excess of toxicity.9 This general experience has led to a consensus that treatment intensity in childhood ALL may have reached the maximum tolerated doses, so that further improvement can only be obtained by precision medicine based on targeted therapies. However, most children with ALL are cured with conventional chemotherapy, which can be further optimized and tailored thanks to the progressive improvement of biology-based stratification. The study by Karol et al. shows that room remains for improvement of chemotherapy, although this cannot be achieved by a simple protocol therapy intensification.1 Asparaginase is a drug with a unique mechanism of action, and there are no suggested alternatives to replace

356

it in patients who cannot be treated with the drug. DFCI studies showed that these patients have a poorer outcome. In this context it quite interesting the finding that patients with low asparaginase dose intensity, a higher systemic methotrexate dose intensity compensated for the low asparaginase dose intensity. The often neglected and yet very relevant aspect of oral medications administered at home is also of note. In the study reported in this issue, there is the apparent paradox of higher relapse rate associated with higher dose intensity for mercaptopurine, which the authors suggest might reflect low treatment adherence for oral medications at home (not measured in this study), in keeping with the findings of the COG AALL03N1 study, in which it was shown that an adherence rate below 90% to maintenance therapy was associated with an increased relapse risk.10 Although the expectation for further improvements in the treatment of childhood ALL is mostly based on innovative immunological or targeted therapies, pharmacological studies remain crucial to improve the therapeutic index of combinations of antineoplastic agents. To this purpose, it must be considered that simple measurement of duration of treatment phases, incidence of severe adverse effects, and dose intensities of single agents may be inadequate or even misleading. What is needed in order to optimize precision personalized treatment in childhood ALL are comprehensive investigations of compliance/adherence for all drugs, drug interactions and bioavailability, and germline and tumor sensitivity. Disclosures No conflicts of interest to disclose. Contributions The two authors contributed equally.

References 1. Karol SE, Pei D, Smith CA, et al. Comprehensive analysis of dose intensity of acute lymphoblastic leukemia chemotherapy. Haematologica. 2022;107(2):371-380. 2. Sallan SE, Gelber RD, Kimball V, Donnelly M, Cohen HJ. More is better! Update of Dana-Farber Cancer Institute/Children's Hospital childhood acute lymphoblastic leukemia trials. Haematol Blood Transfus. 1990;33:459-466. 3. Niemeyer CM, Reiter A, Riehm H, Donnelly M, Gelber RD, Sallan SE. Comparative results of two intensive treatment programs for childhood acute lymphoblastic leukemia: the Berlin-FrankfurtMünster and Dana-Farber Cancer Institute protocols. Ann Oncol. 1991;2(10):745-749. 4. Nachman J, Sather HN, Gaynon PS, Lukens JN, Wolff L, Trigg ME. Augmented Berlin-Frankfurt-Munster therapy abrogates the adverse prognostic significance of slow early response to induction chemotherapy for children and adolescents with acute lymphoblastic leukemia and unfavorable presenting features: a report from the Children's Cancer Group. J Clin Oncol. 1997;15(6):2222-2230. 5. Rizzari C, Valsecchi MG, Aricò M, et al. Effect of protracted high-

haematologica | 2022; 107(2)

Editorials

dose L-asparaginase given as a second exposure in a Berlin-FrankfurtMünster-based treatment: results of the randomized 9102 intermediate-risk childhood acute lymphoblastic leukemia study--a report from the Associazione Italiana Ematologia Oncologia Pediatrica. J Clin Oncol. 2001;19(5):1297-1303. 6. Mondelaers V, Suciu S, De Moerloose B, et al. Prolonged versus standard native E. coli asparaginase therapy in childhood acute lymphoblastic leukemia and non-Hodgkin lymphoma: final results of the EORTC-CLG randomized phase III trial 58951. Haematologica. 2017;102(10):1727-1738. 7. Toft N, Birgens H, Abrahamsson J, et al. Results of NOPHO ALL2008 treatment for patients aged 1-45 years with acute lymphoblastic leukemia. Leukemia. 2018;32(3):606-615.

haematologica | 2022; 107(2)

8. Schrappe M, Reiter A, Ludwig WD, et al. Improved outcome in childhood acute lymphoblastic leukemia despite reduced use of anthracyclines and cranial radiotherapy: results of trial ALL-BFM 90. German-Austrian-Swiss ALL-BFM Study Group. Blood. 2000;95(11):3310-3322. 9. Salzer WL, Burke MJ, Devidas M, et al. Toxicity associated with intensive postinduction therapy incorporating clofarabine in the very high-risk stratum of patients with newly diagnosed high-risk Blymphoblastic leukemia: a report from the Children's Oncology Group study AALL1131. Cancer. 2018;124(6):1150-1159. 10. Bhatia S, Landier W, Hageman L, et al. 6MP adherence in a multiracial cohort of children with acute lymphoblastic leukemia: a Children's Oncology Group study. Blood. 2014;124(15):2345-2353.

357

REVIEW ARTICLE Ferrata Storti Foundation

Integrating genetic and epigenetic factors in chronic myeloid leukemia risk assessment: toward gene expression-based biomarkers Vaidehi Krishnan,1,2 Dennis Dong Hwan Kim,2,3 Timothy P. Hughes2,4,5,6 Susan Branford2,4,7,8 and S. Tiong Ong1,2,9,10,11 Cancer and Stem Cell Biology Signature Research Program, Duke-NUS Medical School, Singapore, Singapore; 2International Chronic Myeloid Leukaemia Foundation, Bexhill on Sea, UK; 3Department of Medical Oncology and Hematology, Princess Margaret Cancer Center, University Health Network, University of Toronto, Toronto, Ontario, Canada; 4School of Medicine, University of Adelaide, Adelaide, Australia; 5South Australian Health & Medical Research Institute, Adelaide, Australia; 6Department of Hematology, Royal Adelaide Hospital, Adelaide, Australia; 7Department of Genetics and Molecular Pathology, Center for Cancer Biology, SA Pathology, Adelaide, Australia; 8School of Pharmacy and Medical Science, University of South Australia, Adelaide, Australia; 9Department of Haematology, Singapore General Hospital, Singapore; 10Department of Medical Oncology, National Cancer Centre Singapore, Singapore and 11Department of Medicine, Duke University Medical Center, Durham, NC, USA 1

Haematologica 2022 Volume 107(2):358-370

ABSTRACT

Correspondence: S. TIONG ONG sintiong.ong@duke-nus.edu.sg Received: July 16, 2021. Accepted: September 28, 2021. Pre-published: October 7 2021. https://doi.org/10.3324/haematol.2021.279317

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

358

ancer treatment is constantly evolving from a one-size-fits-all towards bespoke approaches for each patient. In certain solid cancers, including breast and lung, tumor genome profiling has been incorporated into therapeutic decision-making. For chronic phase chronic myeloid leukemia (CML), while tyrosine kinase inhibitor therapy is the standard treatment, current clinical scoring systems cannot accurately predict the heterogeneous treatment outcomes observed in patients. Biomarkers capable of segregating patients according to outcome at diagnosis are needed to improve management, and facilitate enrollment in clinical trials seeking to prevent blast crisis transformation and improve the depth of molecular responses. To this end, gene expression (GE) profiling studies have evaluated whether GE signatures at diagnosis are clinically informative. Patient material from a variety of sources has been profiled using microarrays, RNA sequencing and, more recently, single-cell RNA sequencing. However, differences in the cell types profiled, the technologies used, and the inherent complexities associated with the interpretation of genomic data pose challenges in distilling GE datasets into biomarkers with clinical utility. The goal of this paper is to review previous studies evaluating GE profiling in CML, and explore their potential as risk assessment tools for individualized CML treatment. We also review the contribution that acquired mutations, including those seen in clonal hematopoiesis, make to GE profiles, and how a model integrating contributions of genetic and epigenetic factors in resistance to tyrosine kinase inhibitors and blast crisis transformation can define a route to GE-based biomarkers. Finally, we outline a four-stage approach for the development of GE-based biomarkers in CML.

Introduction Chronic myeloid leukemia (CML) is a clonal disorder of the hematopoietic stem cell compartment defined and driven by the BCR-ABL1 gene rearrangement and the tyrosine kinase it encodes.1 Clinically, it is accompanied by an expansion of mostly myelo-erythroid progenitors that maintain the ability to differentiate terminally into neutrophils. Prior to the introduction of ABL1 tyrosine kinase inhibitors (TKI), most patients would progress to a terminal blast crisis (BC) stage marked by the acquisition of additional genetic abnormalities within an average of 5-7 years.2 In this stage, the clinico-pathological features were the inexorable accumulation of either myeloid or lymphoid progenitors that had acquired aber-

haematologica | 2022; 107(2)

GE-based biomarkers in CML

rant self-renewal properties, broad resistance to cytotoxic therapies, and eventual patient demise from bone marrow failure.2 The arrival of TKI at the turn of the century resulted in remarkable responses, such that most individuals treated in chronic phase (CP) CML can expect to achieve near-normal life expectancies.3 Nevertheless, CML-related deaths are still reported, mainly due to resistance and progression to BC, especially in the first few years of treatment.4

Current treatment aims and features of an ideal biomarker Current therapeutic aims are directed at achieving sufficiently deep molecular responses that the risks of BC transformation are effectively negligible and, in the longer-term, increasing the rates of treatment-free remission.5-7 Clinical guidelines toward achieving deep molecular responses have been reviewed elsewhere,8 and at their core, prescribe the measurement of BCR-ABL1 transcript levels using the International Scale (IS) every 3 months as a readout of the depth of the response to TKI. In turn, the depth of TKI response serves as a critical biomarker guiding patient management and prognostication (Figure 1). Given current treatment goals, an ideal biomarker would accurately predict patients who will achieve a deep molecular response with first-line TKI, or require a switch to alternative therapy, and, among those who

achieve a deep molecular response, those who will be able to stop TKI successfully (Figure 1). The biomarker would be informative from the time of diagnosis and prior to TKI initiation, since this would enable early stratification of patients for therapy with a first-generation versus a second/third-generation TKI, allosteric BCRABL1 inhibitor, a clinical trial, or preparation for allogeneic transplantation. Additionally, among patients who meet the criteria for stopping TKI therapy, the ideal biomarker would identify additional therapies that would enhance treatment-free remissions. Finally, gene expression (GE)-based biomarkers should be clinically robust, and widely available among centers and regions in both low and high Human Development Index countries.9

Why gene expression-based biomarkers? Contributions from genetic and epigenetic mediators to TKI resistance and BC transformation are well documented,10-14 and it is axiomatic that genetic or epigenetic factors mediating these outcomes will contribute to a cell’s GE signature. Accordingly, GE signatures offer a molecular profile that integrates risk factors encoded by both mutations and epigenetic states. However, faithfully extracting and interpreting GE-based information in clinical settings is challenging. Barriers to adoption include technical limitations, logistical factors, as well as differences in study design and data analysis, and are described below.

Figure 1. Features of an ideal chronic myeloid leukemia biomarker. Curves indicate changes in BCR-ABL1 transcript levels, meausred using the International Scale (BCR-ABL1IS), following initation of tyrosine kinase inhibitor (TKI) therapy in patients with chronic phase (CP) chronic myeloid leukemia (CML). The corresponding molecular response (MR) value is provided next to the BCR-ABLIS value. Green, orange, and red curves are representative of patients in European LeukemiaNet 2020 ‘optimal’, ‘warning’, and ‘failure’ cateogories respectively. Major (MMR) and deep (DMR) molecular remissions are defined as 0.1% (MR 3) and 0.01% (MR4) BCRABLIS, respectively. Green, orange, and yellow bullet points indicate guidelines for each category of response. Gray boxes describe predictive capabilities of an ideal biomarker. HCT: hematopoietic stem cell transplant.

haematologica | 2022; 107(2)

359

V. Krishnan et al.

Despite the barriers, recent advances in technological and computational platforms are enabling the interrogation of patient samples on an unprecedented scale, and are being translated into robust technical assays on patient material that are reproducible in clinical laboratories.15 Such advances may eventually result in the identification of pretreatment biomarkers that not only predict TKI resistance but suggest alternative non-BCR-ABL1-targeting therapies to pre-empt the emergence of clinical resistance. Accordingly, it is timely to review the results of GE studies using primary patient material annotated for clinical outcomes, and assess how genetic and epigenetic factors associated with treatment outcome contribute to GE signatures. In doing so, it is also important to develop models incorporating the interplay between genetic and epigenetic factors, and determine how best to use the resulting GE outputs to understand and predict CML drug resistance and transformation. Finally, it is incumbent on the CML community to outline the practical steps needed for the clinical development of GE-based biomarkers in CML.

Gene expression signatures associated with resistance to tyrosine kinase inhibitors Since the beginning of the TKI era, a variety of diagnostic material from CP patients has been used to discover TKI-resistance GE signatures (Table 1). Here, we review the key conclusions from these studies.

Gene expression using peripheral blood In the earliest research by Kaneta et al.16 and McLean et al.17 microarray studies were conducted on blood from imatinib responders and non-responders. Apart from CBLB, which was downregulated in responders, there was no overlap between the two datasets. De Lavallade et al. conducted microarray studies on peripheral blood mononuclear cells to identify a 105-gene set that was enriched in imatinib non-responders, comprising mainly genes in cell cycle and DNA repair pathways.18 However, the GE signature could be validated only in an imatinibtreated cohort but not in a cohort treated with interferona. As a targeted approach, the expression of 21 genes associated with TKI responses and disease progression was studied by Zhang et al.19 Increased PTGS1 expression was the only gene that differentiated primary imatinib-resistant patients from responders, while 15 genes distinguished CP from BC. Twelve genes distinguished imatinib-responsive from secondary imatinib-resistant CML without BCR-ABL1 mutations, of which LYN, JAK2, PTPN22 and CEBPA downregulation was shared with BC samples. The study concluded that at least some features of secondary imatinib resistance overlap with BC transformation. More recently, Kok et al. conducted microarray-based analysis on diagnostic blood from 96 CP patients from the TIDEL-II trial to predict failure of early molecular response,20 which correlates with inferior long-term outcomes.21,22 Three hundred sixty-five differentially expressed genes were identified which were enriched for ‘cell cycle’ and ‘stemness’ (MYC, HOXA9, b-catenin) but depleted for ‘immune-response’ categories in the group with early molecular response failure. A binary classification model was built to predict early molecular response 360

failure based on 17 genes and the signature was validated in an independent cohort. Of these, eight genes IGFBP2, SRSF11, BAX, CDKN1B, BNIP3L, FZD7, PRSS57, and RPS28 intersected with findings of previous CML TKIresistance and progression studies. This study demonstrated that GE information from diagnostic samples could predict events long in the future, including major molecular response (MMR) at 24 months, MR4.5 at 5 years, and BC transformation.

Gene expression using bone marrow Independently, a series of studies used unselected bone marrow for comparisons of GE between groups of patients with different treatment responses. Frank et al. identified a 128 GE signature associated with imatinib resistance, specifically in an interferon-a pre-treated cohort. Differentially expressed genes were involved in apoptosis (CASP9, TRAP1), DNA repair (MSH3, DDB2), oxidative stress protection (GSS, PON2, VNN1) and centrosomes (ID1).23 Villuendas et al.24 identified 46 differentially expressed genes of which a six-gene prediction score (BIRC4, FZD7, IKBKB, IL-7R, TNC, VWF) that correlated with imatinib resistance after interferon-a failure developed. Differentially expressed genes were involved in cell adhesion (TNC and SCAM-1), drug metabolism (COX1 or PTGS1), protein tyrosine kinases (MKNK1), and phosphatases (BTK and PTPN22). Notably, the MKNK1/2 kinases have been shown by two independent groups to be involved in BC transformation.25,26 In contrast to the prior studies, Crossman et al. found no differentially expressed genes between the imatinib responder categories. The use of mixed peripheral blood and bone marrow samples, unselected white blood cells and a heterogeneous cohort of patients in late CP and heavily pre-treated, were suggested as potential reasons for the negative results.27 The important conclusion was that GE comparisons should be made on purified CD34+ cells. Indeed, in a meta-analysis comparing six published GE studies in CML, DDX11, MSH5, and RAB11FIP3 were the only genes coincident between any two of the studies.28 The small differences in differential GE between responder groups, different GE platforms, different statistical methods and different sources of cells profiled were suggested reasons for the poor intersection. The disappointing results from unselected peripheral blood and bone marrow provided the impetus to isolate and study CD34+ fractions.

Gene expression using CD34+ cells McWeeney et al. were the first group to use CD34+ cells from diagnostic bone marrow.11 Cell adhesion genes were upregulated in imatinib-resistant patients suggesting that CD34+ cells may establish more adhesive interactions with the bone marrow milieu. The enrichment for bcatenin binding targets suggested activated Wnt/b-catenin signaling in imatinib-resistant patients, a feature shared with CD34+ progenitors from BC.26,29 The authors concluded that primary resistance to imatinib might reflect more advanced disease progression. A 75-probe minimal gene classifier predicted 88% of responders and 83% of nonresponders in a validation cohort. Importantly, the authors of this paper compared their GE signatures to those predicting early BC transformation, as discussed below, and provided an important resource for validation and comparison of other CD34+-based GE datasets. haematologica | 2022; 107(2)

Diagnostic

79 DEG were identified. 15 or 30 genes were used to develop a prediction score to separate TKIresponders from non-responders

Time sample taken Unselected/ CD34+ cells; PB/BM Platform

Comments

haematologica | 2022; 107(2)

CP 29, included patients previously on IFN-a

Crossman et al., 2004

No DEG were identified between TKI responders and non-responders

A 6-gene prediction model was constructed which could predict MCyR at 12 months

Predictive genes associated with Wnt signaling, cell adhesion, NK-kB, apoptosis, DNA repair

46 MCyR at 12 months

Microarray (CNIO OncoChip)

Unselected; BM

Diagnostic BM

CP 32, 12 validation

Villuendas et al., 2006

A 128-gene predictor of primary cytogenetic resistance to imatinib was identified

Diagnostic PB & BM Unselected; total WBC from BM & PB Microarray (Affymetrix HG-U133A) 128 R=MCyR (≤35% Ph+); NR= ≥35% Ph+ at 12 months Predictive genes enriched for transcriptional regulation of apoptosis, oxidative stress, DNA repair, centrosomal genes

CP 23 R; 11 NR

Frank et al., 2006

de Lavallade et al., 2010

Unselected; MNC

Diagnostic PB

CP 96 (discovery); CP 88 (validation); CP 132 (nilotinib Rx).

Kok et al., 2019

Predictive genes Predictive genes involved in TKI enriched for DNA influx/efflux, BC repair by progression, recombination BCR-ABL1 signaling; Secondary TKI-R genes similar to BC genes but not primary TKI-R 15 genes Identified a set distinguished of genes whose CP from BC. expression was 12 genes differentially distinguished regulated between in patients secondary TKI-R resistant to vs. optimal imatinib responders. PTGS1 predicted primary TKI-R

A binary classification model based on 17 genes. HR-GES: 77% failure, but missed 2/9. LR-GES: 95% did well, but missed 4/79. 64% sensitivity; 97% specificity. HR-GES had lower rate of EMR failure with nilotinib

GSEA indicated genes associated with poorer outcome enriched for cell cycle, stem cell function, & depleted for immune function.

Microarray Microarray (Affymetrix HG-U133 (Illumina HT-12v4) Plus 2.0) TaqMan LDA 21 105 365 CCyR at 12 months CCyR at 12 months; EMR at 3 months NR (failed to achieve any cytogenetic response)

TaqMan LDA

Unselected; total WBC

CP 63; CP 15 AP 5; secondary TKI-R 29; BC 27 Diagnostic blood Diagnostic PB

Zhang et al., 2009

A 75-probe set classifier that separated the responder groups. PPV 87.7%; NPV 73.7% PPV 94.4%; NPV 75%. CD34+ cell selection & microarray analysis possible, successful in 71% of patients. Predictive genes overlapped with three independent datasets for BC genes (Zheng et al., 2006), genes prediciting early BC transformation (Yong et al., 2006 ), PRC target genes in BC (Ko et al., 2020).

CD34+ selected BM MNC & CD34+ PB MNC in validation group Microarray (Affymetrix HG-U133 Plus 2.0) 885 R=CCyR at 12 months; NR= >66% Ph+ at 12 months Predictive genes enriched for cell adhesion and targets of the Wnt/b-catenin pathway

CP 12 R; 24 NR (discovery); CP17 R 6 NR (validation) Diagnostic PB & BM

McWeeney et al., 2009

CP: chronic phase; AP: accelerated phase; BC: blast crisis; R: responder; NR: non-responder; TKI-R: resistance to tyrosine kinase inhibitors; Rx: treatment; TKI: tyrosine kinase inhibitor; IFN-: interferon-alpha; BM: bone marrow; PB: peripheral blood; PBMC: peripheral blood mononuclear cells; MNC: mononuclear cells; WBC: white blood cells; N.; number; DEG: differentially expressed genes; MCyR: major cytogenetic response; CCyR: complete cytogenetic response; Ph+: Philadelphia chromosome-positive; EMR: early molecular response; GE: gene expression; GSEA: gene set enrichment analysis; DEG: differentially expressed genes; PPV: positive predictive value; NPV: negative predictive value; HR-GES: high-risk gene expression signature; LR-GES: low-risk gene expression signature.

31 genes were used to develop a classifier to separate TKIresponders from non-responders.

Diagnostic

CP 66

McLean et al., 2004

Prior to TKI, but could have been on IFN-a Unselected; Unselected; Unselected; PB & whole blood total WBC MNC from PB & BM cDNA Microarray Microarray Microarray (Affymetrix (Affymetrix HG_U95Av2) HG_U95Av2) N. of DEG 79 55 Time of R=MCyR R=CCyR R=CCyR within predicted event (<35% Ph+); (0% Ph+); 9 months; NR= >65% NR= >65% Ph+ NR= >35% Ph+ Ph+ at 5 months at 12 months after 1 year Biological insights First evidence Predictive genes GE comparisons that GE profiles can enriched for should be made predict sensitivity cell adhesion, on purified CD34+ to imatinib mitogenic cells signaling, apoptosis

CP 18; AP 2; BC 2

Stage & numbers

Kaneta et al., 2002

Table 1. Gene expression profiling studies comparing responders and non-responders to tyrosine kinase inhibitors.

GE-based biomarkers in CML

361

V. Krishnan et al.

Single-cell-based gene expression analysis Recent advances in single-cell analysis have enabled novel GE-based insights on the roles of tumor cell heterogeneity and clonal evolution under the selective pressure of therapeutics, with obvious applications in biomarker development.30 Leukemia stem cell (LSC) heterogeneity was characterized by Warfvinge et al. by combining highthroughput immunophenotyping with single-cell GE profiling with a defined panel of genes.31 LSC sub-fractions with more primitive and quiescent signatures had a higher persistence after TKI therapy with the most TKI-insensitive population identified as Lin–CD34+CD38low/– CD45RA–cKIT–CD26+ stem cells. Giustacchini et al. used the Smart-seq2 platform to combine single-cell RNAsequencing analysis with BCR-ABL1 transcript detection using purified stem cells. A sub-population of BCR-ABL1+ quiescent stem cells enriched for hematopoietic stem cell (HSC) signatures was found to persist during TKI therapy.32 Intriguingly, the BCR-ABL1- cells in CML patients were enriched for inflammatory, tumor growth factor-b and tumor necrosis factor-a hallmarks and discriminated between the TKI-responder groups. In addition to LSC-derived signatures, the GE signature of immune cells can be equally instructive. For example, plasmacytoid dendritic cells, the major producers of interferon-a in vivo, promoted resistance to nilotinib in CML patients.33 These studies imply that the cytokines released by immune cells in the bone marrow microenvironment, and the transcriptomic changes that they bring about on the LSC, may activate cytokine-dependent TKI resistance programs.34 Together, these single-cell studies demonstrate that GE signatures within malignant and nonmalignant compartments in CML are prognostically informative. We anticipate the discovery of additional biomarkers among discrete cell types which have the potential to be assayed by platforms available in standard pathology laboratories, e.g., by flow cytometry or immunohistochemistry.

Gene expression signatures associated with blast crisis progression Transcriptomic comparisons between the CP and BC stages have uncovered progression-related signatures that can herald BC transformation (Table 2). In the pre-TKI era, the time to BC transformation from CP varied between patients, and to understand this difference, Yong et al. compared CD34+ cells from leukapheresis samples provided by patients who progressed to BC within 3 years (aggressive leukemia) or after more than 7 years (indolent leukemia) following diagnosis.13 The study identified that lower CD7 with higher PR3 and ELA2 expression at diagnosis was associated with longer survival. Intriguingly, when the GE signatures identified by Yong et al. and McWeeney et al. were compared, a significant overlap was found.11 This important study demonstrated that biological processes associated with TKI resistance and early BC transformation overlapped, and that CD34+ cells from different sources (bone marrow vs. peripheral blood) contained this information. In the post-TKI era, a landmark study by Radich et al. identified distinct transcriptional programs during BC progression.25 About 3,000 genes were associated with the BC stage with a dysregulated WNT/b-catenin pathway, decreased Jun B and FOS, and higher PRAME expression. The Radich dataset was subsequently used to compute a six-gene signature comprising NOB1, DDX47, IGSF2, LTB4R, SCARB1, and SLC25A3 to predict progression.35 Independently, Zheng et al. isolated CD34+ cells and identified 34 differentially expressed genes as cells transited from CP to BC. Among the misregulated genes, SOCS2 and CD52 were downregulated while HLA-related genes were overexpressed in BC.36 To understand the biological mechanisms underlying GE changes in TKI resistance and BC, a recent study tested the hypothesis that prognostically important genes were enriched for targets of the polycomb repressive complex (PRC; see below).10 Importantly, target genes of PRC-asso-

Table 2. Gene expression profiling studies comparing chronic phase and acute phase.

Yong et al., 2006

Radich et al., 2006

Zheng et al., 2006

Oehler et al., 2009

Stage & numbers

CP 68

CP 11; BC 9

Unselected/CD34+; PBMC/BM Platform N. of genes

CD34+; PBMC Microarray 20

CP 42; AP 17; BC 32 Unselected; BM Microarray 3000+

CD34+; PBMC Microarray 114

CP 42; AP 17; BC 34 Unselected; BM Microarray 6

Comments

Identifies early (≤3 years) vs. late (≥7 years) BC transformation. Low CD7 & high PR-3 predicts higher OS.

Identifies TKI-R in CP (had BC-like signature)

Genes that distinguish CP and BC

Ko et al., 2020

CP 16; MBC 13; LBC 5 CD34+; PBMC Microarray 431 Upregulated LBC 522 downregulated Discriminates between Identifies a core BC early & late CP gene expression signature common to MBC and LBC. PRC-driven transcriptional reprogramming is enriched for poor prognostic genes in CP in the CD34+ datasets of Yong et al. (2006) and McWeeney et al. (2009).

CP: chronic phase; AP: accelerated phase; BC: blast crisis; MBC: myeloid blast crisis; LBC: lymphoid blast crisis; PBMC: peripheral blood mononuclear cells; BM: bone marrow; PR3: proteinase-3: OS: overall survival; TKI-R: resistance to tyrosine kinase inhibitors; PRC: polycomb repressive complex.

362

haematologica | 2022; 107(2)

GE-based biomarkers in CML

ciated silencing in BC progression were enriched for downregulated genes identified in the datasets of both McWeeney et al. and Yong et al.10 The cross-validation of these three independent datasets suggests important lessons for the development of GE-based risk assessment: (i) the discovery of reproducible GE-based biomarkers is possible when homogeneous CD34+ populations are used; (ii) the processes of TKI resistance and BC transformation are biologically convergent despite genetic heterogeneity:10 and (iii). PRC-regulated processes contribute to silencing of prognostically informative genes.

Contribution of somatic mutations to gene expression signatures Recent reviews have described the range and frequency of specific genetic mutations in patients who developed TKI resistance and/or BC.37 For many of these genes, there is strong preclinical information indicating that their associated mutations contribute to or are even sufficient to produce TKI resistance or transformation phenotypes (summarized in Table 3).38-48 These studies imply that genetic mutations alter GE profiles, and here we review their contributions to GE changes in CML since these changes may represent useful GE-based biomarkers. For RUNX1 mutations, the Mustjoki group identified an accompanying GE signature in BC samples.49 They found that RUNX1 mutations were associated with the upregulation of stemness, B-cell markers, interferon and immune signaling and transcription factors regulating plasmacytoid dendritic cell development. In analogous work, the overexpression of an IKZF1 dominant-negative mutant in CD34+ cells from CP patients increased STAT5 expression, a pathway associated with imatinib resistance,50 and enhanced transformation.51 RAG expression status was recently assessed in diagnostic samples, given the role of RAG recombination as a mediator of IKZF1 deletions.52 Notably, RAG1/2 and DNTT upregulation at diagnosis suggested imminent lymphoid BC transformation within 12 months (8/8 patients), demonstrating that GE signatures can reliably predict transformation. Despite limited functional interrogation of ASXL1 using CML patient material, insertion sites within ASXL1 promoted BC progression in a CP mouse model subjected to transposition-based mutagenesis.53 Transgenic expression of truncated protein ASXL1aa1-587 in mice increased HSC self-renewal, and Brd4 occupancy and chromatin accessibility around genes required for stemness, and predisposed mice to myeloid malignancies.54 However, the clinical relevance of diagnostic ASXL1 mutations is still unclear because some patients with ASXL1 variants at diagnosis can achieve a MMR after TKI therapy.14 Furthermore, ASXL1 mutations frequently disappeared when monitored in the long-term during TKI therapy (personal observation by Dr. Dennis Kim). Meanwhile, direct evidence for contributions of other mutations to CML GE signatures is currently lacking, and we have to infer them from studies in other malignancies (Table 3).

Lessons from clonal hematopoiesis Clonal hematopoiesis is the clinical phenomenon by which populations of hematopoietic cells expand and haematologica | 2022; 107(2)

carry a somatic mutation that is at least 2% of the variant allele fraction.55 The common genes comprise DNMT3A, TET2, and ASXL1, and others also found in CML individuals, including RUNX1, BCORL1, and TP53.56 Individuals with clonal hematopoiesis are at increased risk of developing hematologic malignancies, and it is therefore pertinent to ask whether clonal hematopoiesis-related mutations also confer increased risk of TKI resistance or progression. A study by Kim et al. has highlighted important features of clonal hematopoiesis-related mutations in CML.57 Firstly, they may occur in a non-Philadelphia chromosome-positive clone and predate the development of CML, and are unrelated to the CML clone. Secondly, even when a specific mutation occurs in the Philadelphia chromosome-positive clone, it only confers a relative risk of TKI resistance or progression. Indeed, patients with RUNX1 mutations have been documented to achieve MMR (personal observation, Dr. Dennis Kim). Nevertheless, Kim et al. concluded that mutations in genes regulating epigenetic function (TET2, ASXL1 among them) were associated with a higher risk of inferior TKI responses. There are also strong preclinical data indicating that clonal hematopoiesis-related mutations result in subtle but important changes in GE in HSC. For example, Dnmt3a-deficient HSC show a loss of DNA methylation in regions enriched for self-renewal genes such as Meis1, Evi1 and HoxA9.58 In Tet2-deficient mice, the loss of DNA demethylation is accompanied by an expansion of the stem and progenitor cell compartments, and eventual myeloproliferation.45 In ASXL1-deficient mice, an increase in self-renewal capacity of stem cells is observed, through the loss of PRC1-mediated gene repression.59 Another interesting aspect of hematopoietic stem and progenitor cells harboring inactivating mutations of DNMT3A and TET2 is that they both led to increased cytokine production in peripheral myeloid cells, including interleukin-6 and interleukin-1b.60,61 Furthermore, mutations associated with clonal hematopoiesis are frequently found in monocytes, granulocytes, and natural killer cells compared to B or T cells, suggesting that their effects may also be manifest in multiple differentiated cell types within the hematopoietic compartment.62 Together, these observations are relevant to the search for prognostic GE signatures in CML for the following reasons: (i) increased inflammation and cytokine production is associated with LSC persistence,32 and disease progression;10,63 (ii) prognostic GE changes may be found in both CD34+ and CD34– fractions of peripheral blood or bone marrow mononuclear cells; and (iii) changes in natural killer cell function and number may predict treatment-free remissions, and presumably contain informative natural killer cell GE signatures.64-66

Epigenetic contributions to gene expression signatures Polycomb repressive complex-associated gene expression changes Among the most well studied epigenetic complexes in CML are the polycomb group (PcG) proteins.67 The polycomb group proteins assemble into two complexes, PRC2 and PRC1, which modify histones through repressive H3K27 trimethylation (H3K27me3) and H2AK119 363

V. Krishnan et al.

monoubiquitination (H2AK119ub1),respectively, and in general repress gene expression.68 From a mechanistic standpoint, the most commonly occurring mutations in CML appear to converge in their ability to interact with and function in conjunction with the polycomb group proteins. ASXL1 functions in transcriptional repression through its interaction with PRC2 and BAP1.69 BCORL1 is a transcriptional co-repressor that interacts with PCGF1, a core complex of the PRC1.1 complex.70 The RUNX1-CBF-b heterodimer mediates transcription by binding to RUNX sites, but also represses transcription by interacting and recruiting BMI1 of the PRC1 complex to target sites.71 IKZF1 regulates transcription by interacting with repressive epigenetic complexes such as HDAC1, HDAC2, CHD3, CHD4, and SWI/SNF complex, and also recruits PRC2 to target gene loci in T cells.72 Thus, while the commonly mutated genes in CML have their own exclusive roles in transcriptional regulation, they also share a striking commonality as modulators of the PRC. Whether mutated variants of RUNX1,

ASXL1, IKZF1, and BCORL1 drive aberrant PRC recruitment and GE in CML remains to be determined. In this respect, a recent study determined that lymphoid and myeloid BC transcriptomes are highly congruent, and that both undergo PRC-driven epigenetic reprogramming towards a convergent transcriptomic state.10 PRC-dependent epigenetic reprogramming was attributed to gain- and loss-of-function mutations in members of the PRC1 and PRC2 complexes, respectively. Of these, ongoing BMI1/PRC1 activity contributes to maintaining the BC transcriptome, while EZH2/PRC2-binding was instructional for DNA hypermethylation-dependent gene repression. Importantly, the integrative model proposed by Ko et al. suggests that enrichment for PRC-dependent GE signatures at diagnosis can predict disease transformation and TKI resistance, as highlighted above.10 We also note that dysregulated regulation of PRC has been identified as a key feature of TKI-resistant LSC in CP. EZH2 expression was higher in CML LSC than in normal HSC, and CML LSC have a stronger dependence on

Figure 2. Diagrammatic representation of the ‘seed and soil’ model of chronic myeloid leukemia. The model proposes that both acquired mutations and the cell state of the mutation-acquiring cell contribute to the process of full transformation to blast crisis (BC). A ‘strong’ mutation is defined as being able to confer selfrenewal function on a progenitor cell that does not possess inherent self-renewal capacity. A ‘weak’ mutation is unable to confer self-renewal function and can only transform a cell with native self-renewal ability, i.e., a stem cell. For both strong and weak mutations, it is likely that additional genetic and epigenetic events are necessary to confer the full suite of features required for BC transformation. The model is also based on the recent finding that BC progenitors which harbor different somatic mutations share a common or core transcriptome enriched for stemness, quiescence, and inflammatory gene expression signatures.20 HSC: hematopoietic stem cell; LSC: leukemia stem cell; MPP: multipotential progenitor; LMPP: lymphoid-primed multipotent progenitor; CMP: common myeloid progenitor; GMP: granulocyte-macrophage progenitor; TKI-S/R: tyrosine kinase inhibitor-sensitive/resistant cells. Rx: treatment; CP: chronic phase; BC: blast crisis; CML: chronic myeloid leukemia; GE: gene expression.

364

haematologica | 2022; 107(2)

Yang et al., 2018 Katoh, 2013 Balasubramani et al., 2015

Yes Joshi et al., 2014 Beer et al., 2015

Pagan et al., 2007 Wong et al., 2016

Effect of CML BCORL1 variants unknown

BCORL1 depletion increased the re-plating capacity of Runx1depleted Lin− cells

0.9/6.7

TET2

Missense, nonsense and frameshift mutations TET2 mutations may be CHIP-related or a part of the Ph+ clone Impaired 5-methylcytosine hydroxylation and decreased methylation at CpG sites in myeloid cancers with mutant TET2

DNMT3A mutations are mostly CHIP mutations since they are also present in the Ph- clone

PRC2, EVI1, ISGF3, AP2a, ZEB1, HDAC1

Transfer of methyl group to cytosines on DNA

DNA methylation

2.3/4.5

DNMT3A

Zhang et al., 2009 Branford et al., 2018

Kim et al., 2017 Pronier et al., 2011 Crusio et al., 2011

Mayle et al., 2015 Hervouet et al., 2018 Huang et al., 2018 Kim et al., 2017 Branford et al., 2018

DNMT3A-deficient HSC show loss of DNA methylation at the edge of hypomethylated canyon regions enriched for self-renewal genes such as MEIS1, EVI1, HOXA9 MDS and AML GATA2 TET2 silencing in human Nearly a third of CHIPmutants inhibit CD34+ cells increased the related DNMT3A differentiation and monocytic lineage at the mutations reduce protein apoptosis expense of erythroid and stability lymphoid lineages GATA2 deficiency has Conditional TET2 loss Dnmt3a ablation in HSC been recognized as a in the hematopoietic predisposes mice to major MDS compartment leads develop a spectrum of predisposition syndrome to increased stem cell myeloid and lymphoid in humans self-renewal malignancies

Zinc finger domain variants

Methylcytosine dioxygenase Conversion of 5-methylcytosine to 5-hydroxymethylcytosine FOG1 through N-terminal 2-HG, vitamin C, OGT, Zinc finger domain WT1, VPRBP, IDAX

Transcription factor DNA binding via zinc finger domain

0/8.4

GATA2

Transcriptional repressor CML L359V mutant of E-cadherin. inhibits Other targets unknown transactivation by PU.1. GATA2 MDS and AML mutants have altered transactivation activity

Frameshift, nonsense mutations

Transcriptional co-repressor Represses transcription by binding to class II HDAC & CTBP1 PCGF1, the core PRC1.1 component

0.9/8.6

BCORL1

HDAC: histone deacetylases; CML: chronic myeloid leukemia; CHIP: clonal hematopoiesis of indeterminate potential; MDS: myelodysplastic syndrome; AML: acute myeloid leukemia; HSC: hematopoietic stem cell; CP: chronic phase; ALL: acute lymphoblastic leukemia; MPN: myeloproliferative neoplasm.

Yes

Ikaros DNA binding domain inactivation in early pre-B cells leads to ALL

Effect of mutant protein/ RUNX1 H78Q or gene knock out in vivo V91fs-ter94 mutants induced a BC or accelerated phase-like phenotype in mice Effect on CML Yes variant studied References Zhao et al., 2012 Awad et al., 2020 Branford et al., 2018

Increased Brd4 occupancy and chromatin accessibility around genes

Truncated ASXL1 increased proliferation, and decreased differentiation along megakaryocyte and erythroid lineages AML, MPN, MDS-like diseases

Up: JAK-STAT signaling, self-renewal genes. Down: B-cell lineage and DNA repair genes

IK6 expression in CD34+ cells isolated from CP-CML patients enhances their in vitro expansion

Up: Interferon signaling, immune molecules, pDC- TF. Down: DNA repair

Effects on gene expression

Majority are frameshift and nonsense mutations in exon 12

PRC2 , BAP1 complexes

Transcriptional repression Regulator of H3K27me3 & H2AK119ub1 marks

9.7/15.1

ASXL1

Effect of mutant protein/ RUNX1 H78Q orV91 gene knock out in vitro fs-ter94 in 32D-BCR-ABL1 model blocked differentiation

RUNT domain mutations deletions and fusions

Aberration in CML

HDAC1, HDAC2, CHD3, CHD4, PRC2, CtBP1, SWI/SNF Deletionsexon D4–7 (IK6), exon D2–7

Transcription factor DNA binding via zinc finger domain.

Transcription factor DNA binding via RUNT domain.

P300, CBP, PRC1, NuRD, SWI/SNF, MLL/TrxG

6.1/16.0

IKZF1

2.6/18.3

Interactions with other complexes

Mode of action and interactions

Frequency: diagnosis/ progression (%) Function

RUNX1

Table 3. Functional effects of frequently mutated genes in blast crisis.

GE-based biomarkers in CML

365

V. Krishnan et al.

the PRC2-EZH2 axis for survival and TKI resistance.73,74 Likewise, higher BMI1 levels at diagnosis correlated with disease progression from CP to BC.12 while BMI1 overexpression in CP CD34+ cells increased proliferation and self-renewal,75 and transformed B-lymphoid progenitors in vivo.76

DNA methylation-associated gene expression changes Many studies have examined the role of DNA methylation as a regulator of aberrant GE in CML pathogenesis. In candidate-based approaches, genes involved in cell cycle regulation (P16, P53, PLCD1, PER3, HIC1), differentiation (HOXA4, DLX4, DDIT3, SPI1) proliferation (CDH13, DAPK1), apoptosis (BIM), Wnt regulation (sFRP1, CBY1), LSC maintenance (MTSS1), and cell signaling (Jun B, SOCS2) were identified as targets of DNA methylation.67,77 Recent unbiased genome-wide methylome analyses have solidified the concept of aberrant DNA methylation as a driver of resistance and transformation. The number of differentially methylated regions in CP increased from ~600 to ~6,500 CpG sites in BC.78 BC was associated with heightened DNA hypermethylation, and to a lesser extent hypomethylation, around promoters of genes involved in

stem cell fate, differentiation and leukemia-related functions.10 Mechanistically, differential DNA methylation patterns in CML have been attributed to underlying DNMT3A/TET2 mutations, PRC2-dependent epigenetic re-programming, and cytosolic sequestration of Tet2 by BCR-ABL1.79 Notably, the physiological targeting of DNA hypermethylation using 5-aza-2’-deoxycytidine ameliorated disease phenotypes in a mouse model of CP disease,80 while low-dose decitabine displayed clinical activity in patients refractory to imatinib,81 suggesting DNA methylation does indeed contribute to TKI resistance. Based on the biological insights gleaned so far, it is possible that progression-related DNA methylation signatures may already be evident at diagnosis, particularly in patients presenting with advanced CP.10 The DNA methylation status of specific target genes might therefore be useful in the timely identification of such patients for more aggressive therapies. Furthermore, given that DNA methylation is a relatively stable epigenetic and biochemical mark, there are practical advantages to developing DNA methylation-based biomarkers rather than transcript-based readouts, especially for the development of robust clinical-grade tests (Figure 3).

Figure 3. Stages of development of gene expression-based biomarkers. In chronic myeloid leukemia (CML), the development of gene expression-based biomarkers can be divided into three stages following an initial discovery phase. These stages will each determine the analytical validity, clinical validity, and clinical utility of the tests in question. Examples of CML-specific issues or questions that are pertinent to each stage are outlned in boxes under each stage. GE: gene expression; IHC: immunohistochemistry; FC: flow cytometry; RT-PCR: reverse transcriptase polymerase chain reaction; ISH: in-situ hybridization; LCM: laser capture microdissection; scRNA-seq: single-cell RNA sequencing; ATAC-seq: assay for transposase-accessible chromatin sequencing; BM: bone marrow; PBMC: peripheral blood mononuclear cell; MNC: mononuclear cells; FFPE: formalin-fixed paraffin-embedded tissues; PB: peripheral blood; NK: natural killer cells; MDSC: myeloid-derived suppressor cells; TKI: tyrosine kinase inhibitor; NCCN: National Comprehensive Cancer Network; ELN: European LeukemiaNet; BC: blast crisis; EFS: event-free survival; DFS: diseasefree survival; PFS: progression-free survival; OS: overall survival, TFR: treatment-free remission; 95% CI: 95% confidence interval; DMR: deep molecular response..

366

haematologica | 2022; 107(2)

GE-based biomarkers in CML

Gene expression profiles and mutations: ‘seed and soil’ revisited As described above, it will be important to develop CML models that integrate the interaction between genetic and epigenetic factors in driving drug resistance and disease transformation. In this respect, the effects of specific mutations may be cell-context dependent, with differential effects on GE and function depending on the cell type being examined. This is particularly the case for mutations affecting transcription factors, for which cell states, and their attendant chromatin accessibility profile, determine whether the mutated transcription factor has access to its target genes. To integrate contributions from both the above features, we propose a model in which the cell of origin, with its attendant epigenetic and transcriptional program, determines the ability of specific mutations to contribute to biological and clinical outcomes (Figure 2). This model is a derivative of the ‘seed and soil’ hypothesis of cancer initiation.82 The model will be useful for hypothesis testing, and likely explains an important feature of BCRABL1 itself. It has been shown in murine models that only when expressed in HSC, but not more committed progenitors, can BCR-ABL1 induce a myeloproliferative disorder. This is likely because BCR-ABL1 is incapable of conferring self-renewal capacity upon committed progenitors, indicating that CML cells rely on BCR-ABL1-independent mechanisms for stemness programs. These findings are in contrast to those for other leukemia fusion genes (e.g., MLL-ENL, MLL-AF9, MOZ-TIF2) which are capable of conferring self-renewal and transform progenitor cells.83 Relatedly, the model may also explain a naturally occurring phenomenon whereby normal individuals found to carry the BCR-ABL1 fusion in their peripheral blood mononuclear cells apparently never develop CML.84 Here, the model would posit that the BCR-ABL1 fusion is occurring in a long-lived progenitor without selfrenewal function. Analogous to the situation regarding cancer initiation by leukemia fusion genes, mutations devoid of selfrenewal function may only confer an increased risk of BC transformation when they occur in a target cell that already possesses physiological self-renewal function. According to this model, mutations in RUNX1 that are sufficient to induce BC-like disease in mice (Table 3) may be deemed a ‘strong’ biological seed that can transform many cell types within the hematopoietic hierarchy. Such mutations would be expected to induce disease progression in the majority of patients who harbor such mutations, which is indeed the case.37 However, a minority of CP patients with RUNX1 mutations continue to enjoy sustained deep maolecular responses,57 suggesting the existence of other important factors that modulate RUNX1 function. Along the same lines, ASXL1 was recently identified as the most frequently mutated gene at diagnosis in nine patients, the majority (n=6) of whom eventually developed BC, while a minority (n=2) achieved a MMR.14 In contrast to the above examples, the prognostic impact of ‘weak’ seeds is much less clear. In a study by Kim et al., at least four different patterns were observed for TET2 mutations.57 One pattern is seen in patients with TKI resistance when both TET2 and ABL1 variant allele frequencies increased following TKI therapy, while haematologica | 2022; 107(2)

another is seen when the TET2 variant allele frequency reduces after TKI treatment in patients with disease progression. In other cases, TET2 mutations were also detected within Philadelphia chromosome-negative cells, and here, patients showed complex outcomes following TKI therapy, with some achieving MMR and others showing TKI resistance. These observations suggest that the effect of TET2 mutations are highly contextual.

Challenges ahead but room for optimism As described above, the discovery of a limited and tractable set of genes that is prognostic across a majority of CML patients has been challenging for clinical, biological, and technical reasons. Nevertheless, there is room for optimism. In the setting of breast cancer, GE panels comprising 21 genes that encompass various aspects of breast cancer biology have been found to be predictive of therapeutic response, and minimized the use of additional therapy without compromising survival.85 Among liquid tumors, a recent study in acute myeloid leukemia demonstrated that a parsimonious 17-gene GE score, derived from a larger set of stemness-conferring genes, predicts resistance to initial therapy.86 Interestingly, this score was independent of cytogenetic and mutational risk factors, and suggests that biological factors (e.g., stemness) transcend traditional genetics-based groupings.87 Encouragingly in CML, two recent reports suggest that it is possible, using peripheral blood samples taken at diagnosis or 3 months after diagnosis, to predict deep molecular responses and also sustained treatment-free remissions. In the first study, the Adelaide group showed that the rate of decline of BCR-ABL1 transcripts during first-line TKI therapy (calculated from baseline and 3month BCR-ABL1 transcript levels) predicts success of treatment-free remission.88 The time taken for BCR-ABL1 transcripts to halve was the strongest independent predictor of sustained treatment-free remission: 80% in patients with a halving time of <9.35 days versus 4% if the halving time was >21.85 days (P<0.001). In a separate study, Radich et al. reported that GE signatures from peripheral blood taken prior to TKI initiation can distinguish individuals who will achieve a deep molecular response (MR4.5) at 5 years from those who will have suboptimal responses.89 Thus, biological information encoded in GE data can predict very long-term clinical outcomes in CML, and it is therefore conceivable that GE-based data will be able to identify patients in whom TKI therapy can be safely discontinued. More importantly, these early reports suggest that despite the likely existence of diverse resistance mechanisms within the study populations, final common paths, readout either as dynamic measures of BCR-ABL1 transcript levels, or peripheral blood GE signatures are indeed discoverable.

Stages in developing gene expression-based risk assessment The stages of developing GE-based tests has been outlined in recent reviews and consensus statements, and comprise at least three phases that assess: analytical validity (reliably measuring the genotype of interest), clinical validity (ability to segregate patients into biologi367

V. Krishnan et al.

cally and clinically important subsets), and clinical utility (ability to improve clinical decision making).90,91 In this section, we summarize the pertinent stages and highlight issues of particular relevance to GE-based biomarker development in CML (Figure 3). Stage 0 is the discovery stage, which is where the field is currently. Here, we highlight three important components, which include the use of technical approaches for unbiased discovery, the simultaneous interrogation of leukemic and non-leukemic clones from the same sample (since both have been shown to be prognostic), and the use of robust statistical and computational pipelines to discover minimal prognostic genes sets. The advent of single-cell-based technologies and their application to well-annotated cohorts will facilitate this step. In stage I, the minimal gene set has to be converted into a clinical test that accurately and reproducibly measures the GE phenotype. The test platform needs to be robust, as well as sensitive, specific and reliable. The assay should be developed for tissues that are collected as part of routine clinical care. Ideally, any additional processing of material beyond what is routine should be minimized, e.g., CD34+ selection, and should utilize standard procedures available in clinical laboratories, such as flow cytometry and bone marrow immunohistochemistry. An example would be detecting GE signatures of interest by a panel of antibodies for use in flow cytometry or immunohistochemistry applications. It is preferable that the samples used for analytical validation are from wellcharacterized patients representative of ‘real-world’ settings and, ideally, validated in at least one independent cohort. Sample size and power calculations should be determined prior to starting the study, and analytic sensitivity and specificity for the test should be available at the end of the study. At the end of stage I, a locked-down test should be evaluated in stage II, that of clinical validation. In stage II, the locked-down test will be evaluated for its ability to differentiate between clinically meaningful outcomes in modern CML practice. The samples to be tested should be obtained from well-annotated cohorts representative of the broader population, and the test conducted on tissues in a blinded manner with respect to testing and result reporting. Ideal populations include patients who have been treated uniformly in clinical trials. At the end of this stage, the ability of the test to predict clinical outcome should be available as a test score, with clearly defined positive and negative predictive values. The final stage, stage III, will be the determination of clinical utility. This stage would entail the use of the GEbased test to improve clinical decision-making, and would require the study to demonstrate that meaningful outcomes are improved when the test is used compared to when the test is not used. Besides clinical outcomes such as improved progression-free survival and overall survival, additional measures such as cost-effectiveness, avoidance of toxicities, quality of life and psychological parameters should also be assessed. Such studies may also incorporate the contribution of pharmacological factors (e.g., drug metabolism and side effects, patient com-

368

pliance) to overall outcomes. Given the relative rarity of CML, it is envisaged that this will be a multicenter international study.

Conclusion Genetic and epigenetic events contribute to the emergence of BCR-ABL1-independent clones that result in clinical TKI resistance and, if unopposed, BC transformation. Long-term TKI responses, including successful TKI stoppage, can be predicted by slower declines in BCRABL1 transcript levels during first-line TKI therapy,88 suggesting that genetic and epigenetic factors contributing to TKI resistance are present at diagnosis. Recent studies describe a convergent GE signature common to the majority of BC progenitors.10 Elements of this common or core transcriptome can be detected in CD34+ cells from CP patients at risk of TKI resistance or early transformation,11,13 and specific mutations have been shown to contribute additional nuances to the core transcriptome.49 These observations are consistent with a ‘seed and soil’ model that may be helpful for hypothesis generation (Figure 2). Emerging technologies, particularly multimodal single-cell-based approaches, will facilitate the discovery of genetic and epigenetic biomarkers at presentation. This initial discovery phase has to be followed by the translation of GE-based information into validated analytical tests, and subsequently, the determination of clinical validity and utility. This process will be a multi-year, multi-institution international effort akin to that for the development of a genetic-based risk assessment.85,90,91 The integration of both gene mutation- and gene expression-based biomarkers into the care of CML patients will be an important step to achieving the ultimate goal of CML research: the cure of the majority of our patients. Disclosures DK is a member of the advisory boards of Novartis, Pfizer, Paladin, and has received honoraria from Novartis, Pfizer and Paladin, as well as research funding from Novartis, BristolMyers Squibb, Pfizer, and Paladin. TH is a member of a Novartis advisory board and receives research support from Novartis and Bristol-Myers Squibb. SB is a member of the advisory boards of Qiagen, Novartis, and Cepheid and has received honoraria from Qiagen, Novartis, Bristol-Myers Squibb, and Cepheid, as well as research support from Novartis and Cepheid. Contributions VK and STO conceived the topic for review, and wrote the first draft of the manuscript. DDHK, TPH, and SB contributed by the addition of new sections and critical discussions throughout the writing of the review. Funding VK and STO are supported by the National Medical Research Council Singapore (MOH-CSASI18may-0002, MOH-CIRG20nov-0003, NMRC/CIRG/1468/2017).

haematologica | 2022; 107(2)

GE-based biomarkers in CML

References 1. Rowley JD. Letter: A new consistent chromosomal abnormality in chronic myelogenous leukaemia identified by quinacrine fluorescence and Giemsa staining. Nature. 1973;243(5405):290-293. 2. Perrotti D, Jamieson C, Goldman J, Skorski T. Chronic myeloid leukemia: mechanisms of blastic transformation. J Clin Invest. 2010;120(7):2254-2264. 3. Druker BJ. Translation of the Philadelphia chromosome into therapy for CML. Blood. 2008;112(13):4808-4817. 4. Holyoake TL, Vetrie D. The chronic myeloid leukemia stem cell: stemming the tide of persistence. Blood. 2017;129(12): 1595-1606. 5. Hochhaus A, Baccarani M, Silver RT, et al. European LeukemiaNet 2020 recommendations for treating chronic myeloid leukemia. Leukemia. 2020;34(4):966-984. 6. Mahon FX, Rea D, Guilhot J, et al. Discontinuation of imatinib in patients with chronic myeloid leukaemia who have maintained complete molecular remission for at least 2 years: the prospective, multicentre Stop Imatinib (STIM) trial. Lancet Oncol. 2010;11(11):1029-1035. 7. Ross DM, Hughes TP. Treatment-free remission in patients with chronic myeloid leukaemia. Nat Rev Clin Oncol. 2020;17 (8):493-503. 8. Branford S, Fletcher L, Cross NC, et al. Desirable performance characteristics for BCR-ABL measurement on an international reporting scale to allow consistent interpretation of individual patient response and comparison of response rates between clinical trials. Blood. 2008;112(8):3330-3338. 9. Malhotra H, Radich J, Garcia-Gonzalez P. Meeting the needs of CML patients in resource-poor countries. Hematology Am Soc Hematol Educ Program. 2019;2019 (1):433-442. 10. Ko TK, Javed A, Lee KL, et al. An integrative model of pathway convergence in genetically heterogeneous blast crisis chronic myeloid leukemia. Blood. 2020;135 (26):2337-2353. 11. McWeeney SK, Pemberton LC, Loriaux MM, et al. A gene expression signature of CD34+ cells to predict major cytogenetic response in chronic-phase chronic myeloid leukemia patients treated with imatinib. Blood. 2010;115(2):315-325. 12. Mohty M, Yong AS, Szydlo RM, Apperley JF, Melo JV. The polycomb group BMI1 gene is a molecular marker for predicting prognosis of chronic myeloid leukemia. Blood. 2007;110(1):380-383. 13. Yong AS, Szydlo RM, Goldman JM, Apperley JF, Melo JV. Molecular profiling of CD34+ cells identifies low expression of CD7, along with high expression of proteinase 3 or elastase, as predictors of longer survival in patients with CML. Blood. 2006;107(1):205-212. 14. Branford S, Wang P, Yeung DT, et al. Integrative genomic analysis reveals cancerassociated mutations at diagnosis of CML in patients with high-risk disease. Blood. 2018;132(9):948-961. 15. Malone ER, Oliva M, Sabatini PJB, Stockley TL, Siu LL. Molecular profiling for precision cancer therapies. Genome Med. 2020;12 (1):8. 16. Kaneta Y, Kagami Y, Katagiri T, et al. Prediction of sensitivity to STI571 among chronic myeloid leukemia patients by

haematologica | 2022; 107(2)

genome-wide cDNA microarray analysis. Jpn J Cancer Res. 2002;93(8):849-856. 17. McLean LA, Gathmann I, Capdeville R, Polymeropoulos MH, Dressman M. Pharmacogenomic analysis of cytogenetic response in chronic myeloid leukemia patients treated with imatinib. Clin Cancer Res. 2004;10(1 Pt 1):155-165. 18. de Lavallade H, Finetti P, Carbuccia N, et al. A gene expression signature of primary resistance to imatinib in chronic myeloid leukemia. Leuk Res. 2010;34(2):254-257. 19. Zhang WW, Cortes JE, Yao H, et al. Predictors of primary imatinib resistance in chronic myelogenous leukemia are distinct from those in secondary imatinib resistance. J Clin Oncol. 2009;27(22):3642-3649. 20. Kok CH, Yeung DT, Lu L, et al. Gene expression signature that predicts early molecular response failure in chronic-phase CML patients on frontline imatinib. Blood Adv. 2019;3(10):1610-1621. 21. Marin D, Ibrahim AR, Lucas C, et al. Assessment of BCR-ABL1 transcript levels at 3 months is the only requirement for predicting outcome for patients with chronic myeloid leukemia treated with tyrosine kinase inhibitors. J Clin Oncol. 2012;30(3): 232-238. 22. Hughes TP, Saglio G, Kantarjian HM, et al. Early molecular response predicts outcomes in patients with chronic myeloid leukemia in chronic phase treated with frontline nilotinib or imatinib. Blood. 2014;123(9):1353-1360. 23. Frank O, Brors B, Fabarius A, et al. Gene expression signature of primary imatinibresistant chronic myeloid leukemia patients. Leukemia. 2006;20(8):1400-1407. 24. Villuendas R, Steegmann JL, Pollan M, et al. Identification of genes involved in imatinib resistance in CML: a gene-expression profiling approach. Leukemia. 2006;20(6): 1047-1054. 25. Radich JP, Dai H, Mao M, et al. Gene expression changes associated with progression and response in chronic myeloid leukemia. Proc Natl Acad Sci U S A. 2006;103(8):2794-2799. 26. Lim S, Saw TY, Zhang M, et al. Targeting of the MNK-eIF4E axis in blast crisis chronic myeloid leukemia inhibits leukemia stem cell function. Proc Natl Acad Sci U S A. 2013;110(25):E2298-2307. 27. Crossman LC, Mori M, Hsieh YC, et al. In chronic myeloid leukemia white cells from cytogenetic responders and non-responders to imatinib have very similar gene expression signatures. Haematologica. 2005;90(4): 459-464. 28. Burguillo FJ, Martin J, Barrera I, Bardsley WG. Meta-analysis of microarray data: the case of imatinib resistance in chronic myelogenous leukemia. Comput Biol Chem. 2010;34(3):184-192. 29. Jamieson CH, Ailles LE, Dylla SJ, et al. Granulocyte-macrophage progenitors as candidate leukemic stem cells in blast-crisis CML. N Engl J Med. 2004;351(7):657-667. 30. Lim B, Lin Y, Navin N. Advancing cancer research and medicine with single-cell genomics. Cancer Cell. 2020;37(4):456-470. 31. Warfvinge R, Geironson L, Sommarin MNE, et al. Single-cell molecular analysis defines therapy response and immunophenotype of stem cell subpopulations in CML. Blood. 2017;129(17):2384-2394. 32. Giustacchini A, Thongjuea S, Barkas N, et al. Single-cell transcriptomics uncovers distinct molecular signatures of stem cells in chronic myeloid leukemia. Nat Med.

2017;23(6):692-702. 33. Inselmann S, Wang Y, Saussele S, et al. Development, function, and clinical significance of plasmacytoid dendritic cells in chronic myeloid leukemia. Cancer Res. 2018;78(21):6223-6234. 34. Sinnakannu JR, Lee KL, Cheng S, et al. SRSF1 mediates cytokine-induced impaired imatinib sensitivity in chronic myeloid leukemia. Leukemia. 2020;34(7):1787-1798. 35. Oehler VG, Yeung KY, Choi YE, et al. The derivation of diagnostic markers of chronic myeloid leukemia progression from microarray data. Blood. 2009;114(15):32923298. 36. Zheng C, Li L, Haak M, et al. Gene expression profiling of CD34+ cells identifies a molecular signature of chronic myeloid leukemia blast crisis. Leukemia. 2006;20(6): 1028-1034. 37. Branford S, Kim DDH, Apperley JF, et al. Laying the foundation for genomicallybased risk assessment in chronic myeloid leukemia. Leukemia. 2019;33(8):1835-1850. 38. Zhao LJ, Wang YY, Li G, et al. Functional features of RUNX1 mutants in acute transformation of chronic myeloid leukemia and their contribution to inducing murine fullblown leukemia. Blood. 2012;119(12):28732882. 39. Joshi I, Yoshida T, Jena N, et al. Loss of Ikaros DNA-binding function confers integrin-dependent survival on pre-B cells and progression to acute lymphoblastic leukemia. Nat Immunol. 2014;15(3):294304. 40. Balasubramani A, Larjo A, Bassein JA, et al. Cancer-associated ASXL1 mutations may act as gain-of-function mutations of the ASXL1-BAP1 complex. Nat Commun. 2015;6:7307. 41. Pagan JK, Arnold J, Hanchard KJ, et al. A novel corepressor, BCoR-L1, represses transcription through an interaction with CtBP. J Biol Chem. 2007;282(20):15248-15257. 42. Kazenwadel J, Secker GA, Liu YJ, et al. Loss-of-function germline GATA2 mutations in patients with MDS/AML or MonoMAC syndrome and primary lymphedema reveal a key role for GATA2 in the lymphatic vasculature. Blood. 2012;119(5):1283-1291. 43. Zhang SJ, Shi JY, Li JY. GATA-2 L359V mutation is exclusively associated with CML progression but not other hematological malignancies and GATA-2 P250A is a novel single nucleotide polymorphism. Leuk Res. 2009;33(8):1141-1143. 44. Pronier E, Almire C, Mokrani H, et al. Inhibition of TET2-mediated conversion of 5-methylcytosine to 5-hydroxymethylcytosine disturbs erythroid and granulomonocytic differentiation of human hematopoietic progenitors. Blood. 2011; 118(9):2551-2555. 45. Moran-Crusio K, Reavie L, Shih A, et al. Tet2 loss leads to increased hematopoietic stem cell self-renewal and myeloid transformation. Cancer Cell. 2011;20(1):11-24. 46. Mayle A, Yang L, Rodriguez B, et al. Dnmt3a loss predisposes murine hematopoietic stem cells to malignant transformation. Blood. 2015;125(4):629638. 47. Hervouet E, Peixoto P, Delage-Mourroux R, Boyer-Guittaut M, Cartron PF. Specific or not specific recruitment of DNMTs for DNA methylation, an epigenetic dilemma. Clin Epigenetics. 2018;10:17. 48. Huang Y-H, Tovy A, Sundaramurthy V, et al. Nearly a third of clonal hematopoiesis-

369

V. Krishnan et al. associated DNMT3A mutations reduce protein stability and may be associated with poorer prognosis. Blood. 2018;132 (Suppl 1):1315. 49. Adnan Awad S, Dufva O, Ianevski A, et al. RUNX1 mutations in blast-phase chronic myeloid leukemia associate with distinct phenotypes, transcriptional profiles, and drug responses. Leukemia. 2021;35(4): 1087-1099. 50. Warsch W, Kollmann K, Eckelhart E, et al. High STAT5 levels mediate imatinib resistance and indicate disease progression in chronic myeloid leukemia. Blood. 2011;117 (12):3409-3420. 51. Beer PA, Knapp DJ, Miller PH, et al. Disruption of IKAROS activity in primitive chronic-phase CML cells mimics myeloid disease progression. Blood. 2015;125(3): 504-515. 52. Thomson DW, Shahrin NH, Wang PPS, et al. Aberrant RAG-mediated recombination contributes to multiple structural rearrangements in lymphoid blast crisis of chronic myeloid leukemia. Leukemia. 2020;34(8): 2051-2063. 53. Giotopoulos G, van der Weyden L, Osaki H, et al. A novel mouse model identifies cooperating mutations and therapeutic targets critical for chronic myeloid leukemia progression. J Exp Med. 2015;212(10):15511569. 54. Yang H, Kurtenbach S, Guo Y, et al. Gain of function of ASXL1 truncating protein in the pathogenesis of myeloid malignancies. Blood. 2018;131(3):328-341. 55. Jaiswal S, Ebert BL. Clonal hematopoiesis in human aging and disease. Science. 2019; 366(6465):eaan4673. 56. Schmidt M, Rinke J, Schafer V, et al. Molecular-defined clonal evolution in patients with chronic myeloid leukemia independent of the BCR-ABL status. Leukemia. 2014;28(12):2292-2299. 57. Kim T, Tyndel MS, Kim HJ, et al. Spectrum of somatic mutation dynamics in chronic myeloid leukemia following tyrosine kinase inhibitor therapy. Blood. 2017;129 (1):38-47. 58. Jeong M, Sun D, Luo M, et al. Large conserved domains of low DNA methylation maintained by Dnmt3a. Nat Genet. 2014;46(1):17-23. 59. Abdel-Wahab O, Gao J, Adli M, et al. Deletion of Asxl1 results in myelodysplasia and severe developmental defects in vivo. J Exp Med. 2013;210(12):2641-2659. 60. Sano S, Oshima K, Wang Y, et al. CRISPRmediated gene editing to assess the roles of Tet2 and Dnmt3a in clonal hematopoiesis and cardiovascular disease. Circ Res. 2018;123(3):335-341. 61. Sano S, Oshima K, Wang Y, et al. Tet2mediated clonal hematopoiesis accelerates heart failure through a mechanism involving the IL-1beta/NLRP3 inflammasome. J Am Coll Cardiol. 2018;71(8):875-886. 62. Arends CM, Galan-Sousa J, Hoyer K, et al. Hematopoietic lineage distribution and

370

evolutionary dynamics of clonal hematopoiesis. Leukemia. 2018;32(9):19081919. 63. Welner RS, Amabile G, Bararia D, et al. Treatment of chronic myelogenous leukemia by blocking cytokine alterations found in normal stem and progenitor cells. Cancer Cell. 2015;27(5):671-681. 64. Hughes A, Yong ASM. Immune effector recovery in chronic myeloid leukemia and treatment-free remission. Front Immunol. 2017;8:469. 65. Hughes A, Clarson J, Tang C, et al. CML patients with deep molecular responses to TKI have restored immune effectors and decreased PD-1 and immune suppressors. Blood. 2017;129(9):1166-1176. 66. Yong AS, Keyvanfar K, Hensel N, et al. Primitive quiescent CD34+ cells in chronic myeloid leukemia are targeted by in vitro expanded natural killer cells, which are functionally enhanced by bortezomib. Blood. 2009;113(4):875-882. 67. Koschmieder S, Vetrie D. Epigenetic dysregulation in chronic myeloid leukaemia: a myriad of mechanisms and therapeutic options. Semin Cancer Biol. 2018;51:180197. 68. Sparmann A, van Lohuizen M. Polycomb silencers control cell fate, development and cancer. Nat Rev Cancer. 2006;6(11):846-856. 69. Katoh M. Functional and cancer genomics of ASXL family members. Br J Cancer. 2013;109(2):299-306. 70. Wong SJ, Gearhart MD, Taylor AB, et al. KDM2B recruitment of the polycomb group complex, PRC1.1, requires cooperation between PCGF1 and BCORL1. Structure. 2016;24(10):1795-1801. 71. Yu M, Mazor T, Huang H, et al. Direct recruitment of polycomb repressive complex 1 to chromatin by core binding transcription factors. Mol Cell. 2012;45(3):330343. 72. Oravecz A, Apostolov A, Polak K, et al. Ikaros mediates gene silencing in T cells through Polycomb repressive complex 2. Nat Commun. 2015;6:8823. 73. Xie H, Peng C, Huang J, et al. Chronic myelogenous leukemia-initiating cells require polycomb group protein EZH2. Cancer Discov. 2016;6(11):1237-1247. 74. Scott MT, Korfi K, Saffrey P, et al. Epigenetic reprogramming sensitizes CML stem cells to combined EZH2 and tyrosine kinase inhibition. Cancer Discov. 2016;6(11):1248-1257. 75. Rizo A, Horton SJ, Olthof S, et al. BMI1 collaborates with BCR-ABL in leukemic transformation of human CD34+ cells. Blood. 2010;116(22):4621-4630. 76. Sengupta A, Ficker AM, Dunn SK, Madhu M, Cancelas JA. Bmi1 reprograms CML Blymphoid progenitors to become B-ALLinitiating cells. Blood. 2012;119(2):494-502. 77. Behzad MM, Shahrabi S, Jaseb K, et al. Aberrant DNA methylation in chronic myeloid leukemia: cell fate control, prognosis, and therapeutic response. Biochem

Genet. 2018;56(3):149-175. 78. Heller G, Topakian T, Altenberger C, et al. Next-generation sequencing identifies major DNA methylation changes during progression of Ph+ chronic myeloid leukemia. Leukemia. 2016;30(9):1861-1868. 79. Mancini M, Veljkovic N, Leo E, et al. Cytoplasmatic compartmentalization by Bcr-Abl promotes TET2 loss-of-function in chronic myeloid leukemia. J Cell Biochem. 2012;113(8):2765-2774. 80. Amabile G, Di Ruscio A, Muller F, et al. Dissecting the role of aberrant DNA methylation in human leukaemia. Nat Commun. 2015;6:7091. 81. Issa JP, Gharibyan V, Cortes J, et al. Phase II study of low-dose decitabine in patients with chronic myelogenous leukemia resistant to imatinib mesylate. J Clin Oncol. 2005;23(17):3948-3956. 82. Visvader JE. Cells of origin in cancer. Nature. 2011;469(7330):314-322. 83. Huntly BJ, Shigematsu H, Deguchi K, et al. MOZ-TIF2, but not BCR-ABL, confers properties of leukemic stem cells to committed murine hematopoietic progenitors. Cancer Cell. 2004;6(6):587-596. 84. Bose S, Deininger M, Gora-Tybor J, Goldman JM, Melo JV. The presence of typical and atypical BCR-ABL fusion genes in leukocytes of normal individuals: biologic significance and implications for the assessment of minimal residual disease. Blood. 1998;92(9):3362-3367. 85. Sparano JA, Gray RJ, Makower DF, et al. Adjuvant chemotherapy guided by a 21gene expression assay in breast cancer. N Engl J Med. 2018;379(2):111-121. 86. Ng SW, Mitchell A, Kennedy JA, et al. A 17gene stemness score for rapid determination of risk in acute leukaemia. Nature. 2016;540(7633):433-437. 87. Bill M, Nicolet D, Kohlschmidt J, et al. Mutations associated with a 17-gene leukemia stem cell score and the score's prognostic relevance in the context of the European LeukemiaNet classification of acute myeloid leukemia. Haematologica. 2020;105(3):721-729. 88. Shanmuganathan N, Pagani IS, Ross DM, et al. Early BCR-ABL1 kinetics are predictive of subsequent achievement of treatmentfree remission in chronic myeloid leukemia. Blood. 2021;137(9):1196-1207. 89. Radich JP, Larson R, Kantarjian H, et al. Gene expression signature predicts deep molecular response (DMR) in chronic myeloid leukemia (CML): an exploratory biomarker analysis from ENESTnd [Abstract]. Blood. 2019;34;(Suppl_1):665. 90. Kwa M, Makris A, Esteva FJ. Clinical utility of gene-expression signatures in early stage breast cancer. Nat Rev Clin Oncol. 2017;14(10):595-610. 91. Teutsch SM, Bradley LA, Palomaki GE, et al. The Evaluation of Genomic Applications in Practice and Prevention (EGAPP) Initiative: methods of the EGAPP Working Group. Genet Med. 2009;11(1):3-14.

haematologica | 2022; 107(2)

ARTICLE

Acute Lymphoblastic Leukemia

Comprehensive analysis of dose intensity of acute lymphoblastic leukemia chemotherapy

Ferrata Storti Foundation

Seth E. Karol,1,2 Deqing Pei,3 Colton A. Smith,1 Yiwei Liu,1 Wenjian Yang,1 Nancy M. Kornegay,1 John C. Panetta,1 Kristine R. Crews,1 Cheng Cheng,3 Emily R. Finch,1 Hiroto Inaba,2 Monika L. Metzger,2,4 Jeffrey E. Rubnitz,2 Raul C. Ribeiro,2 Tanja A. Gruber,5,6 Jun J. Yang,1 William E. Evans,1 Sima Jeha,2,4 Ching-Hon Pui2 and Mary V. Relling1 Department of Pharmaceutical Sciences, St. Jude Children’s Research Hospital, Memphis, TN; 2Department of Oncology, St. Jude Children’s Research Hospital, Memphis, TN; 3Department of Biostatistics, St. Jude Children’s Research Hospital, Memphis, TN; 4Department of Global Pediatric Medicine, St. Jude Children’s Research Hospital, Memphis, TN; 5Department of Pediatrics, Stanford University School of Medicine, Stanford, CA, and 6Stanford Cancer Institute, Stanford University School of Medicine, Stanford, CA, USA

Haematologica 2022 Volume 107(2):371-380

ABSTRACT

hemotherapy dosages are often compromised, but most reports lack data on dosages that are actually delivered. In two consecutive acute lymphoblastic leukemia trials that differed in their asparaginase formulation, native E. coli L-asparaginase in St. Jude Total 15 (T15, n=365) and pegaspargase in Total 16 (T16, n=524), we tallied the dose intensities for all drugs on the low-risk or standard-risk arms, analyzing 504,039 dosing records. The median dose intensity for each drug ranged from 61-100%. Dose intensities for several drugs were more than 10% higher on T15 than on T16: cyclophosphamide (P<0.0001 for the standard-risk arm), cytarabine (P<0.0001 for the standard-risk arm), and mercaptopurine (P<0.0001 for the low-risk arm and P<0.0001 for the standardrisk arm). We attributed the lower dosages on T16 to the higher asparaginase dosages on T16 than on T15 (P<0.0001 for both the low-risk and standard-risk arms), with higher dose-intensity for mercaptopurine in those with anti-asparaginase antibodies than in those without (P=5.62x103 for T15 standard risk and P=1.43x10-4 for T16 standard risk). Neutrophil count did not differ between protocols for low-risk patients (P=0.18) and was actually lower for standard-risk patients on T16 than on T15 (P<0.0001) despite lower dosages of most drugs on T16. Patients with low asparaginase dose intensity had higher methotrexate dose intensity with no impact on prognosis. The only dose intensity measure predicting a higher risk of relapse on both studies was higher mercaptopurine dose intensity, but this did not reach statistical significance (P=0.03 T15; P=0.07 T16). In these intensive multiagent trials, higher dosages of asparaginase compromised the dosing of other drugs for acute lymphoblastic leukemia, particularly mercaptopurine, but lower chemotherapy dose intensity was not associated with relapse.

Correspondence: MARY V. RELLING mary.relling@stjude.org Received: January 21, 2021. Accepted: April 28, 2021. Pre-published: July 1, 2021. https://doi.org/10.3324/haematol.2021.278411

Introduction Dosages of chemotherapy drugs used in treatment regimens for acute lymphoblastic leukemia (ALL) vary widely.1-8 There is, however, a lack of data comparing administered dosages to planned dosages, and the last comprehensive analysis (for only 209 patients) was published in 1991.9 Hence, it is difficult to compare feasibility of protocol delivery among cooperative treatment groups, between adults and children, and across different countries. Most ALL drugs can cause myelosuppression, and thus comparisons of administered dose intensity across protocols that are limited to a single agent, e.g., mercaptopurine or asparaginase, may be misleading if the dosages of other possibly “compensating” agents are not also accounted for across protocols. As newer immune-based and less myelosuppressive agents are added to

haematologica | 2022; 107(2)

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.

371

S.E. Karol et al.

existing ALL regimens,10 predicting the tolerability of new combination protocols has proven challenging without accurate information on how current conventional ALL therapy is actually administered. Our aim was to tabulate the actual dosages and dose intensities of conventional chemotherapy in two consecutive multiagent front-line pediatric St. Jude ALL trials that used highly similar backbones but different formulations of asparaginase: E. coli L-asparaginase in Total 15 (T15) and pegaspargase in Total 16 (T16).

Methods Patients Patients were enrolled on St. Jude Children’s Research Hospital protocols T15 (ClinicalTrials.gov ID: NCT00137111)11 and T16 (ClinicalTrials.gov ID: NCT00549848)12 for newly diagnosed ALL (Online Supplementary Figure S1). T15 and T16 therapy included remission induction therapy followed by consolidation therapy and 120 weeks of continuation therapy (146 weeks for boys on T15) which included two phases of reinduction (Online Supplementary Table S1). After remission induction, patients were classified for risk-adapted therapy as low-risk (LR, about 40% of patients), standard-risk (SR, about 50% of patients) or high-risk (HR, about 10% of patients).11,12 Patients with Down syndrome received altered methotrexate and leucovorin regimens, and HR patients received intensification phases, and thus both groups were excluded from this analysis. This

report focuses on the ~90% of patients who were treated on either the LR or SR arms. Ninety patients on T15 who received most of their therapy at a collaborating institution11 were also excluded, because their drug administration data were difficult to verify (Online Supplementary Figure S2). The studies were approved by the Institutional Review Board. Informed consent was obtained from either the parents or the patients, consistent with the Declaration of Helsinki.

Treatment Drug administration data were recorded prospectively on protocol-specific forms, generally on a daily basis for induction and reinduction and on a weekly basis for other phases, by clinical staff and research nurses, and entered into centralized St. Jude databases by protocol-specific research data managers. Reasons for dose modifications were protocol-specified (Online Supplement). Treatment regimens have been described previously11,12 and are summarized in Online Supplementary Figure S1 and Online Supplementary Table S1. The two protocols differed primarily by first-line asparaginase formulation, although there were a few other differences (Online Supplementary Figure S1, Online Supplementary Tables S1 and S2).11 Patients received E. coli asparaginase (Elspar) in T15, and pegaspargase (Oncaspar) in T16. In the case of allergic reaction to either E. coli asparaginase or pegaspargase, an asparaginase formulation change was permitted by the protocol (Online Supplement). To compare planned dosages roughly between protocols, Elspar doses were converted to comparable doses of pegaspargase based on protocol-spec-

Figure 1. Planned cumulative doses of drugs used in the low- and standard-risk arms of T15 and T16. The only drugs with at least a 10% difference in planned cumulative dosages between T16 and T15 (within risk arms) were asparaginase (34% higher in the standard risk arm), dexamethasone (17.4% lower in the standard-risk arm), and doxorubicin (50% less in the low-risk arm). Thus, only asparaginase was planned for higher dosages on T16 than on T15. There were no differences between T15 and T16 in planned cumulative dosages of prednisone, vincristine, daunorubicin, cytarabine, cyclophosphamide, mercaptopurine; there was a planned decrease of only 6.0% in low-dose methotrexate dosages in the standard-risk arm on T16 versus T15. See Online Supplementary Table S2 for details on planned dosages.

372

haematologica | 2022; 107(2)

Dose intensity in ALL treatment

ified conversions (i.e., pegaspargase 2,500 units/m2 were considered comparable to 50,000 units/m2 of Elspar and 100,000 units/m2 of Erwinia asparaginase). Total planned dosages were identical for T15 LR and T16 LR as well as for T15 SR and T16 SR for consolidation high-dose methotrexate, cyclophosphamide, cytarabine, daunorubicin, continuation mercaptopurine, and vincristine; they were higher on T16 than on T15 for asparaginase, and were lower on T16 than on T15 for dexamethasone (both LR and SR arms) and for doxorubicin (LR arms; Figure 1, Online Supplementary Table S2). In Total therapy protocols, particularly for the 120 weeks of continuation, dosages of therapy were adjusted or omitted if they caused toxicity (particularly myelosuppression), as detailed (Online Supplementary Table S2), but therapy was not typically delayed. Dose intensities were calculated as the delivered dosage divided by the protocol-specified dosage for each risk group and protocol. Dose intensity was tabulated per phase, and cumulative dose intensity was estimated as the total prescribed dosage per patient divided by the total cumulative protocol-specified dosage (Online Supplement). Absolute neutrophil count (ANC) was generally checked at least weekly throughout treatment (Online Supplement). Time to complete therapy was the time from start of first chemotherapy in induction to last dose of chemotherapy at 120 weeks.

Additional methods Details of genetic ancestry assessment, TPMT phenotype assignment, and statistical methods are available in the Online Supplement.

Results Differences in dosage intensity and delivered dosages between the Total 15 and Total 16 trials The median cumulative dose intensity for each drug ranged from 61% to 100%, with the largest interpatient variability observed for mercaptopurine and methotrexate. Post-induction, dose intensity median absolute deviations ranged from 16% to 23% for mercaptopurine (Online Supplementary Table S3). The cumulative dose intensities for several drugs were higher on T15 than on T16 (Figure 2, Online Supplementary Table S3), and median cumulative dose intensities were over 10% higher on T15 than on T16 for cyclophosphamide (P<0.0001 for SR patients), cytarabine (P<0.0001 for SR patients), mercaptopurine (P<0.0001 for LR and SR patients), and consolidation methotrexate (P<0.0001 for LR patients). Examining phases instead of cumulative dose intensity, mercaptopurine exhibited the lowest dose intensities of any drug in all protocols and arms, and its dose intensity was higher on T15 than on T16 for all phases (P<0.0001 for all) except induction (Online Supplementary Table S4). The dose intensity of cyclophosphamide was higher on T15 than on T16 during the continuation phase (P<0.0001), while that of cytarabine was higher at reinduction II and during continuation (P<0.0001 for both phases; Online Supplementary Table S5 and S6). Tolerated dosages of mercaptopurine were lower on T16 than they had been on T15, with as much as 27% higher dose intensity on T15 than on T16 (Figure 3, Online

Figure 2. Administered cumulative dose intensities for drugs on T15 versus T16, by risk arm. The only significant differences (***P<0.0001) in dose intensities with more than a 10% difference between protocols were for cyclophosphamide and cytarabine (standard-risk arms) and for mercaptopurine (both risk arms), all of which were higher on T15 than on T16 (see Online Supplementary Table S3 for details on all drugs). Bars and whiskers indicate medians and median absolute deviations among each patient population. There was a total of 16 statistical comparisons, thus the Bonferroni significance threshold=0.003.

haematologica | 2022; 107(2)

373

S.E. Karol et al.

Figure 3. Thiopurine dose intensity by phase in T15 versus T16. The box and whisker plots show the quartiles and nonoutlier ranges. Dose intensity was significantly lower on T16 than on T15 (P<0.0001) for all phases except induction (see Online Supplementary Table S4 for exact P values). The largest difference was 27% during continuation weeks 1016 in standard-risk patients. There was a total of 10 statistical comparisons, thus the Bonferroni significance threshold=0.005.

Supplementary Table S4). The median cumulative dosage of mercaptopurine delivered was 7,420 mg/m2 higher on the LR arm and 9,130 mg/m2 higher on the SR arm of T15 compared to T16 (Table 1). Despite these markedly higher dosages on T15 (Table 1), ANC did not differ between the protocols for the LR arm (P=0.18) (Figure 4), and were actually lower on T16 than on T15 for the SR arm (P<0.0001), indicating that our clinicians were not titrating to a higher ANC on T16 than they had on T15. Interestingly, the only drug for which dose intensity was significantly higher on T16 than on T15 was dexamethasone (P<0.0001) (Online Supplementary Table S3), although the administered cumula374

tive dosage was higher on T15 than on T16, reflecting the fact that the planned dosage (i.e., the denominator) was lower on T16 than on T15. For most drugs on each of the LR and SR arms, the planned dosages were identical on T15 and T16 (Figure 1, Online Supplementary Table S2); the drug with the largest planned differences in dosages between the two protocols was asparaginase, with an 8% planned increase in the LR arms and a 34% increase in the SR arms on T16 compared to T15. In actuality, asparaginase was the only drug for which the planned or administered dosages were higher on T16 than on T15, with higher cumulative dosages given on haematologica | 2022; 107(2)

Dose intensity in ALL treatment

Table 1. Actual cumulative dosages administered for drugs on T15 and T16 by risk group.

Drug (units) Asparaginase (U/m2) Cyclophosphamide (mg/m2) Cytarabine (mg/m2) Daunorubicin (mg/m2) Doxorubicin (mg/m2) Vincristine (mg/m2) Thiopurine.Induction (mg/m2) ConsolHDMTX (mg/m2) Dexamethasone (mg/m2) MP Consol_to_Wk120 (mg/m2) MTX Cont_to_Wk120 (mg/m2)

Risk arm

LR SR LR SR LR SR LR SR LR SR LR SR LR SR LR SR LR SR LR SR LR SR

192 173 190 172 190 172 192 173 189 171 192 173 190 172 190 171 190 171 191 171 190 165

Median

T15 5th-%tile

95th-%tile

CV%

Median

T16 5th-%tile 95th-%tile

CV% P Value

12000 26600 1000 4440 599 11300 50 50.3 60 179 60.8 52.2 817 813 11000 18500 1060 1330 51400 37100 3250 2440

8500 15700 917 1190 227 3550 25 38.4 57.7 120 38.9 19.8 260 354 7410 11100 586 287 29600 9340 2210 1280

15500 40900 1040 4650 636 12300 51.9 52.3 60.9 183 66.1 67.2 936 918 14800 23400 1160 1580 69000 48600 4600 3440

6% 7% <1% 6% 2% 13% 2% 2% 1% 2% 8% 34% 9% 10% 19% 12% 9% 22% 21% 23% 15% 17%

254 270 252 269 253 269 254 270 250 265 253 269 253 269 250 266 251 265 251 266 251 261

13100 43300 1000 3830 600 9460 49.4 49.5 30 162 62.1 55.3 813 796 9970 18400 1020 1260 43900 28000 2980 2090

10600 12700 969 2080 295 5510 24.7 25 29.2 118 38.6 26 354 376 8990 13000 481 365 22700 8000 1840 854

17200 58700 1040 4670 638 10400 50.9 50.8 30.9 182 66.8 67.9 915 886 10300 20900 1110 1400 61400 42200 3960 3070

3% 27% <1% 25% 4% 10% 2% 2% 1% 16% 7% 29% 11% 12% 2% 10% 8% 13% 26% 40% 17% 24%

<0.0001 <0.0001 1 <0.0001 1 <0.0001 1 2.07 x10-4 <0.0001 0.22 0.33 0.33 1 0.52 <0.0001 0.92 6.18 x10-4 1.29 x10-4 <0.0001 <0.0001 <0.0001 <0.0001

T15 minus T16 Median n -1100 -16700 0 613 -1.03 1830 0.58 0.85 30 17.7 -1.3 -3.1 4.1 16.5 1020 94.7 37 74.2 7420 9130 273 347

Drugs excluded were prednisone from the induction phase (because of variability in formulations and documentation) and mercaptopurine and methotrexate for boys in T15 from week 120146 (which were not present in T16). The CV% is calculated as median absolute deviation divided by median dosage. Total of 11(drugs)x2(risk arms)=22 comparisons. Bonferroni significance threshold=0.002. T15: Total therapy 15 trial, T16: Total therapy 16 trial; Consol: consolidation; cont: continuation; HDMTX: high-dose methotrexate; CV%: percentage coefficient of variation; MP: mercaptopurine; MTX: methotrexate; wk: week.

T16 than on T15 for both the LR and SR arms (P<0.0001 for both) (Table 1). The higher exposure was due to the planned higher dosages on T16 than on T15, rather than to a change in prescribing practices, in that the dose intensities for asparaginase did not differ on T16 versus T15 for either the LR (P=1) or the SR (P=0.77) arms (Online Supplementary Table S3). The dose intensity for mercaptopurine was higher in those who had antibodies against asparaginase (against Elspar for T15 and against pegaspargase for T16) than in those who did not (P=0.12 for T15 LR, P=0.0056 for T15 SR, P=0.00027 for T16 LR, and P=0.00014 for T16 SR) (Figure 5). The higher dose intensity was likely in response to a higher ANC in those with antibodies compared to those without antibodies, with the difference only reaching nominal statistical significance for T15 SR (P=0.032) (Figure 5).

Differences in dose intensity by inherited genetics and other characteristics of the patients (age, sex, race) The only dosage modification based on a pharmacogenetic characteristic for both protocols was that for thiopurines based on TPMT status (Online Supplementary Table S2), and this modification was made prospectively. Thus, as expected, the median dose intensity was lower for those with a TPMT abnormality (poor or intermediate metabolizers) than for those without a defective TPMT allele (normal metabolizers): 0.73 versus 0.83 for T15 SR, 0.68 versus 0.84 for T15 LR, 0.48 versus 0.63 for T16 SR, and 0.60 versus 0.75 for T16 LR (P=0.001, P=4.3x10-4, P<0.0001, and P=1.92x104 , respectively) (Online Supplementary Figure S3, Online haematologica | 2022; 107(2)

Supplementary Table S7). Importantly, using this prospective precision medicine approach of genetically-driven prescribing prevented excessive thiopurine-induced cytopenias. As a result, neither the dose intensities of other drugs (anthracyclines, asparaginase, methotrexate, cyclophosphamide, cytarabine, and dexamethasone) (Online Supplementary Table S7) nor the ANC (Figure 6) differed (or differed only marginally) by TPMT status, demonstrating that pinpointing the correct drug for dosage adjustments prevented compromising the dosages of other chemotherapeutic agents. When mercaptopurine dose intensity was re-estimated using a TPMT-specific denominator for expected mercaptopurine dosages (see the Online Supplement for details), the dose intensity for mercaptopurine showed a much smaller difference between those with and those without a defect in TPMT (Online Supplementary Figure S4) than when the denominator was not adjusted downward for the expected decrease in dosage due to TPMT status (Online Supplementary Figure S5). We compared dose intensity for all drugs by sex, ancestral group (white, Black, Hispanic, and other), and age, adjusting for protocol and risk arm. Differences in dose intensity by race and by sex were relatively modest (data not shown), particularly compared to the differences by protocol or risk group. There were a few agents for which dose intensity differed by age (Online Supplementary Figure S6). For T15 SR, T16 SR, and T16 LR, the dose intensities for dexamethasone and vincristine were significantly inversely correlated to age and the dose intensity for methotrexate was positively correlated to age; for T15 LR, only the dose intensity of vincristine was inversely correlated with age, 375

S.E. Karol et al.

Figure 4. Absolute neutrophil count by risk group on T15 versus T16. The graphs show the average of fitted absolute neutrophil count (ANC; cells/mm3) data per phase per patient with thick solid lines representing the median per risk group. Based on 46,310 and 64,549 ANC records for T15 and T16, respectively.

but the effect did not reach statistical significance after correction for multiple testing.

Differences in dose intensity between risk arms The most striking difference in dose intensity between risk arms was for mercaptopurine (Figure 2). The median dose intensity was similar between risk groups for most phases on T15 (Online Supplementary Table S8) including the phase immediately after Elspar asparaginase ended (weeks 20-47, P=0.1). In contrast, on T16, in the weeks immediately following reinduction and the completion of pegaspargase (which ended week 29), the median dose intensity was 41% versus 69% for the SR versus the LR arms (P<0.0001) (Online Supplementary Table S9). The cumulative dose intensities for all the drugs by protocol and risk arm are shown in Online Supplementary Tables S10 and S11.

Relapse We explored relationships between dose intensities for all drugs during all phases and cumulatively, and ANC for all phases and cumulatively, and treatment outcomes. No significant associations between dose intensity or ANC and outcomes were observed after adjusting for multiple testing. Only one dose intensity or ANC variable was nominally associated with outcome in the same direction for both protocols in both continuous and tertile analyses: for those who finished 120 weeks of therapy: a higher dose intensity for mercaptopurine was nominally associated with a higher risk of any relapse (unadjusted P=0.03 for T15, P=0.07 for T16, tertile analysis) (Online Supplementary Table S12, Online Supplementary Figure S7). For T15, this tendency for an association between higher mercaptopurine dose intensity and 376

worse outcomes was also true when including patients who did not complete 120 weeks of therapy (Online Supplementary Table S12). Associations between ANC and outcomes were not reproducible across phases or protocols. Notably, there was no association between dose intensities of any other medications, including asparaginase (Online Supplementary Figure S8) and outcomes; however, it should be noted that there was also little interpatient variability in the dose intensity for asparaginase (median absolute deviations, 1% to 7%; Online Supplementary Table S3).

Absolute neutrophil count versus dose intensity For each drug, protocol, and risk arm, we analyzed whether dose intensity was related to ANC for each phase. The strongest associations were for mercaptopurine; in all instances with nominal associations, ANC and dose intensity were positively correlated (Online Supplementary Table S13). This reflects that clinicians followed protocol recommendations (Online Supplementary Table S1) to increase the mercaptopurine dosage in those with high ANC, and to decrease the dosage for those with low ANC. For all associations between dose intensity of other drugs and ANC, the correlations were also positive, with a few exceptions; dexamethasone dose intensity was inversely correlated with ANC in some phases (although not statistically significantly after corrections for multiple testing; data not shown).

Dose intensity versus time on therapy Although the practice was to avoid delays in therapy, there was observed variability in the time required to complete all therapy up to week 120 of continuation (median time to complete therapy including induction, consolidahaematologica | 2022; 107(2)

Dose intensity in ALL treatment

tion, and continuation was 135.5 weeks for both T15 and T16; interquartile range, 134.4-137.3 weeks). To assess the possible impact of delivered dose intensity on time to complete therapy, we analyzed the association between the dose intensity of asparaginase and mercaptopurine and time to complete therapy in patients completing treatment. We found an inverse association between dose intensity and time to complete therapy (P<0.0001 for both drugs), however, the correlation was relatively weak (r2 = 0.03 for both asparaginase and mercaptopurine). More importantly, there was no association between time to complete therapy and relapse risk (P=0.7 on T15, P=0.4 on T16). Of the 889 patients included on the study, 17 (1.9%) discontinued treatment early due to toxicity. With a median follow-up of 7.5 years from diagnosis for these patients, only one of 17 patients (5.9%) experienced a relapse (at 3.9 years after diagnosis).

Discussion This study is the first to comprehensively evaluate dosages of all conventional drugs constituting modern ALL therapy. Our most striking observation was that mercaptopurine dose intensity was lower on T16 than on T15, despite no planned changes to mercaptopurine dosages on T16 versus T15. The most likely explanation for this decreased ability to administer full dosages of mercaptopurine is higher planned and administered doses of asparaginase on T16 than on T15. We suggest that asparaginase interfered with mercaptopurine delivery based on several findings: an inverse association between delivered asparaginase and mercaptopurine dosages between protocols (Figure 2, Table 1, Online Supplementary Table S4), a higher delivered mercaptopurine dosage in those with antibodies to asparaginase than in those without antibodies (Figure 5), and a temporal decrease in dose intensity of mercaptopurine that corresponds with the timing of asparaginase use (Online Supplementary Tables S8 and S9, Online Supplementary Figures S4 and S5). Moreover, we and others have shown that asparaginase can decrease the clearance of other drugs, such as dexamethasone, putatively through its hypoproteinemic effects on hepatic drug metabolizing enzymes and transporters. We found that this interaction is associated with an increased risk of at least one adverse effect of dexamethasone (osteonecrosis) and higher drug exposure both in the clinic13 and in preclinical models.14 Others have also hypothesized that asparaginase can influence the dose intensity or toxicity of thiopurines and/or methotrexate.15,16 Asparaginase is an important part of ALL therapy, which is the reason that we chose to increase exposure to asparaginase on T16 compared to that on T15. Early discontinuation of asparaginase was associated with lower event-free survival in the Dana-Farber 91-01 study1 and in ETV6/RUNX1 ALL,17 and patients treated with E. coli asparaginase had fewer relapses than those treated with Erwinase.18 Patients with allergy to pegaspargase treated on recent Children’s Oncology Group protocols who did not receive all asparaginase therapy had a lower disease-free survival, and drug shortages of Erwinase contributed to not being able to compensate fully for missed pegaspargase doses.19 However, prolonged asparaginase did not improve outcomes in two trials,2,20 and higher doses of pegaspargase did not improve outcome relative to standard doses.12 In the haematologica | 2022; 107(2)

Figure 5. Mercaptopurine cumulative dose intensity and absolute neutrophil count in patients negative or positive for anti-asparaginase antibodies. Mercaptopurine cumulative dose intensity (DI; left y axes) and absolute neutrophil count (ANC) in cells/mm3 (right y axes) for continuation weeks 10-16 in patients who were negative or positive for anti-asparaginase antibodies against Elspar (T15) or Oncaspar (T16) measured at continuation week 7. Boxes and whiskers represent quartiles and non-outlier ranges. Nominal P values were *P<0.05; **P<0.01, ***P<0.001. There was a total of four comparisons, thus the Bonferroni significance threshold=0.01

current analysis, we did not find that asparaginase dose intensity was related to disease free-survival in either T15 (using primarily native E. coli asparaginase) or T16 (using primarily pegaspargase) (Online Supplementary Table S12, Online Supplementary Figure S8). There are several possible explanations for this finding. First, we used more asparaginase on our studies than others did, thus perhaps exceeding some threshold value for exposure, consistent with the lack of influence of pegaspargase dosage on relapse in T16.12 Second, in those with allergy to their front-line asparaginase preparation, substitution with another formulation was aggressive, such that asparaginase dose intensity was not lower for those with or without allergy on T15 or on T16 (Online Supplementary Figure S9) and interpatient variability in dose intensity for asparaginase was low, i.e., <7% (Online Supplementary Table S3). Another possibility is that even in those patients with relatively low asparaginase dose 377

Figure 6. Absolute neutrophil count according to TPMT status on the T15 and T16 trials. Absolute neutrophil count (ANC, cells/mm3) was similar regardless of TPMT status on the T15 (top) and T16 (bottom) trials. ANC is depicted for the low-risk and standard-risk arms. The nominal P values are from the Wilcoxon rank sum test. Boxplots show the 25th and 75th percentiles; whiskers extend from 1.5 times the interquartile range, and data points outside the whiskers are depicted as dots. There was a total of 28 statistical comparisons, thus the Bonferroni significance threshold=0.002.

S.E. Karol et al.

378

haematologica | 2022; 107(2)

Dose intensity in ALL treatment

intensity, the dosages of other medications were increased, and this could have compensated for the lower exposure to asparaginase. In fact, we found that patients with low asparaginase dose intensity had significantly higher dose intensity for methotrexate on T16 (Online Supplementary Figure S10), consistent with our protocol recommendations for substituting methotrexate in those who could not receive asparaginase due to allergy or pancreatitis. These findings represent what is unique about the current study, because we have data to indicate how dosages of each medication may have influenced each other. This is the situation clinicians face: when pressing ahead with one medication is thwarted for some reason (drug shortages, adverse effects), other medications are often substituted. In these trials, it appears that the substitutions made were effective in maintaining efficacy, in that there were no associations between low exposure to any one drug and outcome. In our studies, as in other ALL studies, the primary drugs for which dosage adjustments are routinely made are mercaptopurine and methotrexate, which are titrated to a desired ANC, and indeed the greatest variability in dosages was observed for these two drugs (Online Supplementary Table S3). There is controversy as to whether increasing the dose intensity of mercaptopurine and/or methotrexate increases,21 decreases22 or has no effect16 on the risk of relapse. Complicating the interpretation of the data is that most studies do not systematically assess adherence, and most protocols use both drugs orally; thus, patients with the highest prescribed dose intensity may be the patients who are actually taking the lowest percentage of their therapy, thereby complicating any interpretation of how dose intensity affects relapse. For example, those with lower measured mercaptopurine adherence had higher relapse risk, but there was no association with mercaptopurine dose intensity.23 It is possible that low adherence to mercaptopurine also translates into lower adherence with other drugs not measured (e.g., glucocorticoids, methotrexate, supportive care drugs), thus compounding the risk of relapse due to poor adherence. In the current analysis, dosages of all antileukemic drugs were captured, and the only drug whose prescribed dose intensity was associated with relapse in unadjusted analyses on both T15 and T16 was mercaptopurine (Online Supplementary Table S12, Online Supplementary Figure S7), albeit not in the same phases of therapy for both studies. It should be noted that the direction of association was that higher dose intensity was associated with higher relapse, indicating that it is likely that high dose intensity identified patients who were either noncompliant (and thus clinicians needed to push the dosage to achieve desired neutropenia) or had very fast drug clearance. However, it should be noted than no metrics of dose intensity or of ANC were significantly associated with relapse or outcome using P value thresholds adjusted for the large number of exploratory comparisons. Although our data comprehensively capture prescribed mercaptopurine dose, one limitation of these data is that we did not assess adherence to administration of prescribed drug in patients who were not under direct medical supervision, which applied to most of the oral doses of glucocorticoids and mercaptopurine. Any association between dose intensity and outcomes is likely to be affected by the extent to which adherence is emphasized, the extent of thiopurine monitoring, and the rigor with which ANC targets are pursued. It should be noted that on T15, haematologica | 2022; 107(2)

there was a modest association between higher ANC and increased relapse, but this association was not statistically significant after correction for multiple testing, and was not reproducible, as no such association was observed on T16, and in subanalyses for different therapy phases, higher ANC tended to associate with both increased and decreased relapse. In both our studies, thiopurine starting dose was adjusted based on TPMT status,24 which appropriately resulted in lower dosages of mercaptopurine in those with a genetic defect in TPMT; herein, we show for the first time that this allowed for uncompromised dosing of the other chemotherapeutic agents (Online Supplementary Table S7) and no difference in ANC based on TPMT status (Figure 6). It is also worth noting that the dose intensity for mercaptopurine, especially in the first 6 months of therapy, was lower than reported by others;23 after reinduction, the median dose intensity was as low as 49% for one phase on T16, suggesting that our planned dosages may have been too high (Online Supplementary Table S4). Interestingly, the median dose intensity for the comparable time period on T15, when native asparaginase rather than pegaspargase was used, was higher at 76%. Had we realized a priori the impact of asparaginase on patients’ tolerance of thiopurine therapy, we could have designed a more realistic dosage regimen, and this finding has implications for future protocol design. There are conflicting data on the importance of “intensive” non-antimetabolite therapy in ALL.3,5-7,25-29 Although the lack of association of outcomes with dose intensity for most drugs in our study is fairly consistent with studies touting deintensification strategies, St. Jude differs from many other centers in that patients receive all weekly methotrexate parenterally (and thus return to the clinic every week, in contrast to many centers that see patients only every 4-6 weeks during continuation), LR and SR patients received vincristine/dexamethasone pulses throughout continuation, and a high percentage of patients (~50%) received therapy on the SR and HR arms, which include more asparaginase and other non-antimetabolite agents (cytarabine, cyclophosphamide) than many other treatment protocols. Thus, the lack of association between dose intensity and relapse we observed may not be extensible to centers with less intensive monitoring and/or less chemotherapy-dense and diverse regimens. It also suggests that, for patients intolerant to specific components of therapy (e.g., asparaginase due to pancreatic or hepatic toxicity), substitution with alternative chemotherapy may mitigate the adverse prognosis associated with early discontinuation of the offending agent. Given that therapy-limiting toxicities may preclude delivery of prescribed chemotherapy in 12-25% of patients receiving treatment on modern trials,19 prospective evaluation of chemotherapy substitution to address therapy-limiting toxicity should be considered. We conclude that intentional changes to the dose intensity of one agent, e.g., asparaginase, can have dramatic consequences on the ability to administer other conventional agents. Comprehensive data on chemotherapy actually delivered in cancer clinical trials are needed to fully interpret results and further optimize therapy. Disclosures MVR and HI and St. Jude Children’s Research Hospital receive investigator-initiated research funding from Servier Pharmaceutical. 379

S.E. Karol et al.

Contributions SEK, WEE, CHP, SJ and MVR conceived and designed the study; KRC, ARM, CC, DP, SEK, HI, JER, MLM, RCR, TAG, JJY, WEE, SJ, CHP, and MVR provided study material or patients and collected and/or assembled data; NMK, YL, JCP, ERF, CAS, WY, CC, YL and DP, analyzed and interpreted data; MVR wrote the first draft of the manuscript; all authors contributed to the revision of the manuscript; and all authors approved the final submitted version of the manuscript. Acknowledgments The authors thank the patients and parents who participated in the clinical protocols included in this study, and the participating clinicians and research staff.

References 1. Silverman LB, Gelber RD, Dalton VK, et al. Improved outcome for children with acute lymphoblastic leukemia: results of DanaFarber Consortium Protocol 91-01. Blood. 2001;97(5):1211-1218. 2. Albertsen BK, Grell K, Abrahamsson J, et al. Intermittent versus continuous PEGasparaginase to reduce asparaginase-associated toxicities: a NOPHO ALL2008 randomized study. J Clin Oncol. 2019;37(19): 1638-1646. 3. Moricke A, Zimmermann M, Reiter A, et al. Long-term results of five consecutive trials in childhood acute lymphoblastic leukemia performed by the ALL-BFM study group from 1981 to 2000. Leukemia. 2010;24(2):265-284. 4. Smid EJ, Driessen AJ, Konings WN. Mechanism and energetics of dipeptide transport in membrane vesicles of Lactococcus lactis. J Bacteriol. 1989;171(1): 292-298. 5. Veerman AJ, Kamps WA, van den Berg H, et al. Dexamethasone-based therapy for childhood acute lymphoblastic leukaemia: results of the prospective Dutch Childhood Oncology Group (DCOG) protocol ALL-9 (1997-2004). Lancet Oncol. 2009;10(10): 957-966. 6. Matloub Y, Bostrom BC, Hunger SP, et al. Escalating intravenous methotrexate improves event-free survival in children with standard-risk acute lymphoblastic leukemia: a report from the Children's Oncology Group. Blood. 2011;118(2):243-251. 7. Seibel NL, Steinherz PG, Sather HN, et al. Early postinduction intensification therapy improves survival for children and adolescents with high-risk acute lymphoblastic leukemia: a report from the Children's Oncology Group. Blood. 2008;111(5):25482555. 8. Boissel N, Baruchel A. Acute lymphoblastic leukemia in adolescent and young adults: treat as adults or as children? Blood. 2018;132(4):351-361. 9. Gaynon PS, Steinherz PG, Bleyer WA, et al. Association of delivered drug dose and outcome for children with acute lymphoblastic leukemia and unfavorable presenting features. Med Pediatr Oncol. 1991;19(4):221227. 10. Kantarjian H, Ravandi F, Short NJ, et al. Inotuzumab ozogamicin in combination with low-intensity chemotherapy for older patients with Philadelphia chromosomenegative acute lymphoblastic leukaemia: a

380

Funding This work was supported by National Institutes of Health GM115279, CA35401, CA142665, CA21765, and K08CA250418, and the American Lebanese Syrian Associated Charities. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Data-sharing statement De-identified data used in the preparation of this manuscript are available upon request.

single-arm, phase 2 study. Lancet Oncol. 2018;19(2):240-248. 11. Pui CH, Campana D, Pei D, et al. Treating childhood acute lymphoblastic leukemia without cranial irradiation. N Engl J Med. 2009;360(26):2730-2741. 12. Jeha S, Pei D, Choi J, et al. Improved CNS control of childhood acute lymphoblastic leukemia without cranial irradiation: St Jude Total Therapy Study 16. J Clin Oncol. 2019;37(35):3377-3391. 13. Kawedia JD, Liu C, Pei D, et al. Dexamethasone exposure and asparaginase antibodies affect relapse risk in acute lymphoblastic leukemia. Blood. 2012;119(7): 1658-1664. 14. Liu C, Janke LJ, Kawedia JD, et al. Asparaginase potentiates glucocorticoidinduced osteonecrosis in a mouse model. PLoS One. 2016;11(3):e0151433. 15. Merryman R, Stevenson KE, Gostic WJ 2nd, et al. Asparaginase-associated myelosuppression and effects on dosing of other chemotherapeutic agents in childhood acute lymphoblastic leukemia. Pediatr Blood Cancer. 2012;59(5):925-927. 16. Chessells JM, Harrison G, Lilleyman JS, Bailey CC, Richards SM. Continuing (maintenance) therapy in lymphoblastic leukaemia: lessons from MRC UKALL X. Medical Research Council Working Party in Childhood Leukaemia. Br J Haematol. 1997;98(4):945-951. 17. Usami I, Imamura T, Takahashi Y, et al. Discontinuation of L-asparaginase and poor response to prednisolone are associated with poor outcome of ETV6-RUNX1positive pediatric B-cell precursor acute lymphoblastic leukemia. Int J Hematol. 2019;109(4):477-482. 18. Duval M, Suciu S, Ferster A, et al. Comparison of Escherichia coli-asparaginase with Erwinia-asparaginase in the treatment of childhood lymphoid malignancies: results of a randomized European Organisation for Research and Treatment of Cancer-Children's Leukemia Group phase 3 trial. Blood. 2002;99(8):2734-2739. 19. Gupta S, Wang C, Raetz EA, et al. Impact of asparaginase discontinuation on outcome in childhood acute lymphoblastic leukemia: a report from the Children's Oncology Group. J Clin Oncol. 2020;38(17):1897-1905. 20. Mondelaers V, Suciu S, De Moerloose B, et al. Prolonged versus standard native E. coli asparaginase therapy in childhood acute lymphoblastic leukemia and non-Hodgkin lymphoma: final results of the EORTCCLG randomized phase III trial 58951.

Haematologica. 2017;102(10):1727-1738. 21. Schmiegelow K, Nielsen SN, Frandsen TL, Nersting J. Mercaptopurine/methotrexate maintenance therapy of childhood acute lymphoblastic leukemia: clinical facts and fiction. J Pediatr Hematol Oncol. 2014;36 (7):503-517. 22. Relling MV, Hancock ML, Boyett JM, Pui CH, Evans WE. Prognostic importance of 6mercaptopurine dose intensity in acute lymphoblastic leukemia. Blood. 1999;93(9): 2817-2823. 23. Bhatia S, Landier W, Hageman L, et al. Systemic exposure to thiopurines and risk of relapse in children with acute lymphoblastic leukemia: a Children's Oncology Group study. JAMA Oncol. 2015;1(3):287-295. 24. Relling MV, Schwab M, Whirl-Carrillo M, et al. Clinical Pharmacogenetics Implementation Consortium guideline for thiopurine dosing based on TPMT and NUDT15 genotypes: 2018 update. Clin Pharmacol Ther. 2019;105(5):1095-1105. 25. Vora A, Goulden N, Mitchell C, et al. Augmented post-remission therapy for a minimal residual disease-defined high-risk subgroup of children and young people with clinical standard-risk and intermediate-risk acute lymphoblastic leukaemia (UKALL 2003): a randomised controlled trial. Lancet Oncol. 2014;15(8):809-818. 26. Lange BJ, Bostrom BC, Cherlow JM, et al. Double-delayed intensification improves event-free survival for children with intermediate-risk acute lymphoblastic leukemia: a report from the Children's Cancer Group. Blood. 2002;99(3):825-833. 27. Conter V, Valsecchi MG, Silvestri D, et al. Pulses of vincristine and dexamethasone in addition to intensive chemotherapy for children with intermediate-risk acute lymphoblastic leukaemia: a multicentre randomised trial. Lancet. 2007;369(9556):123131. 28. De Moerloose B, Suciu S, Bertrand Y, et al. Improved outcome with pulses of vincristine and corticosteroids in continuation therapy of children with average risk acute lymphoblastic leukemia (ALL) and lymphoblastic non-Hodgkin lymphoma (NHL): report of the EORTC randomized phase 3 trial 58951. Blood. 2010;116(1):36-44. 29. Stock W, La M, Sanford B, et al. What determines the outcomes for adolescents and young adults with acute lymphoblastic leukemia treated on cooperative group protocols? A comparison of Children's Cancer Group and Cancer and Leukemia Group B studies. Blood. 2008;112(5):1646-1654.

haematologica | 2022; 107(2)

ARTICLE

Hematopoiesis

Reversible switching of leukemic cells to a drugresistant, stem-like subset via IL-4-mediated cross-talk with mesenchymal stroma

Ferrata Storti Foundation

Hae-Ri Lee,1 Ga-Young Lee,1 Eung-Won Kim,1 Hee-Je Kim,2 Min-Ho Lee,3 R. Keith Humphries4,5 and Il-Hoan Oh1 Catholic High-Performance Cell Therapy Center & Department of Medical Life Science, College of Medicine, The Catholic University, Seoul, Republic of Korea; 2Division of Hematology, Department of Internal Medicine, St Mary’s Hematology Hospital, College of Medicine, The Catholic University of Korea, Seoul, Republic of Korea; 3Department of Life Science, Dongguk University-Seoul, Goyang-si, Gyeonggi-do, Republic of Korea; 4 Terry Fox Laboratory, British Columbia Cancer Agency, Vancouver, British Columbia, Canada and 5Department of Medicine, University of British Columbia, Vancouver, British Columbia, Canada 1

Haematologica 2022 Volume 107(2):381-392

ABSTRACT

hemoresistance of leukemic cells has largely been attributed to clonal evolution secondary to accumulating mutations. Here, we show that a subset of leukemic blasts in contact with the mesenchymal stroma undergo cellular conversion into a distinct cell type that exhibits a stem cell-like phenotype and chemoresistance. These stroma-induced changes occur in a reversible and stochastic manner driven by cross-talk, whereby stromal contact induces interleukin-4 in leukemic cells that in turn targets the mesenchymal stroma to facilitate the development of new subset. This mechanism was dependent on interleukin-4-mediated upregulation of vascular cell adhesion molecule1 in mesenchymal stroma, causing tight adherence of leukemic cells to mesenchymal progenitors for generation of new subsets. Together, our study reveals another class of chemoresistance in leukemic blasts via functional evolution through stromal cross-talk, and demonstrates dynamic switching of leukemic cell fates that could cause a non-homologous response to chemotherapy in concert with the patient-specific microenvironment.

Correspondence: IL-HOAN OH iho@catholic.ac.kr

Introduction Acute myeloid leukemia (AML) is a heterogeneous, clonal hematopoietic disorder characterized by excessive proliferation of stem cell-like progenitor cells in the bone marrow (BM). AML has a highly variable prognosis1 and a very high risk of relapse particularly in elderly patients.2 Leukemia progression and relapse are widely viewed to occur via clonal evolution from preleukemic cells to overt leukemia driven by genetic mutations,3 followed by additional mutations leading to treatment-resistant, relapsed clone(s).4 However, indepth clonal analyses have revealed the persistence of founding clones,4 and functional heterogeneity among the developed leukemic clones,5 suggesting that other mechanisms may be involved. Several studies have highlighted leukemic stem cell (LSC) properties contributing to drug resistance:6 AML patients whose leukemic blast exhibit higher levels of stem cell signatures are at greater risk of relapse and have a poorer prognosis.7,8 However, the specific relationship between stemness and functional heterogeneity of LSC related to drug resistance, remains poorly understood.7,9 There is increasing awareness that the microenvironment, including growth factors, cytokines and niche stromal cells, can provide protection to leukemic cells and thereby contribute to the acquisition of chemoresistance.10,11 For example, leukemic cell subsets surviving chemotherapy were localized to the surface of osteoblasts in the BM.12-15 Subsequently, multiple protective signals from the stroma have been shown to enhance leukemic cell survival through activation of receptor tyrosine

haematologica | 2022; 107(2)

Received: August 16, 2020. Accepted: December 22, 2020. Pre-published: January 14, 2021. https://doi.org/10.3324/haematol.2020.269944

381

H.R. Lee et al.

kinases16 or interaction with the extracellular matrix. However, despite these protective signals, a role for the stroma in the clonal development of leukemic blasts for the acquisition of chemoresistance has not been demonstrated. Here, we show that subsets of leukemic cells in stromal contact undergo reversible changes associated with a stem cell-like phenotype and drug-resistant state. These changes are stochastic, and distinct from changes induced by other mechanisms of chemoresistance, thus representing a new class of drug-resistant cells developed in the leukemic microenvironment.

Gene expression analysis

Methods

Statistical analysis

Human sample collection Primary leukemic blasts were collected from newly diagnosed AML patients without prior treatment history. Part of the BM samples are from AML patients who had complete medical records during 5 years of follow-up in clinical courses. Human mesenchymal stromal cells (MSC) were separated from the BM of normal donors under informed consent. This study was approved by the Institutional Review Boards of St. Mary’s Hospital and Catholic University of Korea.

Animals C57/BL6 mice were obtained from the Jackson Laboratories (Bar Harbor, ME). Bis+/+, Bis+/-, Bis-/- mice17 were provided by Dr. Jeong-Hwa Lee (Catholic University of Korea). Mice with disruption of interleukin-4 (IL-4) receptor18 were provided by Dr. Chang Yul Kang (Seoul National University).

Mouse and human acute myeloid leukemia cells and mesenchymal stromal cells Fresh murine or human MSC were analyzed in the BM using flowcytometry. Cultured MSC (passage five to eight) were obtained by serial plating of BM cells in the DMEM containing 10% fetal bovine serum as described.19,20 For generation of murine AML cells, fluorouracil (5-FU) treated BM cells were transduced with MN-1 or Meis1/HoxA9 through retroviral infection as described.21,22 For co-culture, MSC were irradiated (15 Gy) 18-24 hours prior to use and leukemic cells were seeded on the MSC for co-culture. For co-culture with transwell, MSC were seeded into the upper chamber (6-well type, polyethylene terephthalate [PET] membrane with 0.4 mm pores; BD Bioscience, San Diego, USA) and leukemic cells were seeded into the lower well.

Flow cytometry of leukemic cells and mesenchymal stromal cells Murine leukemic cells were analyzed by flow cytometry using the following antibodies: CD45.1-APC (BD PharMingen, USA), Lineage cocktail (StemCell Technologies Inc, Canada), Sca-1-PECy7 and c-kit-PE (BD PharMingen). For human leukemic cells, CD45-APC, CD34-BV421, CD90-FITC (BD PharMingen) antibodies were used. For MSC, anti-CD106 (VCAM-1)-biotin, CD51-PE (eBioscience, CA, USA.), CD140a (PDGFRa)-APC, Sca1-PE-Cy7 (BD PharMingen) were used.

Treatment of antibody and cytotoxic drug Leukemic cells seeded on irradiated MSC were treated with anti-IL-4 antibody (R&D Systems Inc., USA), anti-CD106 (VCAM-1) (R&D Systems Inc.) for 3 days. For in vivo antibody injections, mice received intraperitoneal injection of anti-IL-4 Ab (1 mg/kg) (R&D Systems Inc.) or intravenous injection of antiVCAM-1 antbody (10 mg/kg) (Bio X cell, USA) along with

382

immunoglobulin G (IgG) from rat serum (Bio X cell). Cytotoxicity of in vivo leukemic cells was examined by treatment with 100 mg/kg of Ara-C (Sigma-Aldrich, MO, USA) and 3 mg/kg of doxorubicin hydrochloride (Sigma).

Sequencing libraries of two subjects were prepared according to the TruSeq Stranded Total RNA Sample Preparation guide. Aligned reads were quantified using HTSeq-count.23 Differentially expressed genes, fold ratio, P-value, and false discovery rate were identified by edgeR algorithm24 for each subject. Enriched KEGG pathways were identified by GSEA-P.25

In order to compare the generation of CD90+ subsets from individual primary human leukemia patients’ samples, or the responses of individual patients’ leukemia cells to chemotherapy, we used Mann-Whitney test. In order to compare the differences of means in specific experimental settings, we used a standard unpaired, two-tailed student t-test. The frequencies of leukemia-initiating cells in limiting dilution analysis were calculated by applying Poisson statistics with 95% Confidence Interval (CI) representing ±2 standard error of the mean (SEM).

Results A subset of leukemic cells acquires a stem cell-like phenotype by contact with mesenchymal stroma In order to investigate the influence of stromal cells on the function of leukemic cells, we employed an in vitro coculture model of murine leukemic cells in contact with BM-derived MSC. Murine AML cells were generated by transducing BM mononuclear cells (MNC) with meningioma-1 (MN1)21 or HoxA9-Meis1 (H9M1)22 (Figure 1A). When co-cultured with MSC a subset of MN1 leukemic cells acquired a Sca-1(+) phenotype (Lin-c-kit+sca-1+; LSK) mimicking normal hematopoietic progenitors, while the majority remained Sca-1(-) (Lin-c-kit+sca-1-) (Figure 1B). The acquisition of Sca-1(+) phenotypes was similarly observed in other types of leukemic cells (H9M1) or leukemia cell line (C1498) independent of irradiation (Online Supplementary Figure S1A and B). The emergence of the Sca1(+) subset was dependent on direct contact with the mesenchymal stroma (Online Supplementary Figure S1C), as these cells were not observed in stroma-free conditions or in stromal co-culture with a transwell filter (Figure 1B). In order to determine if acquisition of the Sca-1(+) phenotype occurs in vivo, MN1 leukemogenic cells (Lin-ckit+sca-1-) were transplanted into mice. Consistent with the in vitro results, a subset of leukemic cells (GFP+) in recipient mice acquired a Sca-1(+) (Lin-c-kit+sca-1+) phenotype (Figure 1C). In order to determine whether acquisition of the Sca1(+) phenotype in leukemic cells originated from their fusion with stromal cells, as implicated previously,26 we co-cultured MN1 leukemic cells (GFP+) with MSC transduced with YFP. None of the GFP+ leukemic cells coexpressed YFP (Figure 1D). Moreover, there was no difference in cell size between Sca-1(+) and Sca-1(-) cells, as determined by identical forward scatter in flow cytometry, and no increase in tetraploidy in the Sca-1(+) cells (Figure 1E). Similarly, there was no evidence of cell fusion in this in vivo generated Sca-1(+) subset (Online Supplementary Figure S1D). haematologica | 2022; 107(2)

Functional switching of leukemic cells by stromal contact

A recent study implicated mitochondrial transfer from MSC to leukemic cells during acquisition of chemoresistance.27,28 In order to examine this, MSC were labeled with a mitochondrial tracker and co-cultured with leukemic cells. There was no difference in mitochondrial tracker intensity between the Sca-1(-) and Sca-1(+) subsets (Figure 1F). Altogether, this emergence of a new leukemic subset with a stem cell-like phenotype (Sca-1(+)) represents an intrinsic cellular evolution of leukemic cells that occurs independently of cell fusion or mitochondrial transfer during in vivo leukemogenesis and in vitro culture with stromal cells.

Switching to the Sca-1(+) phenotype is reversible In order to determine if the Sca-1(+) subset is a stable phenotype, we sort-purified Sca-1(+) (LSK) and Sca-1(-) (LK) leukemic cells generated during co-culture with stro-

ma, and replated for a second round of co-culture with or without stroma. The purified Sca-1(-) cell fraction again generated Sca-1(+) cells during the second round selectively in the presence of stroma, whereas purified Sca1(+) cells co-cultured with stroma rapidly decreased in frequency (Figure 2A and B) with the emergence of a major Sca-1(-) cell population. Thus, final stable ratios of Sca-1(+) and Sca-1(-) cells were similarly maintained under secondary stromal co-culture conditions regardless of the phenotype of the initial cell population (Figure 2C). The changes in cell populations occurred rapidly within 3 days of co-culture suggesting that the conversion between Sca-1(-) and Sca-1(+) cells occurs by phenotypic switching rather than selective proliferation in the culture. Thus, the emergence of Sca-1(+) leukemic cells during stromal contact occurs in a reversible manner in any subsets of leukemic cells without clonal predisposition (sto-

Figure 1. Generation of a stem cell-like phenotype in a subset of leukemic cells. (A) Schematic illustration of the experiment. Murine acute myeloid leukemia (AML) cells were generated by transduction of fluorouracil (5-FU)-treated bone marrow (BM) cells with retrovirus encoding oncogene (MN1, or HoxA9/Meis1). Shown are retroviral vectors, experimental procedure for transplantation into mice, and the light microscopy morphology of transformed leukemic cells visualized by Giemsa staining. (B) Generation of Sca-1(+) (Lin-c-kit+sca-1+: LSK) leukemic cells during co-culture with murine mesenchymal stromal cells (mMSC). Co-cultures with mMSC for 3 days were performed in the presence (transwell) or absence (direct contact) of a transwell membrane between the cells in comparison to stroma-free (SF) culture. Phenotypes of leukemic cells (CD45+GFP+) from co-cultured MSC (CD45-GFP-) were analyzed by flow cytometry. Shown are the representative profile (left) and quantification (right) (mean ± standard error of the mean [SEM] , n=7, *P<0.05). (C) In vivo generation of Sca-1(+) leukemic subsets. MN1 leukemic cells (Lin-c-kit+) were transplanted into mice and generation of Sca-1(+) subsets among BM engrafted leukemic cells were examined at 2 weeks post-transplantation (95% green fluorescent protein postive [GFP+] leukemic cells at the point). Representative flowcytometry plot (left) and quantification (right) are shown (mean ± SEM, n=10, *P<0.05). (D) Experimental scheme for analyzing cell fusion between MSC and leukemic cells. MSC transduced with a retroviral vector encoding yellow fluorescent protein (YFP), and leukemic cells transduced with a vector encoding GFP were co-cultured for 3 days. Shown are the experimental scheme (left) and representative flow cytometry profiles showing the absence of double positive (YFP/GFP) populations before and after co-culture (right). (E) Flow cytometry profiles for comparison between LK (Sca-1(-)), and LSK (Sca-1(+)) cell populations of cell size by forward scattering (left), and of DNA content (right) (n=5). (F) Experimental scheme to compare mitochondrial transfer between Sca-1(+) and Sca-1(-) cell populations. Murine MSC were pre-labeled with MitoTracker and co-cultured with MN1 leukemic cells for 3 days. Shown are representative flow cytometry plots from the experiments, each indicated leukemic cell subset (LK or LSK) of leukemic cells (CD45+GFP+) was gated and analyzed for MitoTracker and quantified for difference in mitochondrial transfer (n=4).

haematologica | 2022; 107(2)

383

H.R. Lee et al.

Figure 2. Reversible and equipotent nature for generation of LSK leukemic subsets. LSK (Sca-1(+)) or LK (Sca-1(-)) subsets of MN1 leukemic cells generated by coculture were sort-purified and then replated for 3 days in the absence (SF) or presence of mesenchymal stromal cells (+MSC). (A) Flow cytometry profiles. (B and C) Quantitative analysis for expansion of cell numbers for LSK or LK subsets from input numbers of sorted LSK or LK cell populations during the second round of coculture with murine MSC. Shown are the fold increases of cell numbers compared with input in the second co-culture (B) and final frequencies for LSK from each set of the second co-culture (C) (mean ± standard error of the mean, n=6, *P<0.05).

chastic), but with similar probability of each cells (equipotent) for conversion among the total leukemic cell populations.

Functional heterogeneity acquired in stem cell-like leukemic subsets We next determined whether the Sca-1(+) stem cell-like leukemic subset arising by stromal contact was functionally distinct. When MN1 leukemic cells were treated with the chemotherapeutic Ara-C during co-culture with mesenchymal cells, Sca-1(-) subsets exhibited significant decrease of cell numbers, but the Sca-1(+) subset exhibited higher resistance compared to the Sca-1(-) subsets, with no significant changes in cell numbers (Figure 3A). Drug resistance in the Sca-1(+) subset was similarly reproduced in other leukemia cell types tested (HoxA9/Meis1-induced leukemic cells or C1498 leukemia cell line) (Figure 3A). The chemoresistance of the Sca-1(+) (LSK) leukemic population compared to the rest of the Sca-1(-) (LK) cells was similarly observed in vivo with mice engrafted with MN1 leukemic cells and treated with chemotherapeutic drug (Ara-C and doxorubicin)29 (Figure 3B). Thus, enhanced drug resistance is a common feature of leukemic subsets acquiring a Sca-1(+) phenotype upon stromal contact in a range of leukemic cell models. In order to further investigate the drug resistance of the Sca-1(+) cells, we analyzed their cell cycling in BM and found that % of quiescent cell population (G0) was higher in LSK cells (Figure 3C). We also compared the frequency of leukemic initiating cells (LIC), a functional assay for 384

leukemia stem cells30 in the Sca-1(+) leukemic subset in comparison to the other subsets. Thus, subsets of Lin(+) cells, Lin-c-kit-, LK (Lin-c-kit+sca-1-) cells and LSK (Lin-c-kit+ sca-1+) leukemic cells generated in the BM of MN1 transplanted mice were sort purified and transplanted into secondary recipient mice in a limiting dilution assay. However, the LK and LSK populations exhibited a similar frequency of LIC, while exhibited significantly higher frequencies than the other cell populations (Figure 3D; Online Supplementary Figure S2A). These two populations (LK and LSK) also exhibited comparable levels of in vivo leukemic engraftment or in vitro leukemia colony formation (Online Supplementary Figure S2B and C), indicating that the Sca-1(+) subset developed during in vivo leukemogenesis comprise a subset of LIC that does not display significantly different leukemogenic activity compared to their Sca-1(-) counterparts. Thus, the Sca-1(+) leukemia subset generated from leukemic cells represents a distinct leukemic cell population that has acquired drug-resistance without altering their leukemogenic activity.

Interleukin-4 plays a role in the emergence of drugresistant Sca-1(+) cells We next sought to identify possible signals from the stroma that induce emergence of Sca-1(+) cells. Given that altered production of cytokines and/or growth factors are frequently observed in leukemic cells,10 we examined the cytokine/growth factor gene expression induced by stromal contact of leukemic cells (Online Supplementary Figure S3A). haematologica | 2022; 107(2)

Functional switching of leukemic cells by stromal contact

Figure 3. Property of stem cell-like leukemic subsets exhibiting drug resistance without changes in leukemic initiating cell frequencies. (A) Murine leukemic cells were cultured in the presence of murine mesenchymal stromal cells (MSC), and numbers of surviving cells (annexinV- propidium iodide [PI]) in Sca-1(+) or Sca-1(-) subsets were measured after 2 days of treatment with Ara-C (100 nM for MN1 leukemia, 30 nM for H9M1, 500 nM for C1498 leukemic cells). (B) Mice engrafted with MN1 leukemic cells were treated with chemotherapeutic agents (Ara-C + doxorubicin) for 3 days by intraperitoneal injection at the indicated times and then examined for changes in the numbers of Sca-1(+) (LSK) or Sca-1(-) (LK) subsets in the bone marrow (BM). Experimental scheme (upper) and relative fold changes (lower) in the cell numbers after chemotherapy (chemoTx) compared to the control groups (mean ± standard error of the mean, n=5, *P<0.05). (C) Cell cycling of leukemic subsets in BM. MN1 leukemic cells were transplanted into mice and cell cycling of Sca-1(+) and Sca-1(-) subsets in BM were analyzed by Hoechst33342/pyronin staining. Shown are the representative flow cytometry plots with % of cell population (upper) and quantification of cells in G0 (quiescent cell population) and Non-G0 (G1/S/G2M) phase (lower) (n=3, *P<0.05). (D) Comparisons of leukemia-initiating cell (LIC) frequencies for each leukemic subset. MN1 leukemic cells were transplanted into mice and each subset of leukemic cells in recipient BM were sort-purified for transplantation into secondary recipients in a limiting dilution dose. Shown is the plot of limiting dilution analysis for frequencies of LIC in each leukemic subset analyzed by Poisson statistics. The resulting LIC frequencies are shown in the Online Supplementary Figure S2A with 95% Confidence Intervals in parenthesis.

Upon contact with stroma, the murine leukemic cells exhibited a notable induction of cytokines and growth factors implicated in leukemogenic activity, including IL-4, PDGF-A, PDGF-D, CCL-2, CCL-5, CXCL-1 and stem cell factor,31-39 but not in the presence of transwell filters (Online Supplementary Figure S3B). Among those cytokines, IL-4 was selectively induced in LSK subsets, but not in the majority of remaining cells (LK) as determined by its transcript and protein level (Online Supplementary Figure S3C and D). Thus, we examined whether IL-4 acts as an autocrine signal for generating Sca-1(+) subsets. Addition of recombinant IL-4 increased the frequency of Sca-1(+) subsets (LSK) in a dosedependent manner (Figure 4A). Conversely, addition of an IL-4-neutralizing antibody significantly decreased the frequency of LSK during co-culture (Figure 4A). Injection of an antibody against IL-4 into recipient mice along with MN1 leukemic cells also decreased the LSK population in the BM haematologica | 2022; 107(2)

of recipients without changes in overall engraftment levels (Figure 4B). Moreover, Il-4-neutralizing antibody abrogated resistance of the LSK population to chemotherapeutic drugs (Ara-C and doxorubicin), markedly decreasing the LSK population in recipient BM (Figure 4C), which caused a decrease in the residual burden of surviving LIC that can initiate leukemogenesis (Online Supplementary Figure S4). Together, these results support a key role for IL-4 in the generation of drug resistant Sca-1(+) subset upon stromal contact.

Interleukin-4-dependent generation of Sca-1(+) leukemic cells is generated by stromal cross-talk In order to investigate the mechanisms underlying IL-4mediated generation of Sca-1(+) subsets, we examined the cellular target of IL-4 during the co-culture of leukemic cells and stroma (Figure 5A). First, to see if IL-4 acts directly on 385

H.R. Lee et al.

Figure 4. Role of stroma-induced interleukin-4 in the generation of the stem cell-like leukemic subset. (A) Effects of interleukin-4 (IL-4) during co-culture of leukemic cells. Left: experimental scheme; middle: % LSK generated during 3-day co-culture of leukemic cells with stroma supplemented with recombinant IL-4 (mean ± standard error of the mean [SEM], n= 6); Right: effects of antibody against IL-4 on generation of LSK during stromal co-culture of leukemic cells. Shown are the mean±SEM for % LSK in leukemic cells (green fluorescent protein positive [GFP+] CD45+) (n=7, *P<0.05). (B) Effects of IL-4 antibody on in vivo generation of LSK. Left: experimental design. Antibody against IL-4 was intraperitoneally administered into recipient mice at each indicated time point before and after transplantation of MN1 leukemic cells. Middle: numbers of LSK leukemic (GFP+) cells in the BM (two femurs and two tibia) of recipient mice. Middle and right: % of MN1 leukemic cell engraftment determined by total leukemic cells (GFP+ cells) (middle) and cells with leukemia-initiating cell properties (LK and LSK) (right), respectively (mean ± SEM, n=6, *P<0.05). (C) Effects of IL-4 antibody on the chemosensitivity of the leukemic subsets. Left: experimental design. After engraftment of MN1 leukemic cells (10 days after transplantation), recipients were injected with IL-4 antibody and chemotherapeutic drug (AraC+doxorubicin) at the indicated times. Right: changes in the chemosensitivity of leukemic subsets by in vivo injected IL-4 antibody. Relative fold decrease in the cell numbers of each leukemic subset compared to the control (phosphate buffered saline) group 3 days after exposure to drug and antibody (mean ± SEM, n=5).

leukemic cells, we established MN1 leukemic cells from hematopoietic progenitors of mice lacking IL-4 receptor a (IL-4Ra knockout [KO]). Co-culture of leukemic cells from IL-4Ra KO or wild-type (WT) with stromal cells led to comparable frequencies of LSK or LK subsets in each group (Figure 5B). In contrast, when mesenchymal stromal cells from IL-4Ra KO mice were co-cultured with MN1 leukemic cells, significantly lower frequencies of LSK, but not LK subsets, were observed compared to the WT stroma group (Figure 5C). Thus, IL-4 signals target mesenchymal stromal cells, rather than leukemic cells, to facilitate stroma-mediated generation of the LSK subset, indicating that IL-4-mediated cross-talk promotes the functional evolution of leukemic cells. Next, to investigate the effects of IL-4 on mesenchymal stroma, we examined whether the mode of cellular interaction between MSC and leukemic cells is influenced by IL-4. LSK subsets were predominantly generated among the leukemic cells tightly adherent to the mesenchymal cells, for both MN1 or H9M1 leukemic cells, but seldom among the loosely adherent/suspension leukemic cells (Online Supplementary Figure S5A). Supporting the influence of IL-4 on stromal adherence, the level of vascular cell adhesion 386

molecules 1 (VCAM-1) in MSC, which mediate stromal adherence of leukemic cells,40 were up-regulated by IL-4 in WT MSC, but not in IL-4Ra KO MSC (Figure 5D). Conversely, in vivo injection of IL-4-neutralizing antibody caused a significant decrease of VCAM-1 expressions in BM mesenchymal cells including subsets enriched for mesenchymal progenitors (CD44(-)PDGFRa(+))41-43 (Figure 5E). In order to further examine the influences of stromal VCAM-1 expression level on the generation of LSK subsets, leukemic cells were co-cultured with sort-purified MSC fractions for different levels of VCAM-1. MSC with higher VCAM-1 levels increased LSK generation during co-culture, whereas MSC expressing lower levels of VCAM-1 decreased it, in comparison to LSK cells from unsorted MSC co-cultures (Figure 5F). Similarly, VCAM-1-blocking antibody significantly decreased stromal adherence of leukemic cells (Online Supplementary Figure S5B), which led to a concomitant decrease in the generation of the LSK subset (Figure 5G). Moreover, in vivo administration of VCAM-1 antibody caused a significant decrease of LSK numbers in BM (Figure 5H). These data, together with positive expression of VCAM-1 ligands in leukemic cells (Online Supplementary Figure S6) indicates that tight adherence of haematologica | 2022; 107(2)

Functional switching of leukemic cells by stromal contact

Figure 5. Interleukin-4-mediated cross-talk in mesenchymal stromal cells controlling VCAM-1-mediated generation of leukemic subsets. (A to C) Identification of interleukin-4 (IL-4) target in cross-talk of leukemic cells and mesenchymal stromal cells (MSC). (A) Experimental design. IL-4 receptor (IL-4Ra) knockout (LeukemiaKO) or wild-type (WT) leukemic cells were co-cultured with WT mice-derived murine MSC (MSC WT) (left), or MSC from IL-4Ra KO mice (MSC KO) or WT (right) were co-cultured with leukemic cells from WT mice. (B) Effects of IL-4Ra KO out in leukemic cells on the generation of LSK subsets. Shown are the quantification of % LSK and LK cells during co-culture (mean ± standard error of the mean[SEM], n=9, *P<0.05). (C) Effects of IL-4R KO in MSC on the generation of LSK subsets. Shown are the quantification of % LSK and % LK from the co-culture (mean± SEM, n=15, *P<0.05). (D to H) IL-4 targeting of MSC facilitates generation of leukemic subsets by controlling VCAM-1 expression in MSC. (D) Effects of IL-4 signals on VCAM-1 expression levels of MSC. Murine MSC from WT or IL-4R KO mice were treated with recombinant IL-4 and the fold increase of % VCAM-1(+) were analyzed in comparison to the control group (mean± SEM, n=6, *P<0.05). (E) In vivo changes of VCAM-1(+) cells in the bone marrow (BM) of mice injected with IL-4-neutralizing antibody. Mice were intraperitoneally injected with IL-4 antibody for 4 days and analyzed for % VCAM-1(+) cells in BM stromal cells. Shown are the relative fold differences of % VCAM-1(+) cells in the indicated subsets of BM mesenchymal stromal cells relative to the IgG treated mice group (n=5, *P<0.05). (F) Influence of VCAM-1 expression in MSC for generation of LSK subsets. Murine MSC were sort-purified for differences in VCAM-1 expression levels, and the generation of LSK subsets from MN1 leukemic cells co-cultured with each fraction was analyzed. Shown are the flow cytometry plots for sorting of MSC (left) and relative folds for LSK numbers generated in each co-culture group compared to co-culture with unsorted MSC (right) (n=6). (G) Effects of blocking antibody against VCAM-1 on the in vitro generation of LSK subsets. During co-culture of leukemic cells with stroma, the indicated amount of antibody against VCAM-1 was added and changes in the numbers of LSK generated in the co-culture were analyzed (n= 6, *P<0.05). (H) Effects of blocking antibody against VCAM-1 on the in vivo generation of LSK subsets. Mice transplanted with MN1 leukemic cells were injected with rat immunoglobulin (Ig) or VCAM-1-blocking antibody (intravenous 10 mg/kg) (7 and 10 days after leukemic cell transplantation). Three days after antibody injection, generation of LSK in recipient BM was analyzed (n=3 for IgG, n=4 for anti-VCAM1, *P<0.05).

haematologica | 2022; 107(2)

387

H.R. Lee et al.

Figure 6. Generation of stem-like, drug-resistant leukemic cells in human leukemia models. (A) Generation of stem-like (CD90(+)) leukemic subsets from each indicated human leukemia cell line during stromal co-culture. Leukemic cell populations (CD45(+))were gate separated from mesenchymal stromal cells (MSC) (CD45() for analysis. Shown are the quantification of the frequency of CD90(+) cells in the co-culture (n=7, † <0.2%, *P<0.05). (B) Comparison of drug sensitivity between each leukemic subset. Each indicated human leukemic cell line was exposed to Ara-C (200 nM) for 2 days. Shown are the numbers of surviving (annexinV- propidium iodide-) cells in the culture (mean ± standard error of the mean [SEM], n=6). (C to G) Generation of stem-like, drug-resistant leukemic subsets in human leukemic cells from acute myeloid leukemia (AML) patients. (C) Human leukemia cells from AML patients were co-cultured in the presence or absence of human MSC. Shown are the % CD90 (+) cells in total leukemic cells after co-culture for 3 days and differences were analyzed by Mann-Whitney U test (n=14 from seven individual patients’ samples, P<0.05). (D) Effects of interleukin-4 (IL-4) on the generation of CD90(+) subsets. Leukemic cells were co-cultured with human MSC in the presence or absence of IL-4 (100 ng/mL). Shown are the relative fold increases in % of CD90(+) subsets in leukemic cells (mean± SEM, n= 10 from five individual patient’s samples, *P<0.05). (E) Reversible switching of subsets of human primary leukemic cells to maintain constant equilibrium. CD90(+) and CD90(-) subsets generated during co-culture were sort-purified and re-plated in the co-culture with human MSC. Shown are the fold changes of each subset after plating each purified subset after 3 days of co-culture (n=6, *P<0.05). (F) Comparisons of drug sensitivity between the CD90(+) subset and the rest of the CD90(-) population in primary AML cells after exposure to Ara-C (200 nM). Shown are the relative fold changes in numbers of surviving cells of each population determined by numbers of annexinV-propidium iodide- (PI) cells and differences were analyzed by Mann-Whitney U test (n=14 from seven individual patients’ samples, P<0.05). (G) Enrichment of stem cell signatures in CD90(+) human leukemic cells. Primary human leukemic cells from two AML patients (#4 and #2) were co-cultured with human MSC for 3 days, and the generated CD90(+) and CD90(-) cells were subjected to RNA sequencing analysis. Differentially expressed genes were analyzed by gene set enrichment analysis (GSEA-P) for enrichment of 259 genes specific for leukemia stem cells.47 Shown are the plots of enrichment scores (upper) and ranked list of each gene in the order of log folds ratio (CD90+/90-) with position at zero indicated (lower). 2

leukemic cells to VCAM-1 in MSC facilitates emergence of LSK subsets. Consistent with these findings, gene expression changes in MSC induced by IL-4 treatment during culture revealed 41 differentially expressed genes (DEG), the most profound changes of which were in the gene ontology group related to the ‘binding’ molecular function, supporting their role in the cellular interaction with leukemic cells (Online Supplementary Figure S7A, B). Thus, IL-4 enhances the cellular interaction of stroma and leukemic cells to facilitate stroma-dependent evolution of the Sca-1(+) leukemic subset exhibiting drug resistance.

Stroma-induced changes in human leukemic cell models In order to investigate whether a similar phenomenon 388

can be seen in human leukemic cells, we examined human AML cells for acquisition of CD90(+) as a phenotype for stem-like subsets based on findings that a subset of CD90(+) cells amongst CD34(+) cells represent long-term repopulating hematopoietic stem cells (HSC)44 and that CD90 expression in human leukemic cells represents highrisk leukemia with stem cell properties.45,46 We first examined human leukemic cell lines, MOLM-14 and MV4-11 (M5 type FAB), and HL-60 (M3 type FAB). For each leukemic cell line tested, co-culture with human BMderived MSC resulted in the emergence of leukemic subsets with the CD90(+) phenotype, albeit to variable levels (Figure 6A). Moreover, when chemoresistance was compared between leukemic subsets, significant resistance to Ara-C treatment was observed selectively for CD90(+) cells in all tested leukemic cells (Figure 6B) similarly exhibiting haematologica | 2022; 107(2)

Functional switching of leukemic cells by stromal contact

Figure 7. Dependency on the mesenchymal progenitor cells for generation of stem-like leukemic subsets. (A) Identification of VCAM-1-expressing mesenchymal stromal cells in the bone marrow (BM) of mice. % of VCAM-1(+) cells among murine mesenchymal cells (CD45-31-Ter119-) in fresh BM were compared for CD44(+) and CD44(-) cells. Shown are the representative flow cytometry plots (left) and quantification (right) for frequency of VCAM-1(+) cells among the indicated murine mesenchymal stromal subsets in fresh BM. (B) Comparisons of frequency of VCAM-1(+) cells in mice BM between the mesenchymal progenitor and non-progenitor subsets of murine mesenchymal stromal cells. Mesenchymal progenitor subsets in fresh mice BM were defined by PDGFRa(+)/Sca-1(+) or PDGFRa(+)CD51(+) based on published reports.41-43,50 (C) Leukemogenesis in the Bis knockout (KO) mouse model. MN1 leukemic cells were transplanted into Bis KO mice, where mesenchymal progenitor populations are selectively decreased. Two weeks after transplantation into neonates of each mice model, engraftment of leukemic cells in BM and % LSK among engrafted leukemic cells were analyzed. Shown are the experimental design (upper) and % LSK leukemic subsets among engrafted leukemic cells for each indicated mice recipient (lower, left) and % engraftment of leukemic cells (GFP (+)) in BM (lower, right) (mean ± standard error of the mean [SEM], n=6 for wildtype [WT], n=27 for hetero, n=10 for KO). (D) Comparisons of mesenchymal progenitor cell numbers in BM of acute myeloid leukemia (AML) patients with respect to the clinical course. (upper) Experimental design. Fresh uncultured BM of AML patients without prior treatment were analyzed for cytogenetic abnormalities of leukemic blasts and content of mesenchymal progenitor cells (MPC; CD45-31-235a-146+166-) in fresh BM. Five years after the initial analysis, MPC numbers in patients’ fresh BM were compared with subsequent clinical courses (maintenance of complete remission or relapse) with respect to the karyotype of leukemic blasts. (lower) Mean numbers of MPC (CD146+166-) in fresh BM of AML patients for each indicated clinical course and karyotype (mean ± SEM, n=14 for normal karyotype, n=5 for mixed-lineage leukemia [MLL], n=10 for others).

quiescence in cell cycle (Online Supplementary Figure S8) In order to examine these findings in primary leukemic cells, we examined the response of primary AML blasts from five to seven individual patients to mesenchymal stroma. Primary AML blasts exhibited a significant induction of CD90(+) cells upon stromal contact, which was further increased by IL-4 treatment during co-culture (Figure 6C and D). Sort-purified subsets of CD90(+) and CD90(-) leukemic cells exhibited similar switching of phenotypes to maintain constant ratios in CD90(+) subsets in total leukemic cells, as observed for murine leukemic cells (Figure 6E). Moreover, the CD90 (+) subset generated during stromal contact exhibited higher resistance to Ara-C (Figure 6F) than the remaining CD90(-) cell population in the same coculture, demonstrating a similar drug-resistance of newly emerging leukemic subsets in primary human leukemic cells. Gene expression study on two independent patients showed that CD90(+) subsets of primary human AML cells are significantly enriched with gene sets specific for LSC47 haematologica | 2022; 107(2)

than the remaining CD90(-) cells (Figure 6G), and enriched with gene sets involved in the interaction with the extracellular matrix (ECM) or focal adhesion (Online Supplementary Figure S9). Thus, subsets of human leukemic cells in contact with stroma exhibit a stem-like properties to acquire drug-resistance through interaction with stroma.

Stromal heterogeneity for generation of stem cell-like leukemic subsets Extensive heterogeneity has been documented among mesenchymal populations in BM stroma.48,49 Therefore, we investigated the mesenchymal subpopulations responsible for generation of Sca-1(+) cells. Given that VCAM-1 expressing MSC played a role in the generation of drugresistant subsets, we examined VCAM-1 expression among stromal cell populations in the BM. VCAM-1(+) mesenchymal cells were predominantly enriched by a CD44(-) population, where colony-forming mesenchymal progenitor cells (MPC) are exclusively localized41-43,50 (Figure 7A). 389

H.R. Lee et al.

Similarly, VCAM-1 is exclusively enriched in subsets for MPC as defined by PDGFRa(+)/Sca-1(+) or PDGFRa(+)/CD51(+) subsets,41-43,50 indicating that the VCAM-1(+) cells that can drive emergence of drug-resistant subsets predominantly overlap with mesenchymal progenitor cells (Figure 7B). Supporting this finding, when MN1 leukemic cells were transplanted into the homozygous Bis KO mice, where selfrenewing MPC are decreased in the BM,17,51 a significant decrease in the Sca-1(+) subset (LSK) among BM engrafted leukemic cells was observed compared to WT or heterozygote Bis KO mice (Figure 7C). This decrease was not associated with altered overall engraftment levels (Figure 7C), consistent with the differences between LSK and LK subsets. Similarly, supporting the role of MPC for the development of drug resistant leukemic cells, AML patients wo relapsed after treatment exhibited higher numbers of mesenchymal progenitor subsets (MPC: CD146+/166-)19, 43, 52 in BM, i.e., retrospective studies on AML patients who had undergone relapse within 1 year after complete remission exhibited higher numbers of mesenchymal progenitor subsets in the BM than those who maintained complete remission for 5 years (Figure 7D). Notably, this difference was observed regardless of the underlying cytogenetic abnormality of the leukemic blasts, indicating that the heterogeneity of stromal cells could be an additional factor drugresistance of leukemic cells. Altogether, this heterogeneity in the BM MPC can influence the stroma-dependent generation of stem cell-like leukemic subsets.

Discussion Leukemic cell evolution has prevented the effective management of a diverse spectrum of leukemic disease. Here, using a variety of murine and human leukemia cells both in vitro and in vivo, we show that subsets of leukemic cells can undergo a phenotypic conversion into a stem-like phenotype that exhibit a higher resistance to chemotherapy in the context of stromal contact. This development of chemoresistant subsets by stromal contact was not dependent on cell fusion or changes in leukemogenic activities observed in mitochondrial transfer.27,28 The acquisition of the stem cell-like phenotype was reversible, being rapidly reverted to the non-stem cell phenotype under stroma-free conditions independent of difference in cell cycles or apoptosis (Online Supplementary Figure S10), unlike the stable maintenance of chemoresistance in leukemic clones generated by clonal evolution. Moreover, the frequencies of LSK cells among leukemic cells in contact with stroma were maintained constant regardless of the phenotype of the initial cell populations. This suggests that the stroma-mediated development of the stem cell-like, drug-resistant subpopulation occurs in a stochastic and reversible manner in leukemic cells with similar probabilities among leukemic cells (equipotent) without clonal predisposition. Reminiscent of these findings, recent studies showed that non-stem cancer cells can be spontaneously converted to stem-like state, and these plasticity of cancer cells allows cellular switching between distinct functional states.53,54 Together, these studies raise the possibility that the stochastic development of chemoresistant clones by stromal contact is an intrinsic process of leukemogenesis that could cause a non390

homogenous response to chemotherapy among the leukemic cell populations. The mechanisms for dynamic equilibrium among different subsets of leukemic cells remains still unclear. One possibility is a feedback control mechanism that maintains a constant ratio of stem-like versus non-stem-like leukemic cells, probably through cellular interaction between distinct leukemic subsets, as inferred from clonal interactions between heterogenous subsets.55 Similarly, studies on cancer stem cells have suggested that non-tumorigenic cells regulate the maintenance of cancer stem cells influencing their relative frequencies in the population.56 Since clonal heterogeneity of leukemia or cancer cells underlies the differential response to chemotherapy and emergence of relapsing clones,55,57,58 the kinetics of generating these stemlike subsets could be a factor for differential response to chemotherapy. Interestingly, we show that the development of these drug-resistant leukemic subset is facilitated by bi-directional cross-talk between stroma and leukemic cells mediated by IL-4, exhibiting resistance to apoptosis (Online Supplementary Figure S11). While IL-4 was implicated in inhibition of leukemic cells and apoptosis,59 we did not find increased apoptosis of the non-stem-like population precluding the selective enrichment of stem-like subsets by IL4 (Online Supplementary Figure S12). Moreover, rather than acting directly on the leukemic cells, IL-4 targets stromal cells, which facilitate the generation of LSK subsets. How IL-4 acts on stromal cells to facilitate the generation of a drug-resistant leukemic subset remains unclear. However, we demonstrated a key role for VCAM-1 downstream of IL-4 in MSC leading to tight adherence of leukemic cells and MSC, which was necessary for generation of the LSK subset. Similarly, we found IL-4-dependent induction of gene clusters in MSC whose functions are related to ‘binding function’. This suggests that the mode of interaction between MSC and leukemic cells is altered by IL-4 acting on MSC, which facilitates the development of the drugresistant leukemic subset. Interestingly, we also found functional differences between stromal cells in terms of their capacity to drive development of drug-resistant leukemic cells. We found that expression of VCAM-1 in stromal cell was important for adherence-dependent generation of leukemic subsets, while other adhesion molecules we tested did not influence the process (Online Supplementary Figure S13). Importantly, the VCAM-1-expressing stromal cells were selectively enriched in mesenchymal progenitors. Since mesenchymal stromal cells undergo various degenerative changes during leukemia,19,60,61 it is possible that patient-to-patient heterogeneity in BM mesenchymal progenitor cell content could differentially contribute to the development of drug-resistant clones. Consistent with this, AML patients whose BM has higher levels of a primitive (CD146+) subset of mesenchymal cells19,43,52 that express higher levels of VCAM-1 tend to have a higher risk of leukemic relapse compared to those who maintained complete remission. Thus, independent of oncogenic mutations or cytogenetic abnormalities in the blasts,62 heterogeneity per se in mesenchymal progenitors in BM could be another factor for development of drug-resistant leukemic subsets. In summary, our study reveals an additional mechanism of functional evolution of leukemic cells induced by contact with the mesenchymal stroma that can cause a reversible switch to a stem cell-like, drug-resistant subset independent haematologica | 2022; 107(2)

Functional switching of leukemic cells by stromal contact

of mutation-driven clonal evolution in leukemic blasts (Online Supplementary Figure S14). These findings thus provide further insight into the multiple mechanisms for development of drug-resistance that could generate leukemic cells with distinct characteristics and chemoresistance, highlighting the importance of the microenvironment in this process. This supports the need for better defining the mechanisms of drug resistance in leukemia patients, and could lead to the development of more comprehensive management of leukemic diseases. Disclosures No conflicts of interest to disclose. Contribution HRL, GYL, EWK, HJK performed experiments and collected data, MHL performed experiments on genomics and statistical analysis of data; RHK conceptualized research, provided study

References 1. Estey E, Dohner H. Acute myeloid leukaemia. Lancet. 2006;368(9550):18941907. 2. Craddock C, Tauro S, Moss P, Grimwade D. Biology and management of relapsed acute myeloid leukaemia. Br J Haematol. 2005;129(1):18-34. 3. Welch JS, Ley TJ, Link DC, et al. The origin and evolution of mutations in acute myeloid leukemia. Cell. 2012;150(2):264278. 4. Ding L, Ley TJ, Larson DE, et al. Clonal evolution in relapsed acute myeloid leukaemia revealed by whole-genome sequencing. Nature. 2012;481(7382):506510. 5. 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. 6. Dick JE. Acute myeloid leukemia stem cells. Ann N Y Acad Sci. 2005;1044:1-5. 7. Hope KJ, Jin L, Dick JE. Acute myeloid leukemia originates from a hierarchy of leukemic stem cell classes that differ in selfrenewal capacity. Nat Immunol. 2004;5 (7):738-743. 8. Eppert K, Takenaka K, Lechman ER, et al. Stem cell gene expression programs influence clinical outcome in human leukemia. Nat Med. 2011;17(9):1086-1093. 9. Guenechea G, Gan OI, Dorrell C, Dick JE. Distinct classes of human stem cells that differ in proliferative and self-renewal potential. Nat Immunol. 2001;2(1):75-82. 10. Ayala F, Dewar R, Kieran M, Kalluri R. Contribution of bone microenvironment to leukemogenesis and leukemia progression. Leukemia. 2009;23(12):2233-2241. 11. Konopleva M, Konoplev S, Hu W, Zaritskey AY, Afanasiev BV, Andreeff M. Stromal cells prevent apoptosis of AML cells by up-regulation of anti-apoptotic proteins. Leukemia. 2002;16(9):1713-1724. 12. Katsumi A, Kiyoi H, Abe A, et al. FLT3/ ITD regulates leukaemia cell adhesion through alpha4beta1 integrin and Pyk2 signalling. Eur J Haematol. 2011;86(3):191198. 13. Ninomiya M, Abe A, Katsumi A, et al. Homing, proliferation and survival sites of human leukemia cells in vivo in immunod-

haematologica | 2022; 107(2)

materials and wrote the manuscript; IHO conceptualized idea and research, supervised research, wrote the manuscript and provided financial support Acknowledgements We thank Dr. Lee, Jeong-Hwa (College of Medicine, Catholic University of Korea) for the kind supply of bis knock-out mice and Dr. Kang, Chang-Yul (College of Pharmacy, Seoul National University) for the kind supply of mice lacking the IL-4 receptor and Life Science Editors for manuscript editing. We thank Dr. JinA Kim for help in clinical data processing. We also thank the Department of Biostatistics of the Catholic Research Coordinating Center for statistical support. Funding This study was supported by the NRF of Korea and funded by Ministry of Science, ICT, & Future Planning (2017M3A9B3061947)

eficient mice. Leukemia. 2007;21(1):136142. 14. Saito Y, Uchida N, Tanaka S, et al. Induction of cell cycle entry eliminates human leukemia stem cells in a mouse model of AML. Nat Biotechnol. 2010;28(3):275-280. 15. Ishikawa F, Yoshida S, Saito Y, et al. Chemotherapy-resistant human AML stem cells home to and engraft within the bonemarrow endosteal region. Nat Biotechnol. 2007;25(11):1315-1321. 16. Doepfner KT, Boller D, Arcaro A. Targeting receptor tyrosine kinase signaling in acute myeloid leukemia. Crit Rev Oncol Hematol. 2007;63(3):215-230. 17. Youn DY, Lee DH, Lim MH, et al. Bis deficiency results in early lethality with metabolic deterioration and involution of spleen and thymus. Am J Physiol Endocrinol Metab. 2008;295(6):E1349-1357. 18. Kim IK, Kim BS, Koh CH, et al. Glucocorticoid-induced tumor necrosis factor receptor-related protein co-stimulation facilitates tumor regression by inducing IL9-producing helper T cells. Nat Med. 2015;21(9):1010-1017. 19. Kim JA, Shim JS, Lee GY, et al. Microenvironmental remodeling as a parameter and prognostic factor of heterogeneous leukemogenesis in acute myelogenous leukemia. Cancer Res. 2015;75(11):2222-2231. 20. Kim JH, Lee HS, Choi HK, et al. Heterogeneous niche activity of ex-vivo expanded MSCs as factor for variable outcomes in hematopoietic recovery. PloS One. 2016;11(12):e0168036. 21. Heuser M, Argiropoulos B, Kuchenbauer F, et al. MN1 overexpression induces acute myeloid leukemia in mice and predicts ATRA resistance in patients with AML. Blood. 2007;110(5):1639-1647. 22. Kroon E, Krosl J, Thorsteinsdottir U, Baban S, Buchberg AM, Sauvageau G. Hoxa9 transforms primary bone marrow cells through specific collaboration with Meis1a but not Pbx1b. EMBO J. 1998;17(13):37143725. 23. Anders S, Pyl PT, Huber W. HTSeq--a Python framework to work with highthroughput sequencing data. Bioinformatics. 2015;31(2):166-169. 24. Robinson MD, McCarthy DJ, Smyth GK.

edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 2010;26 (1):139-140. 25. Subramanian A, Kuehn H, Gould J, Tamayo P, Mesirov JP. GSEA-P: a desktop application for Gene Set Enrichment Analysis. Bioinformatics. 2007;23(23): 3251-3253. 26. Cogle CR, Goldman DC, Madlambayan GJ, et al. Functional integration of acute myeloid leukemia into the vascular niche. Leukemia. 2014;28(10):1978-1987. 27. Moschoi R, Imbert V, Nebout M, et al. Protective mitochondrial transfer from bone marrow stromal cells to acute myeloid leukemic cells during chemotherapy. Blood. 2016;128(2):253-264. 28. Wang J, Liu X, Qiu Y, et al. Cell adhesionmediated mitochondria transfer contributes to mesenchymal stem cell-induced chemoresistance on T cell acute lymphoblastic leukemia cells. J Hematol Oncol. 2018;11(1):11. 29. Zuber J, Radtke I, Pardee TS, et al. Mouse models of human AML accurately predict chemotherapy response. Genes Dev. 2009;23(7):877-889. 30. Bonnet D, Dick JE. Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell. Nat Med. 1997;3(7):730-737. 31. Kittang AO, Hatfield K, Sand K, Reikvam H, Bruserud O. The chemokine network in acute myelogenous leukemia: molecular mechanisms involved in leukemogenesis and therapeutic implications. Curr Top Microbiol Immunol. 2010;341:149-172. 32. Hassan HT, Zander A. Stem cell factor as a survival and growth factor in human normal and malignant hematopoiesis. Acta Haematol. 1996;95(3-4):257-262. 33. Mazur G, Wrobel T, Butrym A, KapelkoSlowik K, Poreba R, Kuliczkowski K. Increased monocyte chemoattractant protein 1 (MCP-1/CCL-2) serum level in acute myeloid leukemia. Neoplasma. 2007;54(4): 285-289. 34. Yang J, Liu X, Nyland SB, et al. Plateletderived growth factor mediates survival of leukemic large granular lymphocytes via an autocrine regulatory pathway. Blood. 2010;115(1):51-60. 35. Akashi K, Harada M, Shibuya T, et al.

391

H.R. Lee et al. Effects of interleukin-4 and interleukin-6 on the proliferation of CD34+ and CD34blasts from acute myelogenous leukemia. Blood. 1991;78(1):197-204. 36. Piccaluga PP, Rossi M, Agostinelli C, et al. Platelet-derived growth factor alpha mediates the proliferation of peripheral T-cell lymphoma cells via an autocrine regulatory pathway. Leukemia. 2014;28(8):1687-1697. 37. Tsuyada A, Chow A, Wu J, et al. CCL2 mediates cross-talk between cancer cells and stromal fibroblasts that regulates breast cancer stem cells. Cancer Res. 2012;72(11): 2768-2779. 38. Ding W, Knox TR, Tschumper RC, et al. Platelet-derived growth factor (PDGF)PDGF receptor interaction activates bone marrow-derived mesenchymal stromal cells derived from chronic lymphocytic leukemia: implications for an angiogenic switch. Blood. 2010;116(16):2984-2993. 39. Schulz A, Toedt G, Zenz T, Stilgenbauer S, Lichter P, Seiffert M. Inflammatory cytokines and signaling pathways are associated with survival of primary chronic lymphocytic leukemia cells in vitro: a dominant role of CCL2. Haematologica. 2011;96(3):408-416. 40. Juneja HS, Schmalsteig FC, Lee S, Chen J. Vascular cell adhesion molecule-1 and VLA-4 are obligatory adhesion proteins in the heterotypic adherence between human leukemia/lymphoma cells and marrow stromal cells. Exp Hematol. 1993;21(3):444450. 41. Morikawa S, Mabuchi Y, Kubota Y, et al. Prospective identification, isolation, and systemic transplantation of multipotent mesenchymal stem cells in murine bone marrow. J Exp Med. 2009;206(11):2483-2496. 42. Pinho S, Lacombe J, Hanoun M, et al. PDGFRalpha and CD51 mark human nestin+ sphere-forming mesenchymal stem cells capable of hematopoietic progenitor cell expansion. J Exp Med. 2013;210(7): 1351-1367. 43. Qian H, Le Blanc K, Sigvardsson M. Primary mesenchymal stem and progenitor

392

cells from bone marrow lack expression of CD44 protein. J Biol Chem. 2012;287(31): 25795-25807. 44. Peault B, Weissman IL, Buckle AM, Tsukamoto A, Baum C. Thy-1-expressing CD34+ human cells express multiple hematopoietic potentialities in vitro and in SCID-hu mice. Nouv Rev Fr Hematol. 1993;35(1):91-93. 45. Buccisano F, Rossi FM, Venditti A, et al. CD90/Thy-1 is preferentially expressed on blast cells of high risk acute myeloid leukaemias. Br J Haematol. 2004;125(2): 203-212. 46. Yamazaki H, Nishida H, Iwata S, Dang NH, Morimoto C. CD90 and CD110 correlate with cancer stem cell potentials in human T-acute lymphoblastic leukemia cells. Biochem Biophys Res Commun. 2009;383(2):172-177. 47. Saito Y, Kitamura H, Hijikata A, et al. Identification of therapeutic targets for quiescent, chemotherapy-resistant human leukemia stem cells. Sci Transl Med. 2010;2(17):17ra19. 48. Oh IH, Kwon KR. Concise review: multiple niches for hematopoietic stem cell regulations. Stem Cells. 2010;28(7):1243-1249. 49. Wei Q, Frenette PS. Niches for hematopoietic stem cells and their progeny. Immunity. 2018;48(4):632-648. 50. Mendez-Ferrer S, Michurina TV, Ferraro F, et al. Mesenchymal and haematopoietic stem cells form a unique bone marrow niche. Nature. 2010;466(7308):829-834. 51. Kwon KR, Ahn JY, Kim MS, Jung JY, Lee JH, Oh IH. Disruption of bis leads to the deterioration of the vascular niche for hematopoietic stem cells. Stem Cells. 2010;28(2):268-278. 52. Sacchetti B, Funari A, Michienzi S, et al. Self-renewing osteoprogenitors in bone marrow sinusoids can organize a hematopoietic microenvironment. Cell. 2007;131(2):324-336. 53. Chaffer CL, Brueckmann I, Scheel C, et al. Normal and neoplastic nonstem cells can spontaneously convert to a stem-like state.

Proc Natl Acad Sci U S A. 2011;108 (19):7950-7955. 54. da Silva-Diz V, Lorenzo-Sanz L, BernatPeguera A, Lopez-Cerda M, Munoz P. Cancer cell plasticity: impact on tumor progression and therapy response. Semin Cancer Biol. 2018;53:48-58. 55. McGranahan N, Swanton C. Biological and therapeutic impact of intratumor heterogeneity in cancer evolution. Cancer Cell. 2015;27(1):15-26. 56. Sprouffske K, Athena Aktipis C, Radich JP, Carroll M, Nedelcu AM, Maley CC. An evolutionary explanation for the presence of cancer nonstem cells in neoplasms. Evol Appl. 2013;6(1):92-101. 57. Jamal-Hanjani M, Quezada SA, Larkin J, Swanton C. Translational implications of tumor heterogeneity. Clin Cancer Res. 2015;21(6):1258-1266. 58. Quek L, David MD, Kennedy A, et al. Clonal heterogeneity of acute myeloid leukemia treated with the IDH2 inhibitor enasidenib. Nat Med. 2018;24(8):11671177. 59. Peña-Martínez P, Eriksson M, Ramakrishnan R, et al. Interleukin 4 induces apoptosis of acute myeloid leukemia cells in a Stat6-dependent manner. Leukemia. 2018;32(3):588-596. 60. Schepers K, Pietras EM, Reynaud D, et al. Myeloproliferative neoplasia remodels the endosteal bone marrow niche into a selfreinforcing leukemic niche. Cell Stem Cell. 2013;13(3):285-299. 61. Zhang B, Ho YW, Huang Q, et al. Altered microenvironmental regulation of leukemic and normal stem cells in chronic myelogenous leukemia. Cancer Cell. 2012;21(4): 577-592. 62. Byrd JC, Mrozek K, Dodge RK, et al. Pretreatment cytogenetic abnormalities are predictive of induction success, cumulative incidence of relapse, and overall survival in adult patients with de novo acute myeloid leukemia: results from Cancer and Leukemia Group B (CALGB 8461). Blood. 2002;100(13):4325-4336.

haematologica | 2022; 107(2)

ARTICLE

Hematopoiesis

Aging of human hematopoietic stem cells is linked to changes in Cdc42 activity

Ferrata Storti Foundation

Amanda Amoah,1 Anja Keller,1° Ramiz Emini,2° Markus Hoenicka,2 Andreas Liebold,2 Angelika Vollmer,1 Karina Eiwen,1 Karin Soller,1 Vadim Sakk,1 Yi Zheng,3 Maria Carolina Florian1° and Hartmut Geiger1 Institute of Molecular Medicine, Ulm University, Ulm, Germany; 2Department of Cardiothoracic and Vascular Surgery, Ulm University Hospital, Ulm, Germany and 3 Cincinnati Children’s Hospital Medical Center and University of Cincinnati, Cincinnati, OH, USA 1

°AK current address: Zentrum für Integrative Psychiatrie, Lübeck Campus, Lübeck, Germany

Haematologica 2022 Volume 107(2):393-402

°RE current address: Herzzentrum, Uniklinik Köln, Köln, Germany °MCF current address: Bellvitge Institute for Biomedical Research, IDIBELL, Barcelona, Spain

ABSTRACT

n this study, we characterize age-related phenotypes of human hematopoietic stem cells (HSC). We report increased frequencies of HSC, hematopoietic progenitor cells and lineage negative cells in the elderly but a decreased frequency of multi-lymphoid progenitors. Aged human HSC further exhibited a delay in initiating division ex vivo though without changes in their division kinetics. The activity of the small RhoGTPase Cdc42 was elevated in aged human hematopoietic cells and we identified a positive correlation between Cdc42 activity and the frequency of HSC upon aging. The frequency of human HSC polar for polarity proteins was, similar to the mouse, decreased upon aging, while inhibition of Cdc42 activity via the specific pharmacological inhibitor of Cdc42 activity, CASIN, resulted in re-polarization of aged human HSC with respect to Cdc42. Elevated activity of Cdc42 in aged HSC thus contributed to age-related changes in HSC. Xenotransplant, using NBSGW mice as recipients, showed elevated chimerism in recipients of aged compared to young HSC. Aged HSC treated with CASIN ex vivo displayed an engraftment profile similar to recipients of young HSC. Taken together, our work reveals strong evidence for a role of elevated Cdc42 activity in driving aging of human HSC, and similar to mice, this presents a likely possibility for attenuation of aging in human HSC.

Introduction Aging is associated with tissue degeneration, aging-related diseases and an increased susceptibility to infections.1,2 These hallmarks of aging have been linked to aging-related changes within somatic stem cell compartments, and primarily investigated in animal models like mice.3,4 One of the most extensively studied somatic stem cell-based system is the hematopoietic system. Hematopoietic stem cells (HSC) maintain blood homeostasis and show an age-related decline in overall function in mice,5 which includes an increase in myelopoiesis,6 accumulation of DNA damage,7 changes in epigenomic and transcriptional programs,8 decreased cell polarity and aberrant activity of the small RhoGTPase Cdc42.9 Although significant progress has been achieved in elucidating mechanisms of aging of murine HSC, it remains unclear whether these mechanisms can be simply extrapolated to other species, including humans. Early studies on HSC number and function in larger mammals showed, based on stochastic modeling, clear differences in HSC biology and aging to the murine model.10,11 These data suggested that HSC from non-human primates cycle more slowly and that fewer numbers of HSC clones actively contribute to hematopoiesis in humans at steady-state than in mice. This connotes that non-human primates and murine HSC may undergo different stress intensities such as the rate of accruing DNA damage and, as a result, may exhibit dissimilarities in aging.

haematologica | 2022; 107(2)

Correspondence: HARTMUT GEIGER, hartmut.geiger@uni-ulm.de Received: August 13, 2020. Accepted: December 22, 2020. Pre-published: January 14, 2021. https://doi.org/10.3324/haematol.2020.269670

393

A. Amoah et al.

In addition, mice show an increase in HSC frequency with age, while the rhesus monkey, shows a decrease with age.12 Moreover, even among distinct murine inbred strains, HSC number and function is distinct upon aging, like C57BL/6 mice present with an elevated number of HSC upon aging, but not so in DBA/2 animals.13,14 For these reasons, novel studies into understanding mechanisms of aging of human HSC are warranted and are a prerequisite to bolster the transition of this knowledge into the clinic. Age-related changes in the frequency and function of HSPC have been in part previously described by a small number of groups. One study for example reported no changes in the re-population potential of aged HSC and a decreased propensity for myeloid differentiation while another recorded a decline in the reconstitution capacity of aged HSC with an increased myeloid differentiation potential.15,16 Both groups used NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) animals as recipients for in vivo xenotransplantation to study the function of human HSC. This model requires irradiation of the recipient animal for the successful establishment of xenotransplants17 which might contribute to variable secondary effects not linked to the transplanted HSC. New and improved mouse models have been created that do not require pre-conditioning of the recipients for achieving human xeno+ NOD.Cg-KitW-41J Tyr Prkdcscid chimerism.18,19 Il2rgtm1Wjl/ThomJ (NBSGW) animals bear in addition to the NSG genotype a mutation in the Kit gene.19 The Kit mutation enables donor cells to efficiently engraft without irradiation.20 We characterize here aged human HSC with a special focus on likely shared hallmarks of age-related changes among human and murine HSC and describe a novel approach to attenuate aging of human HSC. Our data support the possibility of rejuvenating the function of aged human HSC due to similarities between aging of murine and human HSC.

Colony forming unit assay In order to assess the myeloid and erythroid generative potential of samples, 200 HSC were seeded in methylcellulose medium (further details are provided in the Online Supplementary Methods).

Xenotransplantation All animal experiments were carried out in accordance to institutional guidelines and approved by the Regierungspräsidium Tübingen (TVA 1412). Five hundred HSC were injected via the tail vein into non-conditioned or low dose irradiated (1.6 Gy) NBSGW mice. At 8 and 12 weeks, aspirates were drawn from the bone marrow (BM) of mice after administering anesthesia. Human cells were identified using human-specific antibodies (see the Online Supplementary Methods for further details) and analyzed on LSR Fortessa SORP flow cytometer (BD Biosciences). Human chimerism was determined as a percentage of total CD45+ cells and mature cells, as a percentage of human CD45+ cells.

CASIN treatment HSC were collected in serum-free expansion media and incubated at 37°C, 3% oxygen for 1 hour. Cells were then transferred into media ± CASIN, incubated for 4 hours and washed. Cells were then used in subsequent experiments.

Immunofluorescent staining Cells were seeded in serum-free expansion media, fixed and polarity for Cdc42 or tubulin assessed as previously described by Florian et al.9 Cdc42-GTP in HSC was determined using the antibody described by Althoff et al.22 (see the Online Supplementary Methods for further details).

Western blot A Rac/Cdc42 assay reagent (# 14-325, Millipore) was used in pull down assays according to the manufacturer’s protocol (see the Online Supplementary Methods for further details).

Statistical analysis Methods Primary cells Bone marrow cells were isolated from young (range, 23-39 years; median age 27 years) donors acquired from Cincinnati Children’s Hospital Medical Center and aged (range, 58-82 years; median age 66 years) individuals undergoing heart surgery at the Ulm University Clinic, Department of Heart, Thoracic and Vascular Surgery (additional details of age strata are provided in the Online Supplementary Methods). All donors were hematologically healthy. Sample collection and investigation was approved by the Internal Review Board (Ethikkomission) of Ulm University (392/16).

Flow cytometric analysis and cell sorting Mononuclear cells (MNC) were thawed and stained in phosphate-buffered saline (PBS) supplemented with 3% fetal bovine serum (FBS) with human specific antibodies (see the Online Supplementary Methods for details). Different cell populations were identified and sorted on a BD FACS ARIA II 4L SORP (BD Biosciences) according to the markers used by van Galen et al.21

Single cell division assay Single HSC were sorted into Terasaki plates and checked every 12 hours under a light microscope (further details are provided in the Online Supplementary Methods).

394

Statistical analyses were performed with GraphPad Prism 8 (version 8.1.2) and are presented as mean ± standard deviation (SD) or mean ± standard error of the mean (SEM) and box plots as minimum and maximum points. Kendall’s correlation analysis was performed with R version 4.0.3, RStudio Team (2020) version 1.3.1093. *P<0.05, **P<0.005, ***P<0.0005.

Results Changes in the immunophenotypic frequencies of hematopoietic populations occur with age Data on whether there are changes in the frequency of HSC in the BM of humans upon aging remains controversial. This might be, at least in part, due to different gating strategies employed to identify human HSC.15,16 Using a more recently established and improved marker profile for the identification of human HSC21 (Online Supplementary Figure S1A), we first determined the frequency of HSC (Lin-CD34+SSc low CD38-CD90+CD45ra-), HSPC (Lin-CD34+SSc low CD38-) and Lin-CD34+SSc low cells in BM cells from the sternum of the elderly. While the frequency of the Lin-CD34+SSc low population did not change with age within our cohort, the HSPC frequency within Lin-CD34+SSc low population and HSC frequency within HSPC population increased with age (Figure 1A to C). haematologica | 2022; 107(2)

Cdc42 and aging of human HSC

Figure 1. Flow cytometric analysis of different bone marrow populations isolated from young (white) and aged (grey) donors. Populations of interest within the lowdensity mononuclear cell fraction (MNC) were identified and frequencies of (A) Lin-CD34+ssc low, (B) hematopoietic stem progenitor cells (HSPC), (C) hematopoietic stem cells (HSC) and (D) Lin- cells from young and aged donors were determined. *P<0.04, **P<0.005; Mann-Whitney and t-test with Welch’s correction. Bars represent the mean ± standard error of the mean (SEM). (E) Representative image of common myeloid progenitor/ megakaryocyte–erythroid progenitor (CMP/MEP), multipotent progenitor (MPP) and multipotent lymphoid progenitor (MLP) gates. (F) The frequency of CMP/MEP, MPP and MLP in the Lin- fraction of donors. **P=0.003; Mann-Whitney test. Bars represent the mean ± SEM. 17< n >20; 23< n >29. Donor age: young =23-39 years (yr), median =27 yr; aged =58-82 yr, median =65 yr. n : number of young donors; n : number of aged donors. young

young

These results support an increase in the HSPC population with age previously described for the iliac crest.16,23 Our observations further indicate that the age-related increase in frequency is not restricted to a single anatomical site or, in our studies, influenced by sex (Online Supplementary Figure S1B and C). In addition, the frequency of HSC within the Lin-CD34+SSc low population in the sternum was also significantly higher than in the young16 (Online Supplementary Figure S1D). We also found a not yet described increase in the frequency of lineage negative cells within the MNC population in aged donors (Figure 1D). The frequency of common myeloid progenitor/ megakaryocyte–erythroid progenitor (CMP/MEP, LinCD34+SSc low CD38+CD90-CD45ra-) and multipotent progenitor (MPP, Lin-CD34+SSc low CD38-CD90-CD45ra-) did not change upon aging, while the frequency of immunophenotypic multipotent lymphoid progenitor (MLP, Lin-CD34+SSc low CD38-CD90-/low CD45ra+) within the Lin- population decreased significantly (Figure 1E and F). Our results demonstrate and confirm that aging is associated with an increase in the frequency of hematopoietic progenitor and HSC, but with a decrease in the frequency of MLP. haematologica | 2022; 107(2)

aged

Aged hematopoietic stem cells are delayed in the initiation of division We next tested whether the age-related increase in the HSC frequency might be linked to an elevated division rate of aged HSC. To this end, we determined the dynamics of first or second divisions of individual HSC ex vivo (Figure 2A). In general, BM-derived HSC showed a delayed initiation of division when compared to HSC from cord blood (CB) or HSC mobilized to blood (Online Supplementary Figure S2A and B). Surprisingly, aged HSC actually showed a delay until the first 50% of HSC underwent their first division compared to young HSC (Online Supplementary Figure S2B and C). This delay in initiation of division of aged HSC was still imminent in the presence of a different combination of cytokines (Online Supplementary Figure S2C to E) as well as under normoxic conditions (Online Supplementary Figure S2 F and G). The overall rate of division after initiation though was similar for both young and aged HSC for both the first division (Figure 2E) as well as the second division (Figure 2D to E; Online Supplementary Figure S2E). We next examined the proportion of cells in the distinct phases of the cell cycle by staining for DNA (Hoechst 33342) and Ki-67 (Ki-67 antibody) 395

A. Amoah et al.

Figure 2. Kinetics of single cell division of young (blue) and aged (green) hematopoietic stem cells cultured ex vivo. (A) Experimental design of cells singly sorted into plates containing M4 media and incubated at 37°C, 3% oxygen. (B) Cumulative first division of live young and aged hematopoietic stem cells (HSC) and (C) time by which 50% of the cells have undergone the first division. *P=0.03; t-test with Welch’s correction. Bars represent the mean ± standard error the mean (SEM). (D) Cumulative second division of live young and aged HSC. (E) Slope was derived as linear regression fits and probability values calculated from the correlation coefficients (M4 calculations in the box). Each curve was derived from cumulative gaussian fits with robust regression. n =6; n =5 different donors. Donor age: young =27-39 years (yr), median =28 yr; aged =64-75 yr, median =69 yr. young

(Online Supplementary Figure S3A and B). The proportions of young and aged HSC in the G , G and S-G -M phases of the cell cycle were similar, hence the delayed initiation of division was not simply driven by a higher frequency of quiescent cells upon aging. In aggregation, aged HSC show a delayed initiation of division that is not linked to a higher frequency of cells in G or G , while once division is initiated, there is no difference in the overall kinetics of the division. 0

Cdc42 activity is increased in human hematopoietic stem cells from the elderly and correlates with an increase in hematopoietic stem cell frequency The small RhoGPTase Cdc42 cycles between an active (GTP bound) and inactive (GDP bound) state24 and has been shown to have essential roles in HSC regulation.25 We previously demonstrated that the active form (GTPbound form) of the small RhoGTPase Cdc42 was increased in murine low-density bone marrow (LDBM) cells as well as in HSC upon aging, and that this increase in HSC resulted in the age-related increase in HSC frequency in mice.9 We therefore determined the level of activity of Cdc42 in human LDBM from the elderly by a 396

aged

standard pull down assay (Figure 3A; Online Supplementary Figure S4A). Pull-down replicates on the same samples from randomized donors demonstrated overall high reproducibility of the pull-down assay also on primary human samples (Online Supplementary Figure S4B). We observed an approximate 4-fold increase in the relative Cdc42 activity in LDBM cells from elderly donors compared to the young (Figure 3B) which was re-affirmed in correlation analysis of Cdc42 activity and age (Online Supplementary Figure S4C). Using immunofluorescence imaging, we found aged HSC had significantly higher Cdc42-GTP expression than young HSC (Figure 3C to E; Online Supplementary Figure S4D), implying primitive stem cells also undergo Cdc42 activity changes upon aging. In addition, we identified a positive association between Cdc42 activity and HSC frequency but not between Cdc42 activity and HSPC frequency (Figure 3F and G). Furthermore, the spread of values and, thus, the standard deviation (Figure 3B) was higher in the aged group than in the young, which points to a strong increase in heterogeneity of the hematopoietic system in individuals upon aging. Heterogeneity upon aging in mice is, for example, less pronounced due to their inbred nature.26–28 In aggregahaematologica | 2022; 107(2)

Cdc42 and aging of human HSC

Figure 3. Relative expression of Cdc42 activity in young and aged hematopoietic stem cells. (A) Representative image of western blot. (B) Quantitative expression of relative Cdc42 activity normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH). *P= 0.010, if the two points with activity higher than 20 are excluded, then P=0.017; Mann-Whitney test. Bars represent the mean ± standard error of the mean (SEM). n =13; n =41. Donor age: young =23-39 years (yr), median = 27 yr; aged = 60-82 yr, median = 66 yr. (C) Representative confocal image of Cdc42-GTP expression in hematopoietic stem cells (HSC). Cdc42-GTP quantification of young and aged HSC normalized to (D) DAPI intensity and (E) cell size. ***P=0.0001, **P=0.0012; Mann-Whitney test. Bars = mean ± SEM and n =66; n =67, from three different donors per cohort. Donor age: young =27-31 years (yr), median =27 yr; aged =63-76 yr, median =76 yr. Scale bar represents 2 mm. (F) Correlation analysis (Spearman) of relative Cdc42 activity and HSC frequency (r=0.4, P= 0.05, n=37) and (G) hematopoietic stem progenitor cell (HSPC) frequency (r= 0.3, P=0.175, n=22), broken grey lines represent 95% Confidence Interval. young

aged

young

tion, the data show that similar to mice, primitive hematopoietic cells from elderly humans show elevated Cdc42 activity. The level of Cdc42 activity in individuals correlates positively with the frequency of HSC, supporting a possible role for Cdc42 activity in causing the elevated HSC frequency, similar to what has been described in the mouse.8,29

The frequency of hematopoietic stem cells polar for Cdc42 and tubulin declines upon aging Another established age-related hallmark for murine HSC is the reduction in frequency of cells polar for cytosolic polarity proteins like tubulin and Cdc42 (Figure 4A). This “apolarity” of aged murine HSC is a direct consequence of the elevated activity of Cdc42 itself in aged HSC9 and likely results in a change in the mode of the division of aged murine HSC.8 We therefore determined the frequency of aged human HSC that showed a polar distribution of haematologica | 2022; 107(2)

aged

Cdc42 and tubulin by immunofluorescence analyses (Figure 4B to D). Approximately 70% of young HSC showed a polar distribution of Cdc42 while approximately 70% of the aged HSC showed an apolar distribution (Figure 4C). The frequency of aged HSC polar for tubulin was also reduced (Figure 4D). Our findings establish that aged human HSC present with a reduced frequency of cells polar for polarity proteins. Oddly, we did not find a significant association between Cdc42 polarity and tubulin polarity in human HSC (Online Supplementary Figure S5A) suggesting these two parameters might not be directly correlated as has been shown for murine HSC.9 Nonetheless, we observed a strong negative association between Cdc42 polarity and age (Online Supplementary Figure S5B), implying the frequency of cells that remain polar decreases with increasing age. In order to determine whether the level of Cdc42 activity in human cells might be linked to the frequency of cells polar for Cdc42, linear regression analyses 397

A. Amoah et al.

of Cdc42 activity and Cdc42 polarity were performed on human HSC. The data revealed that indeed Cdc42 activity and the frequency of polar HSC were inversely correlated (Figure 4E), suggesting a causative link between Cdc42 activity and polarity also for human HSC.

for Cdc42, though not fully to the frequency seen in young HSC (Figure 4C). This data confirms that elevated activity of Cdc42 is, at least in part, causative for the reduced polarity of aged human HSC similar to aged murine HSC.8,9,32

Aged hematopoietic stem cells are repolarized by pharmacological inhibition of Cdc42 activity

CASIN treatment induces early division onset in aged hematopoietic stem cells

CASIN (Cdc42 activity specific inhibitor) is a pharmacological compound that specifically inhibits the activity of Cdc42.30,31 We thus tested whether inhibition of Cdc42 activity in aged human HSC via CASIN might result in an increase in the frequency of polar HSC. CASIN indeed increased the frequency of HSC polar for Cdc42 in a dose-dependent fashion (Figure 4F) but not for tubulin (Online Supplementary Figure S5C). Inhibition of Cdc42 activity, thus, increases the frequency of aged HSC polar

We previously demonstrated that HSC from aged donors were delayed in division onset in comparison to HSC isolated from the young (Figure 2B and C), hence we assessed the effect CASIN treatment would have on the division of aged HSC. Our results show that upon CASIN treatment, aged HSC commence division hours before untreated HSC (Figure 4G) with no additional effect on the second division (Online Supplementary Figure S6D to E). We observed that the time taken for 50% of untreated

Figure 4. Hematopoietic stem cells polarity assessment and division kinetics in the presence and absence of CASIN. (A) Illustration of a polar and apolar cell. (B) Representative immunofluorescent images taken with a confocal microscope. Quantification of the proportion of young (white) and aged (grey) donor cells polar for (C) Cdc42 and (D) tubulin. ***P=0.00018, **P=0.002; t-test, false discovery rate approach. Bars represent the mean ± standard error of the mean (SEM). Scale bar represents 2 mm; n >5; n >8. Donor age: young =26-39 years (yr), median =28 yr; aged =58-82 yr, median =65 yr. (E) Correlation analysis (Pearson) of Cdc42 polarity and relative Cdc42 activity (r=0.6, P=0.03, n =13), broken grey lines represent 95% Confidence Interval. (F) Quantification of Cdc42 polarity of hematopoietic stem cells (HSC) after treatment with CASIN. *P=0.03; Mann-Whitney test. Bars represen the mean ± SD. n >2. Donor age: aged =61-81 years (yr), median = 62 yr. (G) Cumulative first division of aged HSC with and without CASIN and (H) time by which 50% of the cells have undergone the first division. Curve was derived from cumulative gaussian fits with robust regression. n and n =4. Donor age: aged =63-71 yr, median =66 yr. young

aged

398

aged+CASIN

haematologica | 2022; 107(2)

Cdc42 and aging of human HSC

HSC to undergo first division was considerably longer than cells treated with 5 mM or 10 mM CASIN albeit not statistically significant (P=0.06, Figure 4H) suggesting the inhibition of Cdc42 activity may facilitate or support early division onset.

Murine xenotransplant models that are used to test the potential of human HSC usually require irradiation of the recipient animal for establishing human engraftment.17 Irradiation of the recipients though might contribute to variable secondary effects not linked to the potential of the transplanted HSC. In order to circumvent the necessity for irradiation of the recipients in xenotransplants, we utilized NBSGW animals as recipients33 to assess the function of young, aged and CASIN-treated aged human HSC in vivo. To this end, 500 HSC were injected via the tail vein into mice and BM chimerism was analyzed at week 8 and 12 post xenotransplantation. Mice were considered to be successfully engrafted when the level of human hematopoietic cells (huCD45+) detected was higher than 0.1% of total CD45+ cells.34 In general, transplants with human BM-

derived HSC result in a much lower level of chimerism in comparison to transplants with CB-derived HSC (data not shown). Animals transplanted with aged human HSC showed a significantly higher level of engraftment (mean 3.7%) compared to recipients that received young HSC (mean 0.6%) and the aged + CASIN groups (mean 1.6%, 2.1%) (Figure 5A). Chimerism driven by aged HSC further increased at 12 weeks post-transplantation (mean 5%) compared to chimerism driven by young HSC (Figure 5B). This result was surprising, as the small number of previous studies in which aged human cells were transplanted into irradiated recipients demonstrated a lower16 or at least similar chimerism15 stemming from aged compared to young human HSC. In order to test whether irradiation of recipients might influence the reconstitution potential of aged HSC, we also transplanted young and aged human HSC into irradiated (1.6 Gy) NBSGW mice. Recipients of aged HSC displayed similar levels of chimerism compared to recipients of young HSC, and only marginal differences in their differentiation profile (Online Supplementary Figure S6A to C). Irradiation of recipients might thus influence the reconstitution potential of aged human HSC in xenotransplantation experiments. In summary, our results demonstrate that aged HSC show an overall elevated contribution to BM upon transplantation into NBSGW animals.

Aged hematopoietic stem cells show an elevated level of chimerism but balanced differentiation in NBSGW xenotransplants

Figure 5. Analysis of xenotransplantation experiment under non-irradiated settings. Frequency of human cells in the bone marrow (BM) of NBSGW mice at (A) 8 weeks and (B) 12 weeks post-transplant. (C) Emergence of human myeloid cells and (D) B cells post-transplant from young (blue), aged (green) and CASIN-treated aged (orange, brown) -recipients. Red line represents averages of non-transplanted controls. Bars represent the mean "±" standard deviation (SD). n = 6 different donors per cohort, 46 mice at 8 weeks and 42 mice at 12 weeks. Donor age: young =24-35 years (yr), median =27 yr; aged =60-81 yr, median =62 yr.

haematologica | 2022; 107(2)

399

A. Amoah et al.

Aged HSC treated with CASIN at 5 mM or 10 mM were more similar in their contribution to chimerism to young HSC at 8 weeks post transplantation. Aged HSC treated with the higher concentration of CASIN, 10 mM, were able to sustain a significantly lower, young-like contribution to chimerism at 12 weeks post transplantation. This mirrors our finding that aged HSC, treated with 10 mM of CASIN, showed the highest increase in the repolarization of aged HSC (Figure 4F). Assessment of the frequency of human myeloid (CD33+) and B cells (CD19+) in the BM of recipient mice revealed no significant differences among any of the experimental groups (Figure 5C and D) implying that aged human HSC, at least in NBSGW animals, show no differentiation skewing as reported for aged murine HSC. Colony forming unit assays also revealed the myeloerythroid potential of aged and CASIN-treated aged HSC were preserved (Online Supplementary Figure S6F to H) which is in line with the report by Pang et al.16 In summary, our xenotransplantation experiments demonstrate that, to our surprise, aged human HSC are able to confer an elevated level of chimerism upon transplantation into NBSGW animals, which is likely a result of the elevated activity of Cdc42 in aged HSC, as aged HSC treated with 10 mM of CASIN are more similar to young HSC than to aged HSC.

Discussion In this study, we characterized aging-related changes in human HSC and tested an intervention to target these aging-related changes. Using a well-established marker panel for the identification of primitive hematopoietic cells in BM, we show that there is an increase in the HSC frequency and a decrease in lymphoid progenitors in BM of the sternum of the elderly. We specifically selected the low side scatter CD34+ cells as our reference population in accordance to the guidelines of the International Society of Hematotherapy and Graft Engineering (ISHAGE).35 Additionally, the gating of low side scatter CD34+ cells minimizes variation across labs thereby increasing reproducibility and reliability.35 It is a possibility that the disparities in previous reports on changes in the frequency of human HSC15,16 may result from the use of different reference populations. Our data also demonstrates a delayed entry of aged HSC into division, an observation similarly made by Flach but not us for aged murine HSC,8,36 but no overall delay in division kinetics once the cells start to divide. This finding is in contrast to observations by Pang et al.16 which concluded that aged HSC are more prone to divide. Here, Pang determined the proportion of G and non-G cells using Hoechst 33342 and Pyronin Y while we used Hoechst 33342 and Ki-67. While Pyronin Y correlates with RNA content, Ki-67 is seen as a marker for proliferation in itself. It is thus a possibility that upon aging the correlation between elevated RNA content and proliferation might be somewhat diminished. The molecular mechanisms of the delayed entry of aged human HSC into cell division upon stimulation might be linked to elevated Cdc42 activity as cells treated with a pharmacological inhibitor of Cdc42 activity (CASIN), proceeded into division ahead of their untreated counterparts. In mutant yeast, hyperactive levels of Cdc42 result in the formation of multiple budding sites, however in wild- type cells, foci 0

400

competition is quickly resolved, with only one axis maturing and recruiting other proteins such as Bem1, to facilitate division.37 It is therefore a possibility that also in aged human HSC, the correct allocation of proteins to distinct positions within cells that are critical for proper initiation of division, might be altered upon elevated activity of Cdc42, resulting in a delay. Knowledge on the molecular mechanisms underlying aging of HSC have supported the development of therapeutic approaches to mitigate aging of HSC. For example, CASIN shows great promise in rejuvenating aged HSC9 and also in reverting aging-associated immune remodeling (AAIR) in murine models.32 A multivariate analysis model of gene expression profiles and biological age identified Cdc42 as a strong predictor of survival and that a higher Cdc42 level is associated with higher mortality.38 We showed previously that the level of Cdc42 activity in human blood cells correlates positively with age39 and in this study, we identified a similar association in LDBM cells. Aged human HSC also show elevated activity of Cdc42 and a low frequency of HSC polar for polarity proteins. We also report a negative correlation between the activity of Cdc42 in individual donors and the frequency of HSC polar for polarity proteins like Cdc42 in that donor. We could demonstrate that the inhibition of the activity of Cdc42 is sufficient to increase the frequency of chronologically aged HSC polar for Cdc42 but not for tubulin which suggests in human cells, Cdc42 and tubulin axes may not be closely linked as shown in murine cells.9 Furthermore, the number of HSC in a given donor and the activity of Cdc42 in hematopoietic cells of that donor are positively correlated, which suggests that Cdc42 activity may affect the frequency of human HSC upon aging and could therefore, directly or indirectly, contribute to the increased HSC frequency observed. Given that in mice, apolar distribution of Cdc42 (a consequence of elevated activity of Cdc42) drives HSC to divide symmetrically,8 and that elevated activity of Cdc42 in human leukemia stem cells is linked to more symmetric divisions,40 we postulate here an increased frequency of symmetric divisions also for HSC of the elderly, which though will require further investigations. In order to assess changes in function upon aging, HSC were transplanted into NBSGW recipients. By employing this model, secondary elements that may be introduced due to irradiation were avoided.41 We noted higher chimerism levels in non-conditioned recipients of aged HSC whereas conditioned recipients showed no increase. These observations also corroborate findings that irradiation, even at low doses, induces functional changes in mesenchymal stromal cells which influences their hematopoietic-supporting ability.42 Our data suggests that the elevated Cdc42 activity in aged HSC drives the age-related high level of chimerism observed in non-irradiated NBSGW recipients since recipients of young and 10 mM CASINtreated aged HSC exhibited close similarities in the level of human chimerism. The higher level of chimerism driven by aged human HSC further supports a model in which human aged HSC might predominantly undergo symmetrical divisions. Another possibility remains that an elevated repopulation potential of aged human HSC could be driven by the presence of individual clones43 linked to aging-related clonal hematopoiesis reported for a certain percentage of the elderly.44 Most human studies though demonstrated clonal hematopoiesis only in peripheral blood and not haematologica | 2022; 107(2)

Cdc42 and aging of human HSC

among HSC, thus it remains unclear whether changes in clonality exist already among aged HSC.44–46 We did not determine in this study clonality among the small number of HSC transplanted into animals, as this remains technically very challenging. In summary, we identified novel age-related phenotypes of human HSC and provide evidence of inter-species parallels as well as differences to support translational studies in the aging field. Our data supports that age-related phenotypes that are indicators of the function of aged HSC (Cdc42 activity, polarity, reconstitution potential in xenografts) are malleable in human HSC by inhibition of the age-related elevated activity of Cdc42. This might therefore present a new possibility to improve autologous stem cell transplants of aged donors.

Contributions AA, HG was involved in study design, interpretation and manuscript writing; AA performed and analyzed experiments;

1. Kovtonyuk L V., Fritsch K, Feng X, Manz MG, Takizawa H. Inflamm-aging of hematopoiesis, hematopoietic stem cells, and the bone marrow microenvironment. Front Immunol. 2016;7:502. 2. Ponnappan S, Ponnappan U. Aging and immune function: molecular mechanisms to interventions. Antioxid Redox Signal. 2011;14(8):1551-1585. 3. De Haan G, Lazare SS. Aging of hematopoietic stem cells. Blood. 2018;131(5):479-487. 4. Geiger H, de Haan G, Florian MC. The ageing haematopoietic stem cell compartment. Nat Rev Immunol. 2013;13(5):376-389. 5. Chambers SM, Shaw CA, Gatza C, Fisk CJ, Donehower LA, Goodell MA. Aging hematopoietic stem cells decline in function and exhibit epigenetic dysregulation. PLoS Biol. 2007;5(8):1750-1762. 6. Liang Y, Van Zant G, Szilvassy SJ. Effects of aging on the homing and engraftment of murine hematopoietic stem and progenitor cells. Blood. 2005;106(4):1479-1487. 7. Beerman I, Seita J, Inlay MA, Weissman IL, Rossi DJ. Quiescent hematopoietic stem cells accumulate DNA damage during aging that is repaired upon entry into cell cycle. Cell Stem Cell. 2014;15(1):37-50. 8. Florian MC, Klose M, Sacma M, et al. Aging alters the epigenetic asymmetry of HSC division. PLoS Biol. 2018;16(9):1-35. 9. Florian MC, Dörr K, Niebel A, et al. Cdc42 activity regulates hematopoietic stem cell aging and rejuvenation. Cell Stem Cell. 2012;10(5):520-530. 10. Abkowitz JL, Golinelli D, Harrison DE, Guttorp P. In vivo kinetics of murine hemopoietic stem cells. Blood. 2000;96(10):33993405. 11. Catlin SN, Busque L, Gale RE, Guttorp P, Abkowitz JL. The replication rate of human hematopoietic stem cells in vivo. Blood. 2011;117(17):4460-4466. 12. Lee CCI, Fletcher MD, Tarantal AF. Effect of age on the frequency, cell cycle, and lineage maturation of rhesus monkey (Macaca mulatta) CD34+and hematopoietic progenitor cells. Pediatr Res. 2005;58(2):315-322.

haematologica | 2022; 107(2)

Acknowledgements We would like to thank the Flow Cytometry Core and Imaging Core Facilities and the Tierforschungszentrum at the University of Ulm for their support. We also thank Jeffrey Bailey for excellent training in the bone marrow aspiration technique, Aishlin Hassan and Kalpana Nattamai for their technical support at Cincinnati Children’s Hospital Medical Center (CCHMC). Funding This study was supported with funding from the DFG, GRK1789 (CEMMA) to HG and AA.

Disclosures No conflicts of interest to disclose.

References

RE, MK and AL provided aged samples; AK and KS performed experiments; KS, AV and KE assisted in cell sorting procedures; VS supported in transplantation and bone marrow aspiration; YZ and MCF assisted in study design; AL, RE, MK and MCF reviewed and edited the manuscript.

Data sharing statemen The data that support the findings of this study are available from the corresponding author upon reasonable request.

13. Geiger H, Rennebeck G, Van Zant G. Regulation of hematopoietic stem cell aging in vivo by a distinct genetic element. Proc Natl Acad Sci. 2005;102(14):510-5107. 14. Geiger H, True JM, De Haan G, Van Zant G. Age- and stage-specific regulation patterns in the hematopoietic stem cell hierarchy. Blood. 2001;98(10):2966-2972. 15. Kuranda K, Vargaftig J, de la Rochere P, et al. Age-related changes in human hematopoietic stem/progenitor cells. Aging Cell. 2011;10(3):542-546. 16. Pang WW, Price EA, Sahoo D, et al. Human bone marrow hematopoietic stem cells are increased in frequency and myeloid-biased with age. Proc Natl Acad Sci U S A. 2011;108(50):20012-20017. 17. Doulatov S, Notta F, Laurenti E, Dick JE. Hematopoiesis: a human perspective. Cell Stem Cell. 2012;10(2):120-136. 18. Cosgun KN, Rahmig S, Mende N, et al. Kit regulates HSC engraftment across the human-mouse species barrier. Cell Stem Cell. 2014;15(2):227-238. 19. McIntosh BE, Brown ME, Duffin BM, et al. Nonirradiated NOD,B6.SCID Il2rγ-/kitW41/W41(NBSGW) mice support multilineage engraftment of human hematopoietic cells. Stem Cell Rep. 2015;4(2):171-180. 20. Waskow C, Madan V, Bartels S, Costa C, Blasig R, Rodewald HR. Hematopoietic stem cell transplantation without irradiation. Nat Methods. 2009;6(4):267-269. 21. van Galen P, Kreso A, Wienholds E, et al. Reduced lymphoid lineage priming promotes human hematopoietic stem cell expansion. Cell Stem Cell. 2014;14(1):94106. 22. Althoff MJ, Nayak RC, Hegde S, et al. Yap1Scribble polarization is required for hematopoietic stem cell division and fate. Blood. 2020;136(16):1824-1836. 23. Beerman I, Maloney WJ, Weissmann IL, Rossi DJ. Stem cells and the aging hematopoietic system. Curr Opin Immunol. 2010;22(4):500-506. 24. Etienne-Manneville S, Hall A. Rho GTPases in cell biology. Nature. 2002;420(6916):629635. 25. Yang L, Zheng Y. Cdc42: A signal coordina-

tor in hematopoietic stem cell maintenance. Cell Cycle. 2007;6(12):1444-1449. 26. Gekas C, Graf T. CD41 expression marks myeloid-biased adult hematopoietic stem cells and increases with age. Blood. 2013;121(22):4463-4472. 27. Glauche I, Thielecke L, Roeder I. Cellular aging leads to functional heterogeneity of hematopoietic stem cells: a modeling perspective. Aging Cell. 2011;10(3):457-465. 28. Akunuru S, Geiger H. Aging, clonality, and rejuvenation of hematopoietic stem cells. Trends Mol Med. 2016;22(8):701-712. 29. Klose M, Florian MC, Gerbaulet A, Geiger H, Glauche I. Hematopoietic stem cell dynamics are regulated by progenitor demand: lessons from a quantitative modeling approach. Stem Cells. 2019;37(7):1-22. 30. Liu W, Du W, Shang X, et al. Rational identification of a Cdc42 inhibitor presents a new regimen for long-term hematopoietic stem cell mobilization. Leukemia. 2019;33(3):749761. 31. Du W, Liu W, Mizukawa B, et al. A nonmyeloablative conditioning approach for long-term engraftment of human and mouse hematopoietic stem cells. Leukemia. 2018;32 (9):2041-2046. 32. Leins H, Mulaw M, Eiwen K, et al. Aged murine hematopoietic stem cells drive agingassociated immune remodeling. Blood. 2018;132(6):565-576. 33. McIntosh BE, Brown ME, Duffin BM, et al. Nonirradiated NOD,B6.SCID Il2rγ-/kitW41/W41 (NBSGW) mice support multilineage engraftment of human hematopoietic cells. Stem Cell Rep. 2015;4(2):171-180. 34. Martin MG, Welch JS, Uy GL, et al. Limited engraftment of low-risk myelodysplastic syndrome cells in NOD/SCID gamma-C chain knockout mice. Leukemia. 2010;24(9): 1662-1664. 35. Sutherland DR, Anderson L, Keeney M, Nayar R, Chin-Yee I. The ISHAGE guidelines for CD34+ cell determination by flow cytometry. J Hematother. 1996;5(3):213-226. 36. Flach J, Bakker ST, Mohrin M, et al. Replication stress is a potent driver of functional decline in ageing haematopoietic stem cells. Nature. 2014;512(7513):198-202.

401

A. Amoah et al. 37. Witte K, Strickland D, Glotzer M. Cell cycle entry triggers a switch between two modes of Cdc42 activation during yeast polarization. Elife. 2017;6:e26722. 38. Kerber RA, O’Brien E, Cawthon RM. Gene expression profiles associated with aging and mortality in humans. Aging Cell. 2009;8(3):239-250. 39. Florian MC, Klenk J, Marka G, et al. Expression and activity of the small RhoGTPase Cdc42 in blood cells of older adults are associated with age and cardiovascular disease. J Gerontol Ser. A 2017;72(9):1196-1200. 40. Mizukawa B, O’Brien E, Moreira DC, et al.

402

The cell polarity determinant CDC42 controls division symmetry to block leukemia cell differentiation. Blood. 2017;130(11): 1336-1346. 41. Costa S, Reagan MR. Therapeutic irradiation: consequences for bone and bone marrow adipose tissue. Front Endocrinol (Lausanne). 2019;10:587. 42. Preciado S, Muntión S, Rico A, et al. Mesenchymal stromal cell irradiation interferes with the adipogenic/osteogenic differentiation balance and improves their hematopoietic-supporting ability. Biol Blood Marrow Transplant. 2018;24(3):443-451. 43. Belderbos ME, Jacobs S, Koster TK, et al.

Donor-to-donor heterogeneity in the clonal dynamics of transplanted human cord blood stem cells in murine xenografts. Biol Blood Marrow Transplant. 2020;26(1):16-25. 44. Jaiswal S, Natarajan P, Silver AJ, et al. Clonal hematopoiesis and risk of atherosclerotic cardiovascular disease. N Engl J Med. 2017;377 (2):111-121. 45. Jaiswal S, Fontanillas P, Flannick J, et al. Agerelated clonal hematopoiesis associated with adverse outcomes. N Engl J Med. 2014;371 (26):2488-2498. 46. Sano S, Wang Y, Walsh K. Clonal hematopoiesis and its impact on cardiovascular disease. Circ J. 2019;83(1):2-11.

haematologica | 2022; 107(2)

ARTICLE

Acute Myeloid Leukemia

APR-246 induces early cell death by ferroptosis in acute myeloid leukemia

Ferrata Storti Foundation

Rudy Birsen,1,2 Clement Larrue,3 Justine Decroocq,1,2 Natacha Johnson,1 Nathan Guiraud,4,5 Mathilde Gotanegre,4,5 Lilia Cantero-Aguilar,1 Eric Grignano,1 Tony Huynh,1 Michaela Fontenay,1,6 Olivier Kosmider,1,6 Patrick Mayeux,1 Nicolas Chapuis,1,6 Jean Emmanuel Sarry,4 Jerome Tamburini,1,2,3 and Didier Bouscary1,2 1

Université de Paris, Institut Cochin, CNRS UMR8104, INSERM U1016, Paris, France; Assistance Publique-Hôpitaux de Paris, Centre-Université de Paris, Service d’Hématologie Clinique, Hôpital Cochin, Paris, France; 3Translational Research Center in Onco-Hematology, Faculty of Medicine, University of Geneva, Geneva, Switzerland; 4 Centre de Recherches en Cancérologie de Toulouse, UMR1037, INSERM, Equipe Labellisée LIGUE 2018, Toulouse, France; 5Université de Toulouse, Institut National des Sciences Appliquées de Toulouse, INSERM, Toulouse, France and 6Assistance PubliqueHôpitaux de Paris, Centre-Université de Paris, Service d’Hématologie Biologique, Hôpital Cochin, Paris, France 2

Haematologica 2022 Volume 107(2):403-416

ABSTRACT

PR-246 is a promising new therapeutic agent that targets p53 mutated proteins in myelodysplastic syndromes and in acute myeloid leukemia (AML). APR-246 reactivates the transcriptional activity of p53 mutants by facilitating their binding to DNA target sites. Recent studies in solid cancers have found that APR-246 can also induce p53-independent cell death. In this study, we demonstrate that AML cell death occurring early after APR-246 exposure is suppressed by iron chelators, lipophilic antioxidants and inhibitors of lipid peroxidation, and correlates with the accumulation of markers of lipid peroxidation, thus fulfilling the definition of ferroptosis, a recently described cell death process. The capacity of AML cells to detoxify lipid peroxides by increasing their cystine uptake to maintain major antioxidant molecule glutathione biosynthesis after exposure to APR-246 may be a key determinant of sensitivity to this compound. The association of APR-246 with induction of ferroptosis (either by pharmacological compounds, or genetic inactivation of SLC7A11 or GPX4) had a synergistic effect on the promotion of cell death, both in vivo and ex vivo.

Introduction Acute myeloid leukemias (AML) are highly heterogeneous diseases with a constant activation of oncogenic signaling.1 Recent years have witnessed major breakthroughs in their treatment with the approval of midostaurin, venetoclax and IDH mutant inhibitors.2-5 However, AML has a poor prognosis and there is still an urgent need for new treatments. APR-246, also known as PRIMA-1MET, is a promising new therapeutic agent that targets TP53 mutated cancers.6-8 This compound is being evaluated in AML and myelodysplastic syndromes (MDS) with TP53 mutations and appears to be highly effective against this poor prognosis disease.8-11 Mechanistically, APR-246 is converted to a reactive product (methylene quinuclidinone, MQ) that reacts with nucleophiles and thus alkylates thiol groups in proteins.12 APR-246 reactivates the transcriptional activity of p53 mutants by facilitating their binding to DNA target sites. Specific cysteines in the core domain of mutant p53 proteins are critical targets for their reactivation by APR-246/MQ.13 APR-246 also triggers p53-independent cell death mechanisms.14,15 Accordingly, using esophageal cancer as a model, it has been shown that APR-246 causes a decrease in glutathione (GSH) content resulting in an increased amount of reactive oxygen species (ROS) and of lipid peroxides in particular.16 In this report, we investigated the mechanisms of cell death induced by APR-246 in AML and we demonstrated that early cell death in AML is due to ferroptosis induction.

haematologica | 2022; 107(2)

Correspondence: RUDY BIRSEN rudy.birsen@inserm.fr DIDIER BOUSCARY didier.bouscary@aphp.fr Received: May 18, 2020. Accepted: December 28, 2020. Pre-published: January 7, 2021. https://doi.org/10.3324/haematol.2020.259531

403

R. Birsen et al.

Methods Cell lines and reagents HL60, MOLM14, SET2, MV4-11, OCI-AML2, OCI-AML3, K562, THP1, UT7-EPO, SKM1, NB4 and KASUMI AML cell lines were used. Patients provided written informed consent in accordance with the Declaration of Helsinki. Bone marrow (BM) samples were obtained from five patients with newly diagnosed AML (characteristics provided in the Online Supplementary Table S1). Cells were cultured in RPMI with glutamine (Gibco61870, Life Technologies® Saint Aubin, France) supplemented with 10% fetal bovine serum (FBS) and 4 mM glutamine. All AML cell lines were certified using their microsatellite identity (characteristics provided in Online Supplementary Table S2). Ferrostatin-1, necrostatin-1, necrostatin-1S, necrosulfonamide, QVD-OPH, APR-246 for the in vitro study, erastin and RSL3 were sourced from Selleckchem (Houston, TX). Chloroquine and doxycycline were obtained from Sigma–Aldrich (Saint-Louis, MO). FINO2 was purchased from Cayman Chemicals (Ann Arbor, MI). The APR-246 reagent used in the in vivo study was provided by APREA therapeutics (Solna, Sweden).

Constructs Inducible short hairpin RNA (shRNA) targeting SLC7A11 or GPX4 were constructed as previously described17 using the following sequences: SLC7A11#2, GCTGAATTGGGAACAACTATA; SLC7A11#3, GCAGTTGCTGGGCTGATTTAT; GPX4#1, GTGAGGCAAGACCGAAGTAAA; GPX4#2, CTACAACGTCAAATTCGATAT.

Lentivirus production and acute myeloid leukemia cell line infection 293-T packaging cells were used to produce lentiviral constructs through co-transfection with plasmids encoding lentiviral proteins. Supernatants were collected and ultracentrifuged for 48 hours (h) after transfection over two consecutive days and subsequently stored at -80°C. AML cell lines were plated at 2x106/mL and 2-10 mL aliquots of lentiviral supernatants were added for 3 h. Cells were then grown in 10% fetal calf serum medium and further selected with puromycin. For shRNA induction, 200 mg/mL doxycycline was added to the culture medium.

Flow cytometry-based assay Data acquisition and data analysis were conducted at the Cochin Cytometry and Immunobiology Facility. For glutathione measurements using monochlorobimane (MCB; Thermofischer; Waltham, MA), 2x105 cells were labeled with 40 mM MCB in 1 mL of warm complete medium for 20 minutes (min) in a tissue culture incubator (37°C, 5% CO2) in the dark. The reaction was terminated using 1 mL of cold complete medium, followed by centrifugation (200xg, 1 min). The pelleted cells were then resuspended in 0.5 mL of cold complete medium and placed on ice in the dark until analysis by flow cytometry (FCM). The MCBGSH signal was detected using a 355 nm laser through a 450/50 nm filter. FCM data were collected using a BD Fortessa flow cytometer with DIVA software. 10,000 events were recorded for analysis. Data analysis was then carried out with KALUZA software. For lipid peroxide production measurements using C11BODIPY (581/591) (2 mM) (Thermofischer, Waltham, MA), 2x105 cells were labeled with C11-BODIPY in 1 mL of warm complete medium for 10 min in a tissue culture incubator (37°C, 5% CO2) in the dark. Cells were then washed twice and resuspended in 200 mL of fresh PBS. For cystine uptake measurements using BioTracker Cystine-FITC Live Cell Dye (5 mM) (Thermofischer, Waltham, MA), 2x105 cells were labeled with

404

Biotracker cystine in 1 mL of warm complete medium for 120 minutes in a tissue culture incubator (37°C, 5% CO2) in the dark. FCM data were collected using a C6 Accuri flow cytometer (Becton Dickinson, Le Pont de Claix, France) with CFlow Plus software. 10,000 events were captured for subsequent analysis with CFlow Plus software (Becton Dickinson, Le Pont de Claix, France).

Western blotting Whole-cell extraction and western blotting were performed as previously described.17 Anti-GPX4 antibody was purchased from Proteintech (Manchester, UK). Anti-PARP, caspase 8, caspase 3, cleaved caspase 3, MLKL, pMLKL, p53 and SLC7A11 antibodies were sourced from Cell Signaling Technology (Danvers, MA, USA).

Viability assay AML cells were plated at 20x104/mL in 100 µl of 10% FBSsupplemented RPMI prior to the addition of compounds. Cells were cultured in the presence of the test compounds for 24 to 48 h at 37°C. Viability was quantified using the fluorescence based Uptiblue assay (Interchim, Montluçon, France). Uptiblue was added to each well in 10 mL aliquots. Fluorescence was then measured with a Typhoon FLA9500 scanner (GE Healthcare; IL). Fluorescence values were normalized to dimethyl sulfoxide (DMSO)-treated controls for each AML cell line. Half maximal inhibitory concentration (IC50) values were calculated using a four parameter non-linear regression curve with Graph Pad Prism v8 (GraphPad, La Jolla, CA, USA). For primary AML cells, viability was assayed by FCM analysis using forward scatter (FSC) versus side scatter (SSC).

Measurement of synergistic effects Cell viability was calculated for every dose combination of APR-246 and ferroptosis inducer using the Synergy Finder webtool (https://synergyfinder.fimm.fi/) and compared to each agent alone. Calculations were based on the ZIP model.18

Tumor xenografts in NOD/SCID IL-2 receptor g-chain-null mice Xenograft tumors were generated by randomly injecting 1×106 MOLM14 shCTRL or shSLC7A11 cells into the tail veins of NOD/SCID IL-2 receptor g-chain-null mice (NSG) aged 6–9 weeks. Fourteen days after injection, doxycycline (200 mg/mL) and sucrose (1% weight:volume) were added to the drinking water of these animals. After 3 days, the mice were randomly treated with a daily intraperitoneal injection of APR-246 (100 mg/kg) or vehicle (phosphate-buffered saline [PBS]) for 4 days. All experiments were conducted in accordance with the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care International. Animals were used in accordance with a protocol reviewed and approved by the Institutional Animal Care and Use Committee of Région MidiPyrénées (France). BM (mixed from tibias and femurs) were dissected and flushed in Hanks balanced salt solution with 1% FBS. MNC from BM were labeled with PE-conjugated anti-hCD33, PerCP-Cy5.5-conjugated anti-mCD45.1 and APC-conjugated anti-hCD45 (all antibodies from Becton Dickinson, BD) to determine the fraction of human blasts (hCD45+mCD45.1−hCD33+ cells) using FCM. Acquisition of data was performed on a CytoFLEX (Beckman Coulter) flow cytometer with CytExpert software. The number of AML cells in the BM was determined using CountBright beads (Invitrogen, CA, USA) in accordance with the manufacturer’s protocol. Data analysis was performed with flowJo software.

haematologica | 2022; 107(2)

APR-246 induced ferroptosis in AM

Figure 1. APR-246 induces cell death in acute myeloid leukemia cells irrespective of their TP53 mutational status. (A) Viability curves for the indicated cells at 24 hours (h) post APR-246 treatment. Error bars, ± standard deviation. (B) Viability curves for the indicated primary acute myeloid leukemia (AML) cells at 24 h post APR-246 treatment. Patient characteristics are provided in the Online Supplementary Table S1. (C) Half maximal inhibitory concentration percentage (IC50%) of APR246 treatment for 24 h across a panel of AML cell lines based on cell viability (n=3). In our subsequent experiments, we selected five AML cell lines sensitive to APR246 in these concentration ranges, with or without TP53 mutations (highlighted in bold font). wt: wild-type; fs: frameshift; del: deletion.

haematologica | 2022; 107(2)

405

R. Birsen et al.

Figure 2. Legend on following page.

406

haematologica | 2022; 107(2)

APR-246 induced ferroptosis in AM

Figure 2. APR-246 induces ferroptosis in acute myeloid leukemia cells. (A) Cell viability (%) for the indicated cells at 16 hours (h) post-APR-246 treatment (60 mM) with or without ferrostatin-1 (10 mM), deferoxamine (DFO) (100 mM), necrostatin-1 (20 mM), chloroquine (20 mM) or QVD-OPH (25 mM) (n=3). Error bars ± standard error of the mean [SEM]. All compounds were added 2 h (h) prior to APR-246 in the medium. Statistics, 2-way ANOVA; *P<0.05, **P<0.01, ***P<0.0001. (B) Immunoblotting analysis of PARP, caspase 8 and caspase 3 in MOLM-14 cells treated for 16 h with dimethyl sulfoxide (DMSO), APR-246 (60 µM) or puromycin (1 mg/mL). b-actin was used as a loading control (n=2). (C) Viability curves for the indicated cells at 16 h post APR-246 treatment with or without ferrostatin-1 (10 mM), DFO (100 mM), necrostatin-1 (20 mM), chloroquine (20 mM) or QVD-OPH (25 mM) (n=3). Error bars ± SEM. (d) Cell death (%) of the indicated cells at 16 h and 24 h post-APR-246 treatment (50 mM) with or without ferrostatin-1 (10 mM) (n=3). Error bars ± standard deviation.

Figure 3. APR-246 induces ferroptosis in acute myeloid leukemia cells. (A) Electron microscopy analysis of MOLM-14 cells treated with or without APR-246 (60 mM, H16). The white arrowhead indicates a mitochondrion showing membrane rupture and reduced cristae. (B and C) Detection of lipid peroxidation using C11-BODIPY and flow cytometry (FCM) at 14 hours post APR-246 treatment in acute myeloid leukemia (AML) cell lines (B) and in primary AML cells (C). APR-246 was used at a 100 mM concentration for MOLM-14 and 50 µM for other AML cell lines. Left panels show representative FCM quantification (n=3). Error bars ± standard deviation.

haematologica | 2022; 107(2)

407

R. Birsen et al.

Figure 4. APR-246 induces glutathione depletion in acute myeloid leukemia cells. (A) Summary of the cellular pathways involved in ferroptosis. Ferroptosis execution is triggered by an iron-catalyzed excessive peroxidation of polyunsaturated fatty acids (PUFA)-containing phospholipids (PL-PUFA). Glutathione (GSH) and glutathione peroxidase 4 (GPX4) are the two key elements controlling the elimination of lipid peroxides. Solute carrier family 7 member 11 (SLC7A11) encodes the transporter subunit of the heterodimeric cystine-glutamate antiporter named system xc-. System xc- mediates cystine entry into the cell in exchange for glutamate.26 Once inside the cell, cystine is rapidly reduced to cysteine which is the limiting amino acid for GSH synthesis. SLC7A11 inhibition results in cellular cysteine depletion, which leads to the exhaustion of intracellular pool of GSH. GPX4 is a pleiotropic selenoprotein that uses GSH to selectively reduce lipid hydroperoxides to lipid alcohols, in order to protect the cells against membrane lipid peroxidation.29 GPX4 inhibition is either due to its direct inhibition or downregulation, or to GSH depletion via direct or indirect processes. The inhibition of GPX4 results in uncontrolled polyunsaturated fatty acid phospholipid (PL-PUFA) oxidation and fatty acid radical generation, leading to ferroptotic cell death. ACSL4: acyl-CoA synthetase long chain family member 4; LPCAT 3: lysophosphatidylcholine acyltransferase 3; ALOX: arachidonate lipoxygenase; PUFA: polyunsaturated fatty acid; PL: phospholipids; PE: phosphatidylethanolamine; GPX4: glutathione peroxidase 4. (B) GSH (mBCI) measurement in acute myeloid leukemia (AML) cell lines by flow cytometry (FCM) at 14 hours (h) post APR-246 and fer1 (10 mM) treatment. APR-246 was used at 100 mM for MOLM-14 and 50 mM for other AML cell lines. Fer1 was associated to prevent cell death and allowed analysis of GSH depletion. Left panels show representative FCM quantification. n=3. Error bars + standard deviation. Statistics by t-test. *P<0.05, **P<0.01, ***P<0.0001. (C and D) Cell death (%) (C) and GSH (mBCI) measurement (D) for MOLM-14 at 24 h post-APR-246 treatment (60 mM) with or without B-ME (50 mM), cysteine (50 mM) or Fer1 (10 mM). Error bars + standard deviation. Statistics by t-test; *P<0.05, **P<0.01, ***P<0.0001.

408

haematologica | 2022; 107(2)

APR-246 induced ferroptosis in AM

Results APR-246 induces ferroptosis in acute myeloid leukemia cells In order to determine the activity of APR-246 in AML, we assayed a set of 12 AML cell lines carrying diverse and representative molecular abnormalities, and five primary AML samples (Online Supplementary Table S1). Most of the AML cell lines and primary AML cells were sensitive to cell death induction by APR-246 (Figure 1A and B). The IC50 of APR-246 at 24 h for each cell line ranged from 11 mM to more than 200 mM, independently of their TP53 mutational status (Figure 1C). Previous in vivo human data have shown that the plasma concentrations of APR-246 in the 12 h following its intravenous administration range from 50 to 500 mM, suggesting that concentrations above 50 mM are suitable for in vitro studies of the early effects of APR-246.19 In our subsequent experiments, we selected five AML cell lines sensitive to APR-246 in these concentration ranges, with or without TP53 mutations. In order to investigate the mechanisms underlying APR-246 activity against AML cells, we exposed the cells treated with APR-246 to inhibitors of various cell death pathways (Figure 2A). The decrease in cell viability was almost completely rescued by either iron chelation via deferoxamine (DFO) or by the lipophilic antioxidant ferrostatin-1 (Fer1), demonstrating that cell death is both iron and reactive- oxygen species (ROS) dependent. Necrostatin-1 (Nec1) also consistently prevented cell death induced by APR-246 after short-term incubation in all AML cells lines. Nec1 has been used to define necroptotic cell death, but can also protect against ferroptosis through a target which is as yet unknown.20 Induction of necroptosis after treatment with APR-246 in our cells was excluded due to the absence of protection by more specific necroptosis inhibitors (Necrostatin 1s and Necrosulfonamide),21,22 and due to the absence of MLKL phosphorylation, a key marker in necroptosis21 (Online Supplementary Figure S1A and B). Inhibitors of autophagy (chloroquine) or apoptosis (QVD-OPH) failed to block APR-246-induced cell death. We confirmed that the mechanism of APR-246 induced cell death is distinct from apoptosis, as evidenced by the absence of caspases 3/8 or PARP1 cleavage, including in TP53 mutated AML cell lines (Figure 2B; Online Supplementary Figure S2A to C). Notably, the protection against cell death observed with Fer1, DFO or Nec1 was partially lost at higher doses of APR-246 (Figure 2C) and at later time points (Figure 2D). However, apoptosis did not appear to be the mechanism of this late death, since caspases 3/8 and PARP1 were not cleaved and QVD-OPH was still unable to prevent the cell death (Online Supplementary Figure S2A to E). Examination of the ultrastructural changes induced by APR-246 treatment did not reveal any characteristic features of apoptosis (i.e., no plasma membrane blebbing, chromatin condensation or nuclear fragmentation) or autophagy (absence of autophagolysosomes). Necrotic cells were rare, and some mitochondria showed membrane rupture and reduced cristae (Figure 3A; Online Supplementary Figure S3). We analyzed the levels of endogenous lipid peroxidation – a hallmark of ferroptosis induction – following APR-246 treatment by flow cytometry analysis with C11-BODIPY staining. We observed a high induction of lipid ROS in AML cell lines and primary AML cells from two patients treated with APR-246 ex haematologica | 2022; 107(2)

vivo. This increase in lipid ROS was fully reversed by Fer1, indicating that lipid peroxides had been newly generated (Figure 3B and C). All these results allowed to conclude that APR-246 induces early cell death by ferroptosis, a recently described non-apoptotic form of regulated cell death that links together membrane lipid peroxidation, cysteine and iron metabolism, glutathione peroxidase activity and oxidative stress (as summarized in Figure 4A23–25). As previously reported,14,16 we observed that APR-246 treatment induced a dramatic decrease in GSH levels in AML cell lines (Figure 4B). Cysteine is the main biosynthetic precursor of GSH. Cysteine can be transported into cells via membrane transporters for neutral amino acids. However, in the extracellular space, cysteine is rapidly reduced to cystine. Thus the main source of intracellular cysteine comes from the entry of cystine into the cell via system xc-.26 b-mercaptoethanol (b-ME) can promote cystine uptake through an alternative pathway.27 b-ME was able to completely rescue the cell death and GSH depletion induced by APR-246 (Figure 4C and D). Cysteine treatment showed similar results as Fer1 treatment (Figure 4C and D). Altogether, these results demonstrate that APR-246 induces GSH depletion which induces ferroptosis in AML cells irrespective of their TP53 mutational status, and that ferroptosis induction is the main mechanism of cell death after early exposure to APR-246.

Cystine uptake determined the sensitivity of acute myeloid leukemia cells to APR-246 Since cysteine is a biosynthetic precursor of GSH, we asked whether the ability of cells to provide cysteine for GSH synthesis underlies the sensitivity to APR-246. Using FITC-labeled cystine, we showed that after exposure to APR-246, AML cells increased their uptake of cystine from the extracellular space (Figure 5A). Western blot analysis of the protein levels of SLC7A11 showed an increased amount of SLC7A11 (Figure 5B). This suggests that the cells enhanced cystine uptake by increasing SLC7A11 protein levels to maintain intracellular GSH levels after exposure to APR-246. We modulated the cystine uptake through SLC7A11 overexpression or inhibition. SLC7A11 overexpression in MOLM-14 and OCI-AML2 (Online Supplementary Figure S4) decreased the cell death and prevented the depletion of GSH induced by APR-246 (Figure 5C and D). Then we showed that targeting the SLC7A11 cystine transporter by RNA interference reduced the basal uptake of cystine (Online Supplementary Figure S4B and C) and had a very little effect on cell death (Online Supplementary Figure S4D), but strongly reduced cell proliferation in AML cells in vitro (Online Supplementary Figure S4E). Inhibition of SLC7A11 with RNA interference increased cell death and viability impairment, GSH depletion, and the accumulation of lipid peroxides induced by APR-246 (Figure 6A to C; Online Supplementary Figure S4F). Interestingly, basal GSH levels were not consistently affected by b−ME addition, cysteine addition, or the overexpression or inhibition of SLC7A11, which suggests that the amount of GSH in basal conditions is not a reliable marker of cell cystine uptake ability (Figures 4D, 5D and 6C). Finally, targeting SLC7A11 with erastin, a potent inhibitor of system xc-28 that showed variable sensitivity in our AML cell lines (Online Supplementary Figure S4G), had synergistic activity with APR-246 both on cell death and on cell viability 409

R. Birsen et al.

impairment (Figure 6D and E; Online Supplementary Figure S5). The association of erastin and APR-246 also had a synergistic effect on cell viability in five primary AML samples (Figure 6F). There was no correlation between the basal levels of GPX4 and SLC7A11 proteins and the sensitivity to APR-246 (Online Supplementary Figure S6A and B). Altogether, these data suggest that the ability of AML cells to prevent lipid peroxides accumulation by increasing their cystine uptake to support GSH after

exposure to APR-246 is a key determinant of the sensitivity to this compound.

The association of APR-246 and ferroptosis inducers has a synergistic anti-leukemic activity in vitro We next determined whether targeting ferroptosis pathways in combination with APR-246 can increase the antileukemic activity of this compound, mimicking SLC7A11 inhibition. Downregulation of GPX4 by RNA interference

Figure 5. SLC7A11 overexpression prevents glutathione depletion and cell death following APR-246 exposure. (A) Cystine uptake in MOLM-14 and OCI-AML2 cells lines at 16 hours (h) post-APR-246 (100 mM) and Fer1 (10 mM) treatment. Fer1 was associated to prevent cell death and allowed analysis of cystine uptake. (B) Immunoblotting analysis of SLC7A11 in MOLM-14 cells treated for 16 h with dimethyl sulfoxide (DMSO) or APR-246 (n=2). b-actin was used as a loading control. (C and D) Cell death (%) (C) and glutathione (GSH) (mBCI) measurement (D) of the indicated cells at 20 h post-APR-246 treatment (n=3). For GSH measurement, Fer1 was associated to prevent cell death and allowed analysis of GSH depletion. Error bars ± standard deviation. Statistics by t-test. *P<0.05, **P<0.01, ***P<0.0001.

410

haematologica | 2022; 107(2)

APR-246 induced ferroptosis in AM

Figure 6. Legend on following page.

haematologica | 2022; 107(2)

411

R. Birsen et al.

Figure 6. SLC7A11 inhibition sensitizes cells to APR-246. (A) Viability curves for MOLM-14 and OCI-AML2 with small hairpin RNA control (shSCR) or doxycyclineinducible shRNA (shSLC7A11) cells at 16 hours (h) post APR-246 treatment. Prior to adding APR-246, the cells were treated for 3 days with doxycycline (n=3). Error bars ± standard deviation (SD). (B and C) Cell death (%) (B) and glutathione (GSH) (mBCI) measurement (C) of the indicated cells at 20 h post-APR-246 (MOLM-14 30 mM, OCI-AML2 10 mM) treatment (n=3). For GSH measurement, Fer1 was associated to prevent cell death and allowed analysis of GSH depletion. Prior to adding APR-246, the cells were treated for 3 days with doxycycline (n=3). Error bars ± SD. Statistics by t-test. *P<0.05, **P<0.01, ***P<0.0001. (D) Cell death (%) for the indicated cell types at 24 h post APR-246 (MOLM-14 30 mM, OCI-AML2 10 µM) and or erastin (MOLM-14 100 nM, OCI-AML2 1 mM). Error bars ± SD. Statistics by t-test; *P<0.05, **P<0.01. (E) Illustrative synergy map (left panel) of 24 h co-treatment of MOLM-14 cells with APR-246 and erastin. The mean cell viability of three independent experiments was used. Mean synergy scores of the most synergistic area of 24 h co-treatment of AML cell lines with APR-246 and erastin (n=3). (F) Mean synergy score of the 48 h co-treatment of primary AML cells with APR-246 and erastin (n=1).

resulted in cell death (Online Supplementary Figure S7A and B). Our AML cell lines panel showed variable sensitivity to two ferroptosis-inducing drugs: RSL3, a direct GPX4 inhibitor;29 and FINO2, an indirect GPX4 inhibitor and direct iron oxidant30 (Online Supplementary Figure S8). We observed that knockdown of GPX4 increased the impairment of cell viability induced by APR-246 (Figure 7A). RSL3 and FINO2 in association with APR-246 synergistically decreased cell viability in AML cell lines (Figure 7B and C; Online Supplementary Figures S9 and 10). Collectively, these results show that pharmacological or genetic activation of ferroptosis enhances the antileukemic activity of APR-246 in AML.

Genetic invalidation of SLC7A11 has synergistic anti-leukemic activity with APR-246 in vivo We then examined whether inhibition of GSH synthesis through SLC7A11 inhibition could interfere with AML persistence and could enhance APR-246 activity in vivo. We engrafted MOLM14 cells transduced with either control (shSCR) or anti-SLC7A11 (shSLC7A11) doxycyclineinducible shRNA (Figure 8A). After induction of shRNA expression in vivo, we treated the mice with a 4-day APR246 regimen in order to mimic the therapeutic schedule used in clinical trials of APR-246.16 This treatment scheme varied from those previously published in mice in terms of treatment duration. Indeed, in these studies, APR-246 was administrated for 7 to 28 days, and reduction of tumor volume after 4 days of treatment was minimal at best.7,8,14,16 However, our aim was not to assess the efficacy of APR-246 alone but to demonstrate that its association with SLC7A11 inhibition enhanced its anti-leukemic activity. As expected, APR-246 alone did not reduce BM tumor cell burden. SLC7A11 knockdown significantly reduced tumor cell burden in the BM (Figure 8B and C). Moreover, the decrease in BM tumor cell burden was enhanced when APR-246 treatment was combined to SLC7A11 knockdown (Figure 8B and C). Overall, these results showed that inhibition of anti-ferroptosis mechanisms enhanced the anti-leukemic activity of APR-246 in vivo.

Discussion APR-246 can restore the wild-type conformation of mutant p53 protein, therefore inducing apoptosis and inhibition of tumor growth in mice.6 Thus, APR-246 is one of the most promising compounds in clinical development for TP53 mutated cancers. Controversies exist over the TP53 mutation status dependencies of APR-246.31 Some studies reported that APR-246 acts independently of its ability to reactivate mutant p53 protein.12,32–37 Tessoulin et al. demonstrated that myeloma cells are highly sensitive to APR-246, independently of their TP53 status.14 In this 412

cancer, APR-246 induces cell death by impairing GSH/ROS balance and acts synergistically with L-buthionine sulphoximine to inhibit myeloma growth in vivo.14 In ARID1A-deficient cancers, GSH was the major target of APR-246 and was the basis of the high sensitivity of these cancer cells to this coumpound.38 In esophageal cancer, Liu et al. showed that mutants p53 bind to the antioxidant transcription factor NRF2, leading to a decreased expression of SLC7A11 which sensitizes cells to GSH depletion by APR-246.16 Paradoxically, while APR-246 clinical development is the most advanced in AML with TP53 mutation, the effects of APR-246 have been little studied in this disease. Two studies showed that APR-246 induced in vitro cell death in a large number of leukemic cells from patients, alone or in association with chemotherapies.39,40 In both studies, the cytotoxicity of APR-246 was independent of the TP53 mutational status. The mechanisms of action of APR-246 was investigated in AML cell lines with TP53 mutations, and more specifically studied its association with 5-azacytidine which is currently used in clinical trials.8 It was shown that in TP53-mutated myelodysplastic syndromes (MDS) and AML, APR-246 can reactivate the p53 pathway and induce an apoptotic transcriptional program, with synergistic effects of APR-246 and azacytidine. In this context, our study strongly showed that APR-246 induced cell death in AML cells irrespective of their TP53 mutational status. APR-246 depleted intracellular GSH and induced lipid peroxide production, which led to ferroptosis induction. The ability of AML cells to detoxify lipid peroxides primed their sensitivity to APR-246 treatment. Additionally, we uncovered that inhibition of antiferroptosis mechanisms enhanced the anti-leukemic activity of APR-246 both in vitro and in vivo. We confirm the TP53 independence and GSH depletion and we demonstrated that APR-246 induces ferroptosis.14,16 Ferroptosis is rapidly induced after GPX4 inactivation and cell death occurs in the first 24 hours post-treatment or administration.23,29 The observation that protection against cell death by ferroptosis inhibitors decreases after 24 hours of exposure suggests that other cell death mechanisms might be involved after this early phase and that they may have masked the earlier induction of ferroptosis. The effect of APR-246 might be also different in AML cells in comparison to solid cancers. Our study might have several important implications for the management of MDS and AML patients. First, since APR-246 acts independently of TP53 mutational status, this treatment could be used in a broader panel of AML patients. Future study will need to identify predictive elements of the sensitivity of AML to APR-246 and the induction of ferroptosis. The mechanism of action of APR-246, which is based on GSH depletion and induction of ferroptosis, makes it the first ferroptosis-inducing agent currently used therapeutically in humans. Using ferroptohaematologica | 2022; 107(2)

APR-246 induced ferroptosis in AM

Figure 7. The combination of APR-246 with ferroptosis inducers has synergistic anti-leukemic effects in acute myeloid leukemia in vitro. (A) Viability curves for MOLM-14 and OCI-AML2 cells with or without GPX4 inducible small hairpin RNA (shRNA) at 16 hours (h) post APR-246 treatment. Prior to adding APR-246, the cells were treated for 2 days with doxycycline (n=3). Error bars ± standard deviation. (B) Illustrative synergy map (left panel) of 24 h co-treatment of MOLM-14 cells with APR-246 and RSL3. The mean cell viability of three independent experiments was used. Mean synergy scores of the most synergistic area of 24 h co-treatment of acute myeloid leukemia (AML) cell lines with APR-246 and RSL3 (n=3).(C) Illustrative synergy map (left panel) of 24 h co-treatment of MOLM-14 cells with APR-246 and FINO2. The mean cell viability of three independent experiments was used. Mean synergy scores of the most synergistic area of 24 h co-treatment of AML cell lines with APR-246 and FINO2 (n=3).

haematologica | 2022; 107(2)

413

R. Birsen et al.

Figure 8. The combination of APR-246 with SLC7A11 inhibition has synergistic anti-leukemic effects in acute myeloid leukemia in vivo. (A) Design of the in vivo experiment. (B) Representative flow cytometry analysis of bone marrow cells marked with mCD45.1 and hCD45 from each treatment subgroup. (C) Viable mCD45.1hCD45+ hCD33+ cell counts in the bone marrow of the different treatment subgroups. Bars represent mean of all experiments and errors denote ± standard deviations. Statistics by t-test; *P<0.05, **P<0.01, ***P<0.0001.

414

haematologica | 2022; 107(2)

APR-246 induced ferroptosis in AM

sis induction to treat cancer is an emerging field in oncology research. Renal cancer cells have been reported as highly dependent on the GSH pathway for ROS detoxification, including lipid peroxides, and targeting components of this pathway is an effective strategy for the treatment of this disease.41 Other studies have elegantly highlighted the higher sensitivity to ferroptosis of cancer cells that are resistant to conventional therapy.42,43 In AML, data about ferroptosis are scarce. An in vitro study showed that the ferroptosis inducer erastin enhances sensitivity of AML cells to chemotherapeutic agents.44 Jones et al. recently reported that cysteine depletion leads to GSH exhaustion and ROS-low leukemic stem cell eradication in AML.45 Thus, APR-246 could act on these cell pools that are poorly sensitive to conventional therapy, and which are at the origin of frequent therapeutic failures in AML. From a clinical perspective, this mechanism of action might be relevant. Indeed, iron chelators are frequently used for the treatment of iron overload due to red blood cell transfusions and dyserythropoïesis in MDS/AML patients. Several studies have reported beneficial effects of iron chelation therapy on overall survival in MDS patients with iron overload.46 However, iron chelators are recognized as canonical ferroptosis inhibitors. Therefore, caution should be exercised regarding co-administration of iron chelators which may antagonize the anti-leukemic activity of APR-246, as we observed in vitro. Moreover, the phase II studies of APR-246 in MDS/AML reported the occurrence of neurological adverse events in over a third of patients treated with APR-246.10,11 Recently, ferroptosis has been implicated in the pathogenesis of several neurological disorders, especially neurodegenerative disorders.47,48 One hypothesis could be that the neurological side effects observed after administration of APR-246 are linked to its ability to deplete GSH in neuronal cells. Consequently, anti-ferroptosis agents, such as iron chelators or vitamin E, could be valuable drugs to treat these side effects. Finally, our study highlights that ferroptosis

References 1. Döhner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017;129(4):424-447. 2. Stone RM, Mandrekar SJ, Sanford BL, et al. Midostaurin plus chemotherapy for acute myeloid leukemia with a FLT3 Mutation. N Engl J Med. 2017;377(5):454-464. 3. Stein EM, DiNardo CD, Pollyea DA, et al. Enasidenib in mutant-IDH2 relapsed or refractory acute myeloid leukemia. Blood. 2017;130(6):722-731. 4. DiNardo CD, Stein EM, de Botton S, et al. Durable remissions with Ivosidenib in IDH1-mutated relapsed or refractory AML. N Engl J Med. 2018;378(25):2386-2398. 5. DiNardo CD, Pratz K, Pullarkat V, et al. Venetoclax combined with decitabine or azacitidine in treatment-naive, elderly patients with acute myeloid leukemia. Blood. 2019;133(1):7-17. 6. Bykov VJN, Issaeva N, Shilov A, et al. Restoration of the tumor suppressor function to mutant p53 by a low-molecular-weight compound. Nat Med. 2002;8(3):282-288. 7. Saha MN, Jiang H, Yang Y, Reece D, Chang

haematologica | 2022; 107(2)

induction may represent a new target in AML, opening new therapeutic strategies based on disease-specific vulnerabilities. The effect of ferroptosis induction-based treatments on normal hematopoietic cells and their value compared to standard-of-care AML therapies will be important to evaluate in the future. Disclosures No conflicts of interest to disclose. Contributions JD, LCA, EG, TH performed experiments; CL, JES, JT, NG and MG performed in vivo experiments; NC, OK and MF provided AML samples; NC, PM, JES, NJ and JT analyzed the results and corrected the manuscript; RB performed experiments, analyzed data, and wrote the manuscript; DB designed and supervised the research program, analyzed data, and wrote the manuscript. All authors approved the final version of the manuscript. Acknowledgments We thank Alain Schmitt and the cell imagery department at the Cochin Institute for performing the transmission electron microscopy. We thank Tata Jojo for manuscript proofreading. We also thank the CYBIO cytometry-department at the Cochin Institute. We further thank Aprea Therapeutics for providing the APR-246 used in the in vivo study. Funding This work was supported by grants from the Association de Recherche Contre le Cancer (ARC; aides doctorales RB, grant n°DOC20170505807 and DOC20190508975; aides jeune chercheur TH, grant n°M2R20180507379), from the Institut National du Cancer (JD, grant n° ASC16046KSA), from the Ligue Nationale Contre le Cancer (LNCC; DB, Equipe Labellisée EL2017. N° Projet: ELFUZ17337; NG, grant n° IP/SCG/JD-16129) and from association Laurette Fugain (grant n°ALF2018/02).

H. PRIMA-1Met/APR-246 displays high antitumor activity in multiple myeloma by induction of p73 and Noxa. Mol Cancer Ther. 2013;12(11):2331-2341. 8. Maslah N, Salomao N, Drevon L, et al. Synergistic effects of PRIMA-1Met (APR246) and Azacitidine in TP53-mutated myelodysplastic syndromes and acute myeloid leukemia. Haematologica. 2020; 105(6):1539-1551. 9. Nahi H, Merup M, Lehmann S, et al. PRIMA-1 induces apoptosis in acute myeloid leukaemia cells with p53 gene deletion. Br J Haematol. 2006;132(2):230236. 10. Sallman DA, DeZern AE, Garcia-Manero G, et al. Phase 2 results of APR-246 and Azacitidine (AZA) in patients with TP53 mutant myelodysplastic syndromes (MDS) and oligoblastic acute myeloid leukemia (AML). Blood. 2019;134(Suppl 1):S676. 11. Cluzeau T, Sebert M, Rahmé R, et al. APR246 combined with Azacitidine (AZA) in TP53 mutated myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML). A phase 2 study by the Groupe Francophone Des Myélodysplasies (GFM). Blood. 2019;134(Suppl 1):S677. 12. Lambert JMR, Gorzov P, Veprintsev DB, et al. PRIMA-1 reactivates mutant p53 by

covalent binding to the core domain. Cancer Cell. 2009;15(5):376-388. 13. Zhang Q, Bykov VJN, Wiman KG, Zawacka-Pankau J. APR-246 reactivates mutant p53 by targeting cysteines 124 and 277. Cell Death Dis. 2018;9(5):1-12. 14. Tessoulin B, Descamps G, Moreau P, et al. PRIMA-1Met induces myeloma cell death independent of p53 by impairing the GSH/ROS balance. Blood. 2014; 124(10):1626-1636. 15. Bykov VJN, Zhang Q, Zhang M, Ceder S, Abrahmsen L, Wiman KG. Targeting of mutant p53 and the cellular redox balance by APR-246 as a strategy for efficient cancer therapy. Front Oncol. 2016;6:21. 16. Liu DS, Duong CP, Haupt S, et al. Inhibiting the system xC-/glutathione axis selectively targets cancers with mutant-p53 accumulation. Nat Commun. 2017;8:14844. 17. Jacque N, Ronchetti AM, Larrue C, et al. Targeting glutaminolysis has antileukemic activity in acute myeloid leukemia and synergizes with BCL-2 inhibition. Blood. 2015;126(11):1346-1356. 18. Ianevski A, He L, Aittokallio T, Tang J. SynergyFinder: a web application for analyzing drug combination dose-response matrix data. Bioinformatics. 2017;33(15): 2413-2415.

415

R. Birsen et al. 19. Lehmann S, Bykov VJN, Ali D, et al. Targeting p53 in vivo: a first-in-human study with p53-targeting compound APR246 in refractory hematologic malignancies and prostate cancer. J Clin Oncol. 2012;30(29):3633-3639. 20. Friedmann Angeli JP, Schneider M, Proneth B, et al. Inactivation of the ferroptosis regulator Gpx4 triggers acute renal failure in mice. Nat Cell Biol. 2014;16(12):1180-1191. 21. Sun L, Wang H, Wang Z, et al. Mixed lineage kinase domain-like protein mediates necrosis signaling downstream of RIP3 kinase. Cell. 2012;148(1-2):213-227. 22. Takahashi N, Duprez L, Grootjans S, et al. Necrostatin-1 analogues: critical issues on the specificity, activity and in vivo use in experimental disease models. Cell Death Dis. 2012;3(11):e437. 23. Dixon SJ, Lemberg KM, Lamprecht MR, et al. Ferroptosis: an iron-dependent form of nonapoptotic cell death. Cell. 2012; 149(5):1060-1072. 24. Stockwell BR, Friedmann Angeli JP, Bayir H, et al. Ferroptosis: a regulated cell death Nexus linking metabolism, redox biology, and disease. Cell. 2017;171(2):273-285. 25. Feng H, Stockwell BR. Unsolved mysteries: How does lipid peroxidation cause ferroptosis? PLoS Biol. 2018;16(5):e2006203. 26. Koppula P, Zhang Y, Zhuang L, Gan B. Amino acid transporter SLC7A11/xCT at the crossroads of regulating redox homeostasis and nutrient dependency of cancer. Cancer Commun (Lond). 2018;38(1):12. 27. Ishii T, Bannai S, Sugita Y. Mechanism of growth stimulation of L1210 cells by 2mercaptoethanol in vitro. Role of the mixed disulfide of 2-mercaptoethanol and cysteine. J Biol Chem. 1981;256(23):1238712392. 28. Dixon SJ, Patel DN, Welsch M, et al. Pharmacological inhibition of cystine-glutamate exchange induces endoplasmic reticulum stress and ferroptosis. Elife. 2014;3:e02523. 29. Yang WS, SriRamaratnam R, Welsch ME, et al. Regulation of ferroptotic cancer cell

416

death by GPX4. Cell. 2014;156(1):317-331. 30. Gaschler MM, Andia AA, Liu H, et al. FINO2 initiates ferroptosis through GPX4 inactivation and iron oxidation. Nat Chem Biol. 2018;14(5):507-515. 31. Perdrix A, Najem A, Saussez S, et al. PRIMA-1 and PRIMA-1Met (APR-246): from mutant/wild type p53 reactivation to unexpected mechanisms underlying their potent anti-tumor effect in combinatorial therapies. Cancers (Basel). 2017;9(12):172. 32. Bykov VJN, Zache N, Stridh H, et al. PRIMA-1(MET) synergizes with cisplatin to induce tumor cell apoptosis. Oncogene. 2005;24(21):3484-3491. 33. Lambert JMR, Moshfegh A, Hainaut P, Wiman KG, Bykov VJN. Mutant p53 reactivation by PRIMA-1 MET induces multiple signaling pathways converging on apoptosis. Oncogene. 2010;29(9):1329-1338. 34. Duan W, Gao L, Wu X, et al. MicroRNA34a is an important component of PRIMA1-induced apoptotic network in human lung cancer cells. Int J Cancer. 2010;127 (2):313-320. 35. Roh J-L, Kang SK, Minn I, Califano JA, Sidransky D, Koch WM. p53-Reactivating small molecules induce apoptosis and enhance chemotherapeutic cytotoxicity in head and neck squamous cell carcinoma. Oral Oncol. 2011;47(1):8-15. 36. Izetti P, Hautefeuille A, Abujamra AL, et al. PRIMA-1, a mutant p53 reactivator, induces apoptosis and enhances chemotherapeutic cytotoxicity in pancreatic cancer cell lines. Invest New Drugs. 2014;32(5):783-794. 37. Liu DSH, Read M, Cullinane C, et al. APR246 potently inhibits tumour growth and overcomes chemoresistance in preclinical models of oesophageal adenocarcinoma. Gut. 2015;64(10):1506-1516. 38. Ogiwara H, Takahashi K, Sasaki M, et al. Targeting the vulnerability of glutathione metabolism in ARID1A-deficient cancers. Cancer Cell. 2019;35(2):177-190.e8. 39. Ali D, Jönsson-Videsäter K, Deneberg S, et al. APR-246 exhibits anti-leukemic activity

and synergism with conventional chemotherapeutic drugs in acute myeloid leukemia cells. Eur J Haematol. 2011;86 (3):206-215. 40. Nahi H, Lehmann S, Mollgard L, et al. Effects of PRIMA-1 on chronic lymphocytic leukaemia cells with and without hemizygous p53 deletion. Br J Haematol. 2004; 127(3):285-291. 41. Miess H, Dankworth B, Gouw AM, et al. The glutathione redox system is essential to prevent ferroptosis caused by impaired lipid metabolism in clear cell renal cell carcinoma. Oncogene. 2018;37(40):5435-5450. 42. Tsoi J, Robert L, Paraiso K, et al. Multi-stage differentiation defines melanoma subtypes with differential vulnerability to druginduced iron-dependent oxidative stress. Cancer Cell. 2018;33(5):890-904.e5. 43. Zou Y, Palte MJ, Deik AA, et al. A GPX4dependent cancer cell state underlies the clear-cell morphology and confers sensitivity to ferroptosis. Nat Commun. 2019;10(1):1617. 44. Yu Y, Xie Y, Cao L, et al. The ferroptosis inducer erastin enhances sensitivity of acute myeloid leukemia cells to chemotherapeutic agents. Mol Cell Oncol. 2015; 2(4):e1054549. 45. Jones CL, Stevens BM, D’Alessandro A, et al. Cysteine depletion targets leukemia stem cells through inhibition of electron transport complex II. Blood. 2019;134(4): 389-394. 46. Hoeks M, Yu G, Langemeijer S, et al. Impact of treatment with iron chelation therapy in patients with lower-risk myelodysplastic syndromes participating in the European MDS registry. Haematologica. 2020;105(3):640-651. 47. Wu J-R, Tuo Q-Z, Lei P. Ferroptosis, a recent defined form of critical cell death in neurological disorders. J Mol Neurosci. 2018;66(2):197-206. 48. Cardoso BR, Hare DJ, Bush AI, Roberts BR. Glutathione peroxidase 4: a new player in neurodegeneration? Mol Psychiatry. 2017; 22(3):328-335.

haematologica | 2022; 107(2)

ARTICLE

Acute Myeloid Leukemia

RXRA DT448/9PP generates a dominant active variant capable of inducing maturation in acute myeloid leukemia cells

Ferrata Storti Foundation

Orsola di Martino,1* Margaret A. Ferris,2* Gayla Hadwiger,1 Soyi Sarkar,1 Anh Vu,1 María P. Menéndez-Gutiérrez,3 Mercedes Ricote3 and John S. Welch Department of Internal Medicine, Washington University, St Louis, MO, USA; Department of Pediatrics, Washington University, St Louis, MO, USA and 3Myocardial Pathophysiology Area, Centro Nacional de Investigaciones Cardiovasculares (CNIC), Madrid, Spain

1 2

*OdM and MAF contributed equally as co-first authors.

Haematologica 2022 Volume 107(2):417-426

ABSTRACT

ARA and RXRA contribute to myeloid maturation in both mice and humans, and deletion of Rxra and Rxrb augments leukemic growth in mice. While defining the domains of RXRA that are required for anti-leukemic effects in murine KMT2A-MLLT3 leukemia cells, we unexpectedly identified RXRA DT448/9PP as a constitutively active variant capable of inducing maturation and loss of their proliferative phenotype. RXRA DT448/9PP was associated with ligand-independent activity in reporter assays, with enhanced co-activator interactions, reduced engraftment in vivo, and activation of myeloid maturation transcriptional signatures that overlapped with those of cells treated with the potent RXRA agonist bexarotene, suggestive of constitutive activity that leads to leukemic maturation. Phenotypes of RXRA DT448/9PP appear to differ from those of two other RXRA mutations with forms of constitutive activity (F318A and S427F), in that DT448/9PP activity was resistant to mutations at critical ligand-interacting amino acids (R316A/L326A) and was resistant to pharmacological antagonists, suggesting it may be ligand-independent. These data provide further evidence that activated retinoid X receptors can regulate myeloid maturation and provide a novel constitutively active variant that may be germane for broader studies of retinoid X receptors in other settings.

Correspondence:

Introduction

JOHN S. WELCH jwelch@wustl.edu

Retinoid receptors are highly conserved transcription factors that direct hematopoietic self-renewal and differentiation.1,2 Retinoids (vitamin A metabolites) bind directly to retinoid receptors, converting the receptors from transcriptional repressors to transcriptional activators. There are six types of retinoid receptor (RARA, RARB, RARG, RXRA, RXRB, RXRG) with distinct tissue expression and subtle differences in their binding and response to different ligands. The retinoid receptors RARA and RXRA undergo remarkable upregulation during myeloid maturation in both mice and humans, whereas RXRG is not detected.3-5 Retinoid treatments, both in vitro and in vivo, facilitate hematopoietic stem cell maturation and lineage commitment.1,2 Until now, the clinical application of retinoids in hematology has been restricted to acute promyelocytic leukemia (treated with all-trans retinoic acid, or tretinoin, a pan-RAR ligand) and cutaneous T-cell leukemia (treated with bexarotene, a pan-RXR ligand). However, recent studies have observed activity of all-trans retinoic acid when combined with chemotherapy in acute myeloid leukemias other than the promyelocytic form, and the RARA super-enhancer has emerged as a potential biomarker of retinoid sensitivity.6-10 To better understand the molecular determinants of anti-leukemic retinoid activity, we evaluated a series of RXRA truncations and mutations to determine which might rescue phenotypes observed in Rxra/Rxrb-deficient leukemia cells.5 We unexpectedly identified RXRA DT448/9PP as a constitutively active variant capable of inducing maturation and loss of proliferative capacity in leukemia cells. In this study, we characterized the activity of this variant using cell culture assays, transcription reporter assays,

haematologica | 2022; 107(2)

Received: February 17, 2021. Accepted: May 7, 2021. Pre-published: June 17, 2021. https://doi.org/10.3324/haematol.2021.278603

417

O. di Martino et al.

in vivo engraftment, and RNA sequencing, and present evidence of ligand-independent loss of proliferative capacity, transcriptional signatures of myeloid maturation that overlap with signatures induced by the potent RXR ligand bexarotene, and inhibition of engraftment and leukemic expansion in vivo. These data again suggest that programs of leukemic cell growth and maturation may be susceptible to retinoids and provide a novel constitutively active tool for further delineating the function and activity of retinoid receptors.

package edgeR was used to determine genes that were differentially expressed between groups: RXRA WT cells, RXRA WT cells treated with bexarotene , and RXRA DT448/9PP-transduced cells. Differentially expressed genes were defined as having an absolute log fold change >2 and a P-value <0.0001. Functional pathway analysis was performed using Panther Gene Ontology Analysis software14 with a Fisher exact test using the Bonferroni correction for multiple testing. Pathways were called if P<0.05 and fold enrichment was >2. Gene set enrichment analysis was performed using gene sets curated in the MSigDB database and GSEA 4.0.3 software; RRID:SCR_003199.15

Methods

Study approval

Mice

All animal procedures were approved by the Institutional Animal Care and Use Committee of Washington University.

Mice were maintained in a specific pathogen-free barrier facility with a 12 h light-dark cycle. Upon weaning, all mice were housed in groups of up to five mice of the same sex per cage. Food and a water bottle were provided in a recess of the metal wire lid at the top of the cage. Cages were changed once every week. Six- to tenweek-old mice (C57Bl/6 background) were typically used for the experiments. Equal numbers of male and female mice were used; no gender biases were noted.

Hematopoietic cell culture Murine bone marrow Kit+ cells were isolated using an Automacs Pro (Miltenyl Biotec, San Diego, CA, USA) according to the manufacturer’s protocol. Kit+ cells were plated in progenitor expansion medium (RPMI 1640 medium, 15% fetal bovine serum, stem factor [50 ng/mL], interleukin 3 [10 ng/mL], Flt3L [25 ng/mL], thrombopoietin [10 ng/mL], L-glutamine [2 mM], sodium pyruvate [1 mM], HEPES buffer [10 mM], penicillin/streptomycin [100 units/mL], b-mercaptoethanol [50 mM]) overnight and transduced with MSCV-KMT2A-MLLT3 retrovirus by spinfection with 10 µg/mL polybrene and 10 mM HEPES at 2400 rpm, 30°C for 90 min in an Eppendorf 5810R centrifuge. Cells were transplanted into sublethally irradiated mice and subsequent leukemia harvested 4 to 6 months later. KMT2A-MLLT3 leukemia cells were cultured in vitro using similar media, but without Flt3L or thrombopoietin. KMT2A-MLLT3-RXR-knockout (KO) leukemia cells were derived from Mx1-Cre x Rxraflox/flox x Rxrbflox/flox bone marrow cells11,12 transduced with MSCV-KMT2A-MLLT3 retrovirus and generated as described above. RXR deletion was induced by injecting Mx1-Cre x Rxraflox/flox x Rxrbflox/flox mice intraperitoneally with pIpC 300 mg/mouse; four doses were given, every other day. RXR deletion was confirmed by polymerase chain reaction analysis 4 weeks after mice had been treated with pIpC. THP1, and K562 cells were obtained from the American Type Culture Collection. Monomac6 and OCI-AML3 cells were obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen.

RNA sequencing KMT2A-MLLT3 leukemia cells were transduced with MSCVRXRA-IRES-mCherry or MSCV-RXRA DT448/9PP-IRES-mCherry and treated with or without 250 nM bexarotene. After 24 h, mCherry+ cells were sorted and total RNA was extracted with TRIzol LS (Ambion Life Technologies) and isolated on PureLink RNA Kit columns (Thermo Fisher Scientific). The quality of RNA was measured on a 2100 Bioanalyzer (Agilent). Sequencing libraries, each with individual Illumina indexes, were constructed using the TruSeq Stranded mRNA procedure (Sample Prep Kit v2; Illumina). Libraries were sequenced as paired-end 151 bp reads on an Illumina NovaSeq instrument. Reads were aligned to the mm10 mouse reference genome, and transcription was quantified using kallisto version 0.43.1 13 and Ensembl transcripts version 95. The R

418

Results Effects of RXRA DT448/9PP on maturation We recently evaluated a series of RXRA truncations and mutations to map the domains of RXRA required for antileukemic effects in murine KMT2A-MLLT3 (formerly MLLAF9) leukemia cells derived from Mx-Cre x Rxraflox/flox x Rxrbflox/flox bone marrow cells (RXR-KO KMT2A-MLLT3 leukemia).5 Ghosh et al. had previously observed that RXRA DT448/9PP was associated with increased binding to an NCOR2 co-repressor (nuclear receptor corepressor 2, formerly SMRT) peptide in a GST pull-down assay.16 We included this variant as part of the screen to evaluate corepressor requirements. DT448/9 is positioned at the end of the turn between helix 11 and 12. Helix 11 directly binds to co-repressors and co-activators.17 Helix 12 (the AF2 domain) adopts an extended and unstructured conformation in the absence of ligand and assumes a repositioned and helical conformation following ligand binding (Figure 1A).18 Unexpectedly, retroviral expression of RXRA DT448/9PP (MSCV-RXRA DT448/9PP-IRES-mCherry) in RXR-KO KMT2A-MLLT3 leukemia cells induced the normally round cells to form adherent clumps that were mCherry+ on the tissue culture plate (Figure 1B). mCherry+ cells were also associated with loss of leukemic colony-forming capacity, lack of proliferation, and macrophage-like cytomorphology with ruffled borders and vacuoles (Figure 1C-E). These effects were absent in RXR-KO KMT2A-MLLT3 cells retrovirally transduced with RXRA WT (MSCV-RXRA-IRESmCherry), with other tested RXRA domain deletions or point mutations, or with treatment combinations of RXRA and RARA activating ligands5 (Figure 1B). In addition, we found that retroviral expression of RXRA DT448/9PP (MSCV-RXRA DT448/9PP-IRES-mCherry) in KMT2AMLLT3 WT leukemia cells induced the same phenotype as that observed in RXR-KO KMT2A-MLLT3 leukemia cells (lack of proliferation and macrophage-like cytomorphology) (data not shown). We examined the resulting proteins on western blot after transduction into KMT2A-MLLT3 WT cells (Figures 1F). RXRA DT448/99PP was associated with a similar sized band as wild-type RXRA, but it appeared resistant to the proteolytic cleavage that results in the formation of a truncated 36 kDa fragment.19 Endogenous Rxra levels were significantly lower than levels of retrovirally expressed RXRA (Figure 1G). Thus, retroviral transduction led to similar levels of expression of RXRA WT and RXRA DT448/9PP, at supraphysiological protein levels compared to endogenous RXRA. haematologica | 2022; 107(2)

RXRA active mutation

Figure 1. Monocytic maturation induced by RXRA DT448/9PP. (A) DT448/9 amino acids are highlighted in red within the structure of the RXRA ligand binding domain (PDB 4K4J). An Ncoa2 peptide is highlighted in orange. Note that DT448/9 occurs at the beginning of helix 12, the AF2 domain, and the rigid proline substitution could enhance helix formation. (B, C) RXR-KO KMT2A-MLLT3 leukemia cells were transduced with MSCV-RXRA DT448/9PP-IRES-mCherry or with MSCV-RXRA WTIRES-mCherry. Cells were evaluated under fluorescent and light microscopy at 72 h. (B) Transduced cells were sorted for mCherry+ cells and plated in methylcellulose, and colonies assessed in technical triplicates on day 7. Statistical significance evaluated using the t-test, ***P<0.001 (C). (D) RXR-KO KMT2A-MLLT3 leukemia cells were transduced with MSCV-RXRA DT448/9PP-IRES-mCherry, stained with FxCycle violet, and retention of the dye was assessed by flow cytometry at the indicated time points comparing mCherry+ (RXRA DT 448/9 PP) versus mCherry– (RXR-KO) cells. (E) Cytomorphology of mCherry+ (RXRA DT 448/9 PP) versus mCherry– (RXRA WT endogenous) sorted KMT2A-MLLT3 leukemia cells after transduction with MSCV-RXRA DT448/9PP-IRES-mCherry. (F) RXR-KO KMT2A-MLLT3 leukemia cells were transduced with MSCV-RXRA DT448/9PP-IRES-mCherry or with MSCV-RXRA-IRES-mCherry and protein expression was evaluated through western blot analysis using anti-RXRA antibody (H-10, Santa Cruz). GAPDH was used as a loading control. (G) KMT2A-MLLT3 WT leukemia cells were transduced with MSCV-RXRA WT-IRESmCherry as indicated, and protein expression was evaluated through western blot analysis using anti-RXRA antibody (5388, Cell Signaling). GAPDH was used as a loading control. (H) KMT2A-MLLT3 WT leukemia cells were transduced as indicated and analyzed by flow cytometry. RXRA WT cells were analyzed with and without treatment with 250 nM bexarotene for 24 h. RAW 264.7 cells were used as a positive staining control, but not included in the statistical comparisons. Statistical significance was evaluated using analysis of variance with the Tukey correction for multiple comparisons, *P<0.05. **P<0.01. ***P<0.001.

haematologica | 2022; 107(2)

419

O. di Martino et al.

Because the RXRA DT448/9PP-transduced cells acquired cell adhesion properties and ruffled borders suggestive of macrophage-like maturation, we assessed immunophenotypic changes. Bexarotene treatment of RXRA WT transduced cells resulted in an increased percentage of cells expressing F4/80, CD11b, CD14, and CD64, as determined by flow cytometry, and increased the median fluorescence intensity (MFI) of F4/80, CD11b, CD14, CD64, Ly6c, and Gr1, but had little effect on CD115 expression. RXRA DT448/9PP transduction resulted in increased percentages of cells expressing CD11b, CD14, and CD64, with increased MFI of CD11b, CD14, CD64, Ly6c and Gr1, consistent with myeloid and monocytic maturation, but lacking the classical macrophage marker CD115 (Figure 1H, Online Supplementary Figure S1). Transduction of RXRA DT448/9PP, but not RXRA WT into two KMT2A-MLLT3-associated human cells (THP1 and MonoMac-6) also resulted in adherence to plastic with the development of podocytes and ruffled borders (Online Supplementary Figure S2). These effects required 4 to 6 days to emerge, whereas adhesion and cell clumping in murine KMT2A-MLLT3 cells typically required only 3 or 4 days. In contrast, Kit+ murine bone marrow cells cultured in stem cell media (stem cell factor, interleukin 3, thrombopoietin, Flt3L) and K562 cells were resistant to the effects of RXRA DT448/9PP, and the cells remained round and non-adherent (Online Supplementary Figures S2D and S3A, B).

RXRA DT448/9PP is constitutively active We assessed the functional activity of RXRA DT448/9PP using multiple reporter systems. First, we used a UAS/Gal4 reporter.11 KMT2A-MLLT3 WT leukemia cells derived from UAS-GFP bone marrow cells were retrovirally transduced with a fusion of the Gal4 DNA binding domain and the RXRA wild-type ligand-binding domain (Gal4-RXRA) or a fusion of the Gal4 DNA-binding domain and the RXRA DT448/9PP ligand-binding domain (Gal4-RXRA DT448/9PP). Gal4-RXRA DT448/9PP resulted in constitutive activity uninfluenced by increasing concentrations of bexarotene (Figure 2A). Constitutive activity of RXRA DT449/8PP was also observed in Kit+ UAS-GFP bone marrow cells transduced with Gal4-RXRA retroviruses (Online Supplementary Figure S3C). Second, we evaluated three different luciferase reporters in 293T cells. 293T cells are known to express endogenous RXRA as well as various of its partners; RARA, PPARG, and VDR.20 Reporters assays included a synthetic direct repeat 1 (DR1) peroxisome proliferator response element (PPRE); a DR1 response element from the ApoA1 promoter; and a DR5 response element from the RARB promoter. We noted that RXRA DT448/9PP transfection consistently led to ligand-independent activation of all three constructs (Figure 2B). A control RXRA construct contained a deletion of the AF2 domain (RXRA DAF2), and this construct is unresponsive to ligand. Using this approach we assessed whether RXRA DT448/9PP might be sensitive to inhibition by two pan-RXR antagonists, HX531 and UVI3003. Neither compound inhibited activation of the UAS/Gal4 reporter or the RARE-Luc reporter (Figure 2C). We used a mammalian two-hybrid assay to assess the interaction of RXRA DT448/9PP with the co-activator PGC1a, noting constitutive, ligand-independent binding (Figure 2D, E). This activity remained present with the L2/3A variant (L147A, L148A, L210A, L211A),21 which contains point mutations in the critical LXXLL motifs required 420

for ligand-dependent interactions of PGC1a with nuclear receptors. Thus, RXRA DT448/9PP appears to constitutively engage PGC1a using domains outside the canonical Nterminal LXXLL motifs.

Comparison with other active RXRA variants Two other RXRA variants with constitutively active properties have been described. First, mouse Rxra F318A exhibits constitutively active phenotypes.22 However, when a crystal structure was generated, the ligand-binding pocket contained oleic acid, and Rxra F318A activity could be inhibited by the pan-RXR antagonist HX531, suggesting that the mutation leads to augmented responsiveness to a natural ligand present in the tissue culture, and is not completely ligand-independent.23 We assessed the sensitivity of RXRA DT448/9PP to two pan-RXR antagonists (HX531 and UVI3003) across a series of assays. In RXR-WT KMT2A-MLLT3 leukemia cells transduced with RXRA DT448/9PP, cell adhesion, clumping, and loss of proliferation phenotypes were not abrogated by either compound (Online Supplementary Figure S4). To further assess whether RXRA DT448/9PP phenotypes may result from hyper-responsiveness to intracellular natural ligands, we mutated two amino acids (R316 and L326) that form critical ionic bonds with the carboxylic acid group in ligands (e.g., bexarotene, 9-cis retinoic acid, and long-chain fatty acids).5 RXRA R316A/L326A has been previously shown to abrogate ligand-dependent activation and is unable to rescue response to bexarotene in RXR-KO KMT2A-MLLT3 leukemia cells.5,24 Retroviral expression of the compound variant RXRA R316A/L326A/DT448/9PP again led to overexpression on western blot and resistance to proteolytic cleavage, but did not abrogate cell clumping, loss of proliferation, or loss of colony formation (Figure 3A-D). Second, recurrent RXRA hot-spot mutations (S427F/Y) occur in patients with bladder cancer, and these augment the activity of the PPARG:RXRA heterodimer, but are not capable of independently activating RXRA reporters.25 Retroviral expression of RXRA S427F also led to strong overexpression of the variant, which retained sensitivity to proteolytic cleavage, although cell adhesion, cell clumping, cell proliferation, and colony-forming properties were not altered (Figure 3E-H). To determine whether similar heterodimerization with PPARG or with other nuclear receptors may play essential roles in the activity of RXRA DT448/9PP, we evaluated phenotypes using a series of receptor antagonists. Concurrent treatment with potent antagonists of retinoic acid receptors, LXR, PPARA, and PPARG did not affect clumping, adhesion to plastic, or proliferation phenotypes (Online Supplementary Figure S5), demonstrating that the DT448/9PP creates a nonpermissive receptor that is functionally active, independently of other activated nuclear receptors.

RXRA DT448/9PP inhibits leukemic engraftment and expansion KMT2A-MLLT3 WT cells were transduced with RXRA WT or RXRA DT448/9PP retroviruses labeled with IRESmCherry cassettes (Figure 4A). To limit maturation effects in vitro, the populations were immediately transplanted into recipient mice and residual cells were subsequently analyzed by flow cytometry. After 4 weeks of engraftment and expansion, leukemia cells in the peripheral blood were assessed and we observed that mice transplanted with haematologica | 2022; 107(2)

RXRA active mutation

RXRA WT displayed increased leukocytosis, an increase in neutrophils, and a decrease in lymphocytes compared to mice transplanted with RXRA DT448/9PP-transduced cells (Figure 4B-D). Compared with the pre-engraftment population (RXRA WT, 55% mCherry+; RXRA DT448/9PP, 62% mCherry+), at 4 weeks there was reduced engraftment in RXRA DT448/9PP cells relative to RXRA WT cells: RXRA WT average of 7.6% mCherry+ cells (range, 0.94%-16.4%) versus RXRA DT448/9PP 0.25% mCherry+ cells (range,

0.095%-0.36%) (Figure 4E). When we assessed moribund mice transplanted with cells transduced with RXRA WT, we observed further reduction in the absolute proportion of mCherry+ cells in the bone marrow rather than an expansion: RXRA WT, average 1% (range, 1.21%-1.46%); RXRA DT448/9PP, average 0.741 (range, 0.022%-1.46%) (Figure 4G). Survival was shorter in mice transplanted with RXRA WT-transduced cells than in those transplanted with RXRA DT448/9PP-transduced cells (Figure 4H). At the time of sac-

Figure 2. Constitutive activity of RXRA DT448/9PP. (A) KMT2A-MLLT3 WT leukemia cells derived from UAS-GFP bone marrow cells were transduced with MSCV-Gal4RXRA DT448/9PP-IRES-mCherry or MSCV-Gal4-RXRA-IRES-mCherry, treated with increasing concentrations of bexarotene, and the ratio of GFP+ cells (responding) to total mCherry+ cells (capable of responding) was determined. Data from biological duplicates are shown. (B, C) 293T cells were transfected with the indicated plasmids, treated with bexarotene, HX531, or UVI3003 (all 1 mM), and luciferase was measured after 40 h. Data from biological triplicates are shown. PPRE: peroxisome proliferator response element. ApoA1: DR1 element from the ApoA1 promoter. RARE: retinoic acid receptor response element from the RARB promoter. The ∆AF2 deletion acts as a negative control. (D, E) Mammalian two-hybrid. 293T cells were transfected with the indicated plasmids, treated with bexarotene and GFP was measured by flow cytometry after 40 h. Data from biological triplicates are shown. GFP-N1: a positive control CMV-GFP expression vector. GFP-N0: a negative control derived from GFP-N1 after the deletion of the CMV promoter. Statistical significance was evaluated using analysis of variance with the Bonferroni comparison to control, *P<0.05, **P<0.01, ***P<0.001.

haematologica | 2022; 107(2)

421

O. di Martino et al.

rifice, all but one mouse had developed splenomegaly (average spleen weight 0.5 g) and bone marrow lacked erythroid elements, features that are typical of overwhelming secondary leukemic engraftment (Online Supplementary Figure S6).

Transcriptional consequences of RXRA DT448/9PP To assess the maturation effects of RXRA DT448/9PP more comprehensively and to investigate the overlapping

consequences with RXRA WT activation by bexarotene, KMT2A-MLLT3 WT leukemia cells were transduced with RXRA WT or RXRA DT448/9PP and the transcriptional profiles of the transduced cells were assessed by RNA sequencing (Figure 5A). We noted significant overlap of regulated genes between RXRA WT cells treated with bexarotene and cells transduced with RXRA DT448/9PP (Figure 5B-D). Of 598 genes that were upregulated in RXRA WT cells treated

Figure 3. Comparison with other RXRA variants with activity. KMT2A-MLLT3 WT leukemia cells were transduced with MSCV-3xFlag-RXRA R316A/L326A/DT448/9PPIRES-mCherry or with MSCV-RXRA S427F-IRES-mCherry and evaluated under fluorescent and light microscopy at 72 h (A and E). RXR-KO KMT2A-MLLT3 leukemia cells were transduced with MSCV-3xFlag-RXRA R316A/L326A/DT448/9PP-IRES-mCherry or MSCV-RXRA S427F-IRES-mCherry, stained with FxCycle Violet, and retention of the dye was assessed by flow cytometry at the indicated time points (B and F). RXR-KO KMT2A-MLLT3 leukemia cells were transduced with MSCV-RXRA WT-IRES-mCherry or MSCV-3xFlag-RXRA R316A/L326A/DT448/9PP-IRES-mCherry or MSCV-RXRA S427F-IRES-mCherry, mCherry+ cells were then sorted and plated in methylcellulose, and colonies assessed in technical triplicates on day 7 (C and G). RXR-KO KMT2A-MLLT3 leukemia cells were transduced with MSCV-RXRA-IRESmCherry or MSCV-RXRA DT448/9PP-IRES-mCherry or MSCV-RXRA S427F-IRES-mCherry and protein expression was evaluated through western blot analysis using anti-RXRA antibody (H-10, Santa Cruz). GAPDH was used as a loading control (D and H). Statistical significance was evaluated using the t-test, ***P<0.001.

422

haematologica | 2022; 107(2)

RXRA active mutation

with bexarotene, 525 overlapped with those of RXRA DT448/9PP-transduced cells. Likewise, of 74 downregulated genes in RXRA WT cells treated with bexarotene, 28 overlapped with those in RXRA DT448/9PP-transduced cells. Gene ontology pathway analysis showed enrichment of lipid metabolism and immune system signaling and downregulation of RNA processing (Online Supplementary Tables S1 and S2). Gene set enrichment analysis of a curated list of genes regulated during myeloid development26 indicated that treatment of RXRA WT cells with bexarotene and expression of RXRA DT448/9PP both lead to a myeloid differentiation phenotype (upregulation of “Up” genes and

downregulation of “Down” genes) compared to untreated WT cells (Figure 5E, F). Within this set of differentially expressed genes, we evaluated transcripts associated with cell surface proteins. Sixty-seven cell surface transcripts were differentially expressed, according to RNA sequencing, between RXRA WT cells versus RXRA WT cells treated with bexarotene or RXRA DT448/9PP-transduced cells (Online Supplementary Figure S7A). As in the overall RNA sequencing analysis, these genes were associated with similar or greater upregulation in DT448/9PP-transduced cells than in the wild-type cells treated with bexarotene (Online Supplementary Figure

haematologica | 2022; 107(2)

Figure 4. In vivo engraftment with RXRA WT cells versus RXRA DT448/9PP-transduced cells. (A) Experimental schema. KMT2A-MLLT3 WT leukemia cells were transduced with MSCV-RXRA WT-IRES-mCherry or MSCV-RXRA DT448/9PPIRES-mCherry and transplanted immediately into sublethally irradiated mice. (B-D) Peripheral blood assessment at 4 weeks. WBC: white blood cells. (E) Assessment of proportion of mCherry+ cells before transplantation. (F) Assessment of mCherry in peripheral blood at 4 weeks. (G) Assessment of mCherry in bone marrow of moribund mice (of note, 2 RXRA DT448/9PP mice were found dead and were unable to be assessed and 1 was sacrificed but did not have splenomegaly). (H) Survival of mice after transplantation. *P<0.05, **P<0.01. The Mann-Whitney test was used in (B, F and G) with unequal variance. Otherwise, the ttest was used.

423

O. di Martino et al.

S7B-F). The list of upregulated genes included CD14 and CD64 (Fcgr1), which at the protein level were observed to increase by immunofluorescence (Figure 1H, Online Supplementary Figure S1C-D). The most highly differentiated transcript out of the cell surface markers was the complement factor C3 (Online Supplementary Figure S7D), and additional complement-receptors (C5ar1 and CD59a) also underwent upregulation with RXRA activation (Online Supplementary Figure S7B, F), which correlate with monocytic differentiation. We also noted a variety of upregulated integrins (Bcam, Ceacam1, Itgam, Itgb2, Itgb2l, and Itgb7), and these may play a role in the aggregation phenotype of the DT448/9PP variant (Online Supplementary Figure S7C-F). While the majority of transcripts that were up- or downregulated overlapped between RXRA WT cells treated with

bexarotene and RXRA DT448/9PP-transduced cells (Figure 5B-D), there were 11 genes that were differentially expressed between these two groups (Online Supplementary Figure S8A). In five, RXRA DT448/9PP augmented the bexarotene-induced response: (upregulated Vsig8, Fcgr1 (Cd64), Camp, Gp6; downregulated: Pik3ip1) (Online Supplementary Figure S8B). The other six displayed three different patterns: not expressed in treated or untreated RXRA WT cells and increased in DT448/9PP-transduced cells (Gpr84, S100a8); downregulated from untreated to treated RXRA WT cells and expressed in DT448/9PP-transduced cells (Mpo, Gm28438, Gm10359); or not expressed in untreated RXRA WT with an increase upon bexarotene treatment but not with DT448/9PP (Cited1) (Online Supplementary Figure S8C). This list is too small for pathway

Figure 5. RXRA DT448/9PP transcriptome changes. (A) Schema of experimental design: KMT2A-MLLT3 WT leukemia cells were transduced with the indicated viruses and cultured for 24 h with or without 250 nM bexarotene; mCherry+ cells were sorted, and total RNA was isolated for RNA sequencing analysis. (B) Heatmap of differentially expressed genes (DEG) between all three groups. Each condition was evaluated by biological triplicates. WT: RXRA WT cells; WT + bex: RXRA WT cells treated with bexarotene; DT448/9PP: RXRA DT448/9PP-transduced cells. (C) Venn diagram with the number of upregulated unique DEG from RXRA WT cells versus RXRA WT cells treated with bexarotene (red) and RXRA WT cells versus RXRA DT448/9PP-transduced cells (blue) and DEG common to both groups (purple). (D) Venn diagram with the number of downregulated unique DEG from RXRA WT cells versus. RXRA WT cells treated with bexarotene (red) and RXRA WT cells versus RXRA DT448/9PP-transduced cells (blue) and DEG in common between both groups (purple). (E, F) Gene set enrichment analysis comparing the untreated RXRA WT cells to the bexarotene-treated and DT448/9PP-transduced cells upregulated (E) and downregulated (F) during myelopoiesis (published gene sets from Brown, et al.26). Heatmaps are shown as row z-scores with the color key below the panel (B). FDR: false discovery rate; NES: normalized enrichment score.

424

haematologica | 2022; 107(2)

RXRA active mutation

analysis, but manual curation found that the most of these 11 transcripts are related to immune function and activity.

Discussion Retinoic X receptors and retinoic acid receptors have been shown to influence myeloid maturation.1,2,4,27,28 RXRA protein levels and activity have been linked to cell fate decisions at the neutrophil versus monocyte decision,29 and deletion of Rxra and Rxrb prevents osteoclast maturation.12 and augments KMT2A-MLLT3 leukemogenesis and cell expansion in vivo,5 suggesting a role for retinoic X receptors in regulating myeloid maturation. RXRA DT448/9PP is a serendipitously discovered mutation that results in constitutive activity, leukemic cell maturation, and loss of proliferative capacity. This variant demonstrates ligand-independent activation, augmented co-activator binding, induced maturation transcripts by RNA sequencing, and reduced engraftment in vivo. To our knowledge, RXRA DT448/9PP has not been spontaneously observed in cancer or other pathological states. Two other RXRA mutations with activating phenotypes have been described. Mouse Rxra F318A has increased transcriptional activity, potentially via increased responsiveness to cell-available oleic acid.22 However, Rxra F318A retains sensitivity to RXRA antagonists, whereas RXRA DT448/9PP did not (Figure 2C, Online Supplementary Figure S2). Furthermore, R316 and L326 are critical amino acids that interact directly with bound ligand and are required for ligand activation of RXRA.5,24 Combining these ligandblocking mutations with DT448/9PP, we noted retained KMT2A-MLLT3 maturation phenotypes (Figure 3A-C), further suggesting ligand-independent activity of RXRA DT448/9PP. Recurrent RXRA hot-spot mutations (S427F/Y) have been noted in patients with bladder cancer. These mutations augment the activity of the PPARG:RXRA heterodimer and are not capable of independently activating RXRA in reporter assays.25 We previously found that combinations of RARA and RXRA ligands lead to leukemic maturation and apoptosis, whereas PPARG and RXRA ligands did not.5 Here we found that RXRA S427F did not recapitulate leukemia maturation phenotypes in KMT2AMLLT3 leukemia cells (Figure 3E-G), and the PPARG antagonist T0070907 did not abrogate RXRA DT448/9PP phenotypes (Online Supplementary Figure S5). Thus, in contrast to Rxra F318A and RXRXA S427F, RXRA DT448/9PP may be constitutively active, independently of endogenous, available, natural ligands, and also may not depend on activation through PPARG:RXR heterodimers or other major nuclear receptor heterodimers. RXRA DT448/9PP constitutive activity may be related to at least two phenotypes. RXRA DT448/9PP is resistant to enzymatic cleavage, and this may enable augmented functional protein levels (Figures 1F and 3D, H), but cannot explain resistance to pan-RXR antagonists or co-mutation with R316A/L326A (Figures 2C and 3A-C, Online Supplementary Figure S4). RXRA DT448/9PP was also associated with ligand-independent co-activator binding (Figure 2D), which may enable augmented, or even constitutive activity. A limitation of these studies is that they require a retroviral overexpression system and therefore the effect of DT448/9PP RXRA at physiological levels is unknown. Not all cells tested were susceptible to RXRA DT448/9PP. The two human myelomonocytic acute myeloid leukemia haematologica | 2022; 107(2)

lines with KMT2A-MLLT3 (THP1 and MonoMac-6) were susceptible, whereas the heterogeneous stem/progenitor populations of Kit+ murine bone marrow cells and the blast phase chronic myeloid leukemia cell line K562 were not (Online Supplementary Figures S2 and S3). Maintained in stem cell cytokines, Kit+ bone marrow cells have multipotent potential and their medium lacks cytokines that might provide monocytic maturation support. K562 cells are susceptible to erythroid maturation stimuli rather than myeloid maturation. Other groups have observed differences in sensitivity and resistance to retinoids across cell lines.9,30,31 Thus, different external signals and/or internal priming states may affect the susceptibility to retinoids and to RXRA DT448/9PP. We have previously noted that Rxra and Rxrb expression negatively regulates KMT2A-MLLT3 leukemia and that these cells are exposed to low levels of natural RXRA ligands in vivo.5,32 Similarly, here we found that cells that overexpress RXRA WT consistently engraft in recipient mice, but are associated with a competitive disadvantage relative to untransduced, mCherry-negative cells (Figure 4). This phenotype was augmented by transduction with RXRA DT448/9PP, which further limited engraftment and leukemic outgrowth in vivo, again suggesting the potential of activated retinoic X receptors to inhibit growth of leukemia cells in vivo. RXRA DT448/9PP resulted in several maturation-related phenotypes, including lack of proliferation, loss of colony formation, acquisition of ruffled borders and podocytes, and increased expression of cell surface markers of myeloid and monocytic maturation. Transcriptional analysis of bexarotene-treated cells versus RXRA DT448/9PP-transduced cells suggested strong overlapping myeloid and monocytic maturation signatures, consistent with ligandindependent, constitutively active effects of DT448/9PP. Many maturation-related transcripts were more effectively induced by RXRA DT448/9PP than by bexarotene, and a few myeloid-related transcripts were uniquely induced by RXRA DT448/9PP. Thus, DT448/9PP may more effectively activate the same loci as ligand-activated RXRA, and activity at select novel loci may enable additional phenotypes. Like other nuclear receptors, the retinoic X receptors are ligand-dependent transcription factors and their function and activity change from transcriptional repressors to transcriptional activators in the presence of active ligands. Multiple natural ligands have been proposed for the retinoic X receptors,6 and it is often difficult to know which cells and settings contain active, natural ligands, and when an observed effect of retinoic X receptors may be related to the receptor function in the absence versus presence of ligand. A constitutively active variant provides a helpful genetic comparator and could be used with diverse forms of previously characterized non-functional (e.g., DDBD, E153G/G154S) and ligand-non-responsive variants (e.g., DAF2, R316A/326A). Retinoid receptors have been an attractive therapeutic target in acute myeloid leukemia; however their effects in clinical trials have been modest. The maturation effects (loss of proliferation and colony-forming potential, and morphological changes) of DT448/9PP on KMT2A-MLLT3 leukemia cells are more robust than the maximal effects of retinoid ligands such as all-trans retinoic acid and bexarotene. Although further delineation of mechanistic activity and cell-type susceptibility rules will be required, these data suggest that current small-molecule retinoids 425

O. di Martino et al.

may incompletely activate the maturation pathways regulated by the retinoid receptors and that additional, unrealized potential may exist for retinoids in the treatment of forms of acute myeloid leukemia other than non-acute promyelocytic leukemia. In particular, myeloid maturation programs may be augmented by correctly activated retinoic X receptors and cells with monocytic potential may be susceptible to retinoid-induced maturation. An efficient, constitutively active RXRA variant may enable further elucidation of this potential. Disclosures No conflicts of interest to disclose. Contributions JSW, OdM and MAF designed and performed experiments, and wrote the manuscript. GH, SS, AV, MPMG, and MR designed and performed experiments.

References 1. Oren T, Sher JA, Evans T. Hematopoiesis and retinoids: development and disease. Leuk Lymphoma. 2003;44(11):1881-1891. 2. Evans T. Regulation of hematopoiesis by retinoid signaling. Exp Hematol. 2005;33(9): 1055-1061. 3. Ricote M, Snyder CS, Leung HY, Chen J, Chien KR, Glass CK. Normal hematopoiesis after conditional targeting of RXRalpha in murine hematopoietic stem/progenitor cells. J Leukoc Biol. 2006;80(4):850-861. 4. Welch JS, Klco JM, Gao F, et al. Combination decitabine, arsenic trioxide, and ascorbic acid for the treatment of myelodysplastic syndrome and acute myeloid leukemia: a phase I study. Am J Hematol. 2011;86(9): 796-800. 5. Di Martino O, Niu H, Hadwiger G, et al. Endogenous and combination retinoids are active in myelomonocytic leukemias. Haematologica. 2021;106(4):1008-1021. 6. Martino OD, Welch JS. Retinoic acid receptors in acute myeloid leukemia therapy. Cancers. 2019;11(12):1915 7. Lubbert M, Grishina O, Schmoor C, et al. Valproate and retinoic acid in combination with decitabine in elderly nonfit patients with acute myeloid leukemia: results of a multicenter, randomized, 2 x 2, phase II trial. J Clin Oncol. 2020:38(3):257-270. 8. vSchlenk RF, Frohling S, Hartmann F, et al. Phase III study of all-trans retinoic acid in previously untreated patients 61 years or older with acute myeloid leukemia. Leukemia. 2004;18(11):1798-1803. 9. McKeown MR, Corces MR, Eaton ML, et al. Super-enhancer analysis defines novel epigenomic subtypes of non-APL AML Including an RARalpha dependency targetable by SY1425, a potent and selective RARalpha agonist. Cancer Discov. 2017;7(10):1136-1153. 10. Sakamoto K, Imamura T, Yano M, et al. Sensitivity of MLL-rearranged AML cells to all-trans retinoic acid is associated with the level of H3K4me2 in the RARalpha promoter region. Blood Cancer J. 2014;4(4):e205. 11. Niu H, Chacko J, Hadwiger G, Welch JS. Absence of natural intracellular retinoids in

426

Acknowledgments We thank the Alvin J. Siteman Cancer Center at Washington University School of Medicine and Barnes-Jewish Hospital in St. Louis, (MO, USA). for the use of the Flow Cytometry Core. The Siteman Cancer Center is supported in part by a National Cancer Institute Cancer Center Support grant (P30 CA91842). We thank Christopher Miller, Sai Ramakrishnan, Deborah Laflamme, Conner York, and Sangeetha Vadivelu for technical assistance. Funding This work was supported by National Institutes of Health grant R01 HL128447 (JSW), by the Siteman Investment Program (JSW), the Washington University SPORE DRP (JSW and MAF), the Children’s Discovery Institute (JSW), the Alex’s Lemonade Stand Foundation Young Investigator Award (MAF), the National Institutes of Health 5K12HD07622408 (MAF), and grants from the Spanish Ministerio de Ciencia e Innovación (MCI) (SAF201790604-REDT-NurCaMeIn, RTI2018-095928-BI00) (MR).

mouse bone marrow cells and implications for PML-RARA transformation. Blood Cancer J. 2015;5(2):e284. 12. Menendez-Gutierrez MP, Roszer T, Fuentes L, et al. Retinoid X receptors orchestrate osteoclast differentiation and postnatal bone remodeling. J Clin Invest. 2015;125(2):809823. 13. Bray NL, Pimentel H, Melsted P, Pachter L. Near-optimal probabilistic RNA-seq quantification. Nat Biotechnol. 2016;34(5):525527. 14. Ashburner M, Ball CA, Blake JA, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet. 2000;25(1):25-29. 15. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A. 2005;102(43):15545-15550. 16. Ghosh JC, Yang X, Zhang A, et al. Interactions that determine the assembly of a retinoid X receptor/corepressor complex. Proc Natl Acad Sci U S A. 2002;99(9):58425847. 17. Lee WY, Noy N. Interactions of RXR with coactivators are differentially mediated by helix 11 of the receptor's ligand binding domain. Biochemistry. 2002;41(8):25002508. 18. Cordeiro TN, Sibille N, Germain P, et al. Interplay of protein disorder in retinoic acid receptor heterodimer and its corepressor regulates gene expression. Structure. 2019;27(8):1270-1285. 19. Gao W, Liu J, Hu M, et al. Regulation of proteolytic cleavage of retinoid X receptoralpha by GSK-3beta. Carcinogenesis. 2013;34(6):1208-1215. 20. Fadel L, Reho B, Volko J, et al. Agonist binding directs dynamic competition among nuclear receptors for heterodimerization with retinoid X receptor. J Biol Chem. 2020;295(29):10045-10061. 21. Devarakonda S, Gupta K, Chalmers MJ, et al. Disorder-to-order transition underlies the structural basis for the assembly of a transcriptionally active PGC-1alpha/ERRgamma complex. Proc Natl Acad Sci U S A. 2011;108(46):18678-18683.

22. Kersten S, Dong D, Lee W, Reczek PR, Noy N. Auto-silencing by the retinoid X receptor. J Mol Biol. 1998;284(1):21-32. 23. Bourguet W, Vivat V, Wurtz JM, Chambon P, Gronemeyer H, Moras D. Crystal structure of a heterodimeric complex of RAR and RXR ligand-binding domains. Mol Cell. 2000;5(2):289-298. 24. Hiromori Y, Aoki A, Nishikawa J, Nagase H, Nakanishi T. Transactivation of the human retinoid X receptor by organotins: use of site-directed mutagenesis to identify critical amino acid residues for organotin-induced transactivation. Metallomics. 2015;7(7): 1180-1188. 25. Halstead AM, Kapadia CD, Bolzenius J, et al. Bladder-cancer-associated mutations in RXRA activate peroxisome proliferator-activated receptors to drive urothelial proliferation. Elife. 2017;6:e30862. 26. Brown AL, Wilkinson CR, Waterman SR, et al. Genetic regulators of myelopoiesis and leukemic signaling identified by gene profiling and linear modeling. J Leukoc Biol. 2006;80(2):433-447. 27. Purton LE. Roles of retinoids and retinoic acid receptors in the regulation of hematopoietic stem cell self-renewal and differentiation. PPAR Res. 2007;2007:87934. 28. Mullen EM, Gu P, Cooney AJ. Nuclear receptors in regulation of mouse ES cell pluripotency and differentiation. PPAR Res. 2007;2007:61563. 29. Taschner S, Koesters C, Platzer B, et al. Down-regulation of RXRalpha expression is essential for neutrophil development from granulocyte/monocyte progenitors. Blood. 2007;109(3):971-979. 30. Yadav B, Pemovska T, Szwajda A, et al. Quantitative scoring of differential drug sensitivity for individually optimized anticancer therapies. Sci Rep. 2014;4:5193. 31. Bianchi N, Ongaro F, Chiarabelli C, et al. Induction of erythroid differentiation of human K562 cells by cisplatin analogs. Biochem Pharmacol. 2000;60(1):31-40. 32. Niu H, Fujiwara H, di Martino O, et al. Endogenous retinoid X receptor ligands in mouse hematopoietic cells. Sci Signal. 2017;10(503):eaan1011.

haematologica | 2022; 107(2)

ARTICLE

Cell Therapy & Immunotherapy

Human invariant natural killer T cells promote tolerance by preferential apoptosis induction of conventional dendritic cells

Ferrata Storti Foundation

Hannes Schmid,1* Emmanuelle M. Ribeiro,1* Kathy-Ann Secker,1 Silke Duerr-Stoerzer,1 Hildegard Keppeler,1 Ruoyun Dong,1 Timo Munz,1 Klaus Schulze-Osthoff,2 Stephan Hailfinger,2 Corina Schneidawind1 and Dominik Schneidawind1 1 2

Department of Medicine II, University Hospital Tübingen, Eberhard Karls University, and Interfaculty Institute of Biochemistry, Eberhard Karls University, Tübingen, Germany

*HS and EMR contributed equally as co-first authors.

Haematologica 2022 Volume 107(2):427-436

ABSTRACT

raft-versus-host disease (GvHD) is a major cause of morbidity and mortality after allogeneic hematopoietic cell transplantation. We recently showed in murine studies and in vitro human models that adoptively transferred invariant natural killer T (iNKT) cells protect from GvHD and promote graft-versus-leukemia effects. The cellular mechanisms underlying GvHD prevention by iNKT cells in humans, however, remain unknown. In order to study relevant cellular interactions, dendritic cells (DC) were either generated from monocytes or isolated directly from blood of healthy donors or GvHD patients and co-cultured in a mixed lymphocyte reaction (MLR) with T cells obtained from healthy donors or transplantation bags. Addition of culture-expanded iNKT cells to the MLR-induced DC apoptosis in a cell contact-dependent manner, thereby preventing T-cell activation and proliferation. Annexin V/propidium iodide staining and image stream assays showed that CD4+CD8–, CD4–CD8+ and double negative iNKT cells are similarly able to induce DC apoptosis. Further MLR assays revealed that conventional DC (cDC) but not plasmacytoid DC (pDC) could induce alloreactive T-cell activation and proliferation. Interestingly, cDC were also more susceptible to apoptosis induced by iNKT cells, which correlates with their higher CD1d expression, leading to a bias in favor of pDC. Remarkably, these results could also be observed in GvHD patients. We propose a new mechanism how ex vivo expanded human iNKT cells prevent alloreactivity of T cells. iNKT cells modulate T-cell responses by selective apoptosis of DC subsets, resulting in suppression of T-cell activation and proliferation while enabling beneficial immune responses through pDC.

Correspondence: DOMINIK SCHNEIDAWIND dominik.schneidawind@med.uni-tuebingen.de Received: July 22, 2020. Accepted: December 22, 2020. Pre-published: January 14, 2021.

Introduction Despite significant advances in the field of allogeneic hematopoietic cell transplantation (HCT), graft-versus-host disease (GvHD) still represents a major complication after allogeneic HCT, leading to substantial morbidity and mortality.1,2 GvHD is mediated by donor T cells activated through antigen-presenting cells (APC).3 Dendritic cells (DC) are professional APC that precisely orchestrate adaptive immune responses and their significant role in GvHD pathophysiology has been established previously.4-6 Both donor and host DC present host antigens and promote activation and proliferation of alloreactive donor T cells, which consequently home to GvHD target sites, resulting in tissue destruction and clinical manifestations of GvHD.7,8 The ability of DC to elicit or prevent T-cell responses is tuned by the concomitant expression of stimulatory or inhibitory molecules as well as immunomodulatory cytokines.9 DC also express antigen-presenting molecules such as the major histocompatibility complex-I (MHC-I)-like molecule CD1d that allows for interactions with invariant natural killer T (iNKT) cells. iNKT cells are a small subset of T lymphocytes characterized by the expression of an invariant T-cell receptor in both humans and

haematologica | 2022; 107(2)

https://doi.org/10.3324/haematol.2020.267583

427

H. Schmid et al.

mice.10 Upon activation through glycolipids, iNKT cells regulate immune responses by the instant release of immunoregulatory cytokines or by direct cell killing.11-13 Several studies have shown the ability of iNKT cells to reduce the incidence of GvHD. In murine models, iNKT cells prevent acute and chronic GvHD, while promoting beneficial graft-versus-leukemia (GvL) effects.14-16 In humans, clinical studies have demonstrated that high iNKT-cell numbers are associated with a diminished occurrence of GvHD.17-19 Moreover, we recently showed that culture-expanded human iNKT cells are able to prevent T-cell activation and proliferation while exerting potent anti-leukemic activity.13,20 Nevertheless, the complex cellular and molecular mechanisms of immune tolerance induction through iNKT cells remain poorly understood. In this study, we focused on how culture-expanded human iNKT cells modulate alloreactive T-cell responses through DC in healthy volunteers and GvHD patients.

Biolegend, San Diego, CA, USA) according to the manufacturer’s instructions and tested in a mixed lymphocyte reaction (MLR).

Mixed lymphocyte reaction Major mismatched mo-DC or blood DC were plated together with allogeneic CD3+ T cells at a 1:1 ratio. Culture-expanded iNKT cells were added to the MLR at different doses, either directly or separated from the MLR by 0.4 µm TC-Inserts (Sarstedt, Nuembrecht, Germany). Cells were analyzed by flow cytometry for activation markers (CD69 and CD25) and proliferation (CFSE). Alternatively, T cells were incubated with anti-CD3/CD28-coated beads (ThermoFisher Scientific, Waltham, MA, USA) in the presence or absence of iNKT cells. For blocking assays, iNKT cells or DC were pre-treated with the respective antibodies or IgG control (Online Supplementary Appendix).

Apoptosis assays

Methods

Apoptosis was assessed with an annexin V-FITC/propidium iodide (PI) Staining Kit (BD Bioscience, Franklin Lakes, NJ, USA), by cell cycle analysis modified according to Nicoletti21 or by image stream analysis (Online Supplementary Appendix). The percentage of apoptotic cells was determined by flow cytometry.

Research subjects

Cytokine analysis

Human buffy coats from healthy volunteers were obtained from the Center of Clinical Transfusion Medicine Tuebingen. Samples from hematopoietic cell grafts and peripheral blood mononuclear cells (PBMC) from patients with GvHD were isolated after written informed consent had been obtained. Human leukocyte antigen (HLA) typing was performed by the Center of Clinical Transfusion Medicine Tuebingen or the HLA laboratory of the Department of Medicine II of the University Hospital Tuebingen. The study was approved by our Institutional Review Board to be in accordance with ethical standards and with the Helsinki Declaration of 1975, as revised in 2013 (IRB approvals 483/2015BO2 and 137/2017BO2).

Cell culture supernatants from MLR were collected after 4 and 24 hours (h), respectively. In order to analyze cytokine production bead-based immunoassays were performed according to the manufacturer’s instructions. Cytokine release was measured by a LEGENDplex human CD8/NK-cell panel (BioLegend). Data were acquired using the Lyric flow cytometer with autosampler (BD Biosciences).

Flow cytometry Antibodies and reagents used for flow cytometric analyses are described in the Online Supplementary Appendix.

Invariant natural killer T-cell expansion and enrichment iNKT cells were expanded from third-party PBMC with some minor modifications as previously described (Online Supplementary Appendix).13 Culture-expanded iNKT cells were purified with antiiNKT MicroBeads (Miltenyi Biotech, Bergisch Gladbach, Germany). Alternatively, iNKT cells were stained with DAPI (4',6diamidino-2-phenylindole, Merck, Darmstadt, Germany), antiCD3, anti-CD4, anti-CD8 antibodies and PBS57-loaded CD1d tetramer allowing for enrichment of iNKT cells and their different subsets by fluorescence-activated cell sorting (FACS).

Generation of monocyte-derived dentritic cells and isolation of blood dendritic cells Monocyte-derived dentritic cells (Mo-DC) were generated as described previously.13 Blood DC from healthy donors and patients were isolated using Blood Dendritic Cell Isolation Kit II (Miltenyi Biotech). Where indicated, HLA-DR+ blood DC were further sorted either as CD1c+ conventional DC (cDC) or CD303+ plasmacytoid DC (pDC).

CD3+ T-cell isolation CD3+ T cells were isolated from human PBMC with anti-CD3 MicroBeads (Miltenyi Biotech). For proliferation analysis, T cells were marked with CFSE (carboxyfluorescein succinimidyl ester,

428

Statistical analysis Student’s t-test and analysis of variance (ANOVA) were used for statistical analysis. P<0.05 was considered statistically significant. Data were analyzed with Prism 8 (GraphPad Software, La Jolla, CA, USA). All experiments were performed at least in technical duplicates and repeated independently at least three times using different iNKT-cell donors.

Results Invariant natural killer T cells inhibit T-cell activation and proliferation in a cell contact-dependent manner Human culture-expanded iNKT cells suppress alloreactive T-cell responses when T lymphocytes are stimulated by MHC-mismatched DC.13 As iNKT cells exhibit potent immunoregulatory properties through a rapid release of humoral mediators, we wondered whether this effect might be related to the inhibition of T-cell function.22 Therefore, we measured early (CD69 expression) and late activation (CD25 expression) as well as proliferation (CFSE dilution) of T cells co-incubated with DC in presence or absence of iNKT cells. iNKT cells were either added directly to the MLR or separated through a TCinsert (transwell [TW]). Direct addition of iNKT cells at different ratios to the MLR reduced T-cell activation and proliferation in a dose-dependent manner. However, iNKT cells which were separated by a TW did not prevent T-cell activation and proliferation (Figure 1A to C; Online Supplementary Figure S1). We conclude from these experiments that iNKT cells mostly rely on direct cell contact to efficiently suppress alloreactive T-cell responses. In addition, T-cell activation and proliferation initiated by artifihaematologica | 2022; 107(2)

iNKT cells promote tolerance by cDC apoptosis

Figure 1. Culture-expanded invariant natural killer T cells inhibit T-cell activation and proliferation. Representative dot plots and histograms showing (A) early activated T cells (CD69+, day 1), (B) late activated T cells (CD25+, day 3) and (C) proliferating T cells (carboxyfluorescein succinimidyl ester [CFSE], day 7). T-cell activation and proliferation was measured after incubation with monocyte-derived dendritic cell (mo-DC) in the presence or absence of invariant natural killer T (iNKT) cells. iNKT cells were added to the culture either directly or separately through a transwell insert (TW). (D) Representative dot plots and histograms showing late activated T cells (CD25+, day 3) and (E) proliferating T cells (CFSE, day 7) after stimulation with antiCD3/CD28-coated beads in the presence or absence of iNKT cells. All events were gated on single cells and living lymphocytes. iNKT cells were excluded from the analysis by gating on CD3+ PBS57-loaded CD1d tetramer+ populations. Histograms show the mean of three independent experiments (n=3). Error bars indicate standard error of the mean. ns: not significant, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. DC: dendritic cells; T: T cells.

haematologica | 2022; 107(2)

429

H. Schmid et al.

F Figure 2. Culture-expanded invariant natural killer T cells induce dendritic cell apoptosis in a dose-dependent manner. (A) Representative dot plots showing absence of dendritic cells (DC) (CD11c+HLA-DR+) after co-culture with invariant natural killer T (iNKT) cells. (B) Representative dot plots and pooled data of living DC (annexin V-/propidium iodide [PI]-) after 4 hours (h) of co-culture with increasing numbers of T cells or iNKT cells. Indicated are the ratios of DC to T or iNKT cells. (C) Representative dot plots showing DC apoptosis in co-cultures with iNKT cells after 1, 2, 4, 6 and 8 h of incubation. (D) Histograms showing increased DNA fragmentation in DC after 4 h and 18 h of co-culture with iNKT cells. The gates on the left of each plot show the percentage of apoptotic nuclei. (E) Representative image stream assay and dot plots showing DC apoptosis induced by iNKT cells after 4 h and 18 h (annexin V+, green; 7-AAD+, red). (F) Representative dot plots, relative and absolute numbers of living DC (annexin V-/PI-) after 4 h of co-culture with different iNKT-cell subsets. Histograms show the mean of three independent experiments (n=3). Error bars indicate standard error of the mean. ns: not significant. HLA-DR: human leukocyte antigen DR-isotype.

430

haematologica | 2022; 107(2)

iNKT cells promote tolerance by cDC apoptosis

Figure 3. Culture-expanded invariant natural killer T cells require cell contact to induce apoptosis through degranulated effector molecules. (A) Representative image stream assay illustrating direct cellular contact between invariant natural killer T (iNKT) cells (PBS57-loaded CD1d tetramer+, yellow) and dendrtic cells (DC) (HLA-DR+, pink) and subsequent DC apoptosis induction (annexin V+, green) after 4 hours (h) of co-incubation. (B) Representative dot plots showing DC apoptosis and pooled data of living DC (annexin V-propidium iodide [PI]-) after co-incubation with iNKT cells either directly or separated by a transwell insert (TW). (C) Percentage of DC apoptosis inhibition after blocking of the receptors CD1d, FasL, TRAIL, NKG2D and applying the inhibitors zVAD-fmk (N-benzyloxycarbonyl-Val-Ala-Asp(O-Me) fluoromethylketone), CMA (concanamycin A) and monensin/brefeldin A. (D) Representative dot plots showing DC apoptosis after co-culture with non-degranulated and degranulated iNKT-cell supernatant. (E) IFN-γ, granzyme B, perforin and granulysin release by iNKT cells after encountering DC analyzed by bead-based immunoassay. Histograms show the mean of three independent experiments (n=3). Error bars show standard error of the mean. ns: not significant; *P<0.05, **P<0.01, ***P<0.001; HLA-DR: human leukocyte antigen DR-isotype.

haematologica | 2022; 107(2)

431

H. Schmid et al.

cial antigen-presenting cells (aAPC, e.g., Dynabeads) was only affected when higher numbers of iNKT cells were added to the culture (Figure 1D and E; Online Supplementary Figure S2). In particular, proliferation speed was decreased with a predominance of early daughter generations (Online Supplementary Figure S2). Our findings suggest that the interaction of iNKT cells with DC largely contributes to the control of alloreactive T cells although a minor direct impact of iNKT cells on T cells could be observed.

Invariant natural killer T cells induce apoptosis of allogeneic dendritic cells in a dose-dependent manner We performed flow cytometry to determine the phenotype of DC challenged with iNKT cells. Notably, DC numbers were highly reduced (Figure 2A) and we suspected induction of apoptosis through iNKT cells. Annexin V assays showed that culture-expanded iNKT cells rapidly induced apoptosis of allogeneic DC, while co-culture of DC with conventional allogeneic CD3+ T cells did not, suggesting that apoptosis induction is not only dosedependent but also specific to iNKT cells (Figure 2B and C). The Nicoletti assay revealed that DC start to defragment their DNA after co-culture with iNKT cells, which represents a further hallmark of apoptosis (Figure 2D). Interestingly, an increase of spontaneous DNA defragmentation could be observed in DC without iNKT cells after 18 h which could be explained by the lack of specific stimuli. Image stream analysis also confirmed morphologic changes in DC after co-culture with iNKT cells. Whereas DC cultured alone presented a healthy and round morphology, DC co-cultured with iNKT cells were small, squashed and with a blobbing membrane. Further, upregulation of Annexin V and loss of nuclear integrity in DC co-incubated with iNKT cells could be observed in image stream assays, confirming our previous assumption (Figure 2E). Next, we investigated whether the induction of apoptosis is specific to certain iNKT-cell subpopulations. Therefore, culture-expanded iNKT cells were sorted into double negative, CD4+CD8– and CD4–CD8+ subsets and co-cultured separately with allogeneic DC: all iNKTcell subsets were able to induce apoptosis of DC with comparable efficiency (Figure 2F).

Induction of dendritic cell apoptosis is cell contact-dependent and mediated by cytotoxic effector molecules In order to further elucidate the cellular and molecular mechanisms responsible for iNKT-cell-induced DC apoptosis, we first analyzed image stream data visualizing doublets consisting of DC and iNKT cells. Image stream analysis revealed a direct binding of iNKT cells (PBS57loaded CD1d tetramer+) to the surface of allogeneic DC (HLA-DR+), which subsequently revealed positive surface staining for the apoptosis marker annexin V (Figure 3A). In order to test whether this direct cellular interaction is required, iNKT cells and DC were separated by a transwell insert demonstrating that iNKT cells were unable to induce DC apoptosis anymore (Figure 3B). We further performed blocking experiments of common key molecules to identify critical pathways responsible for apoptosis induction through iNKT cells. It was observed that blocking FasL, TRAIL or NKG2D did not significantly reduce apoptosis of DC exposed to iNKT cells. However, blocking the CD1d and invariant T-cell receptor interaction 432

reduced DC apoptosis significantly indicating that T-cell receptor engagement contributes to efficient lysis. Further, blocking apoptosis via caspase inhibitor zVAD-fmk (Nbenzyloxycarbonyl-Val-Ala-Asp(O-Me) fluoromethylketone) or inhibition of the perforin pathway via CMA (concanamycin A) also diminished iNKT-cell-mediated cell death of DC. Moreover, the inhibition of iNKT-cell degranulation by monensin and brefeldin A was shown to impede apoptosis induction most efficiently (Figure 3C; Online Supplementary Figure S3). By adding supernatant of DC-triggered degranulated iNKT cells to viable DC, we could show that iNKT cells released cytotoxic factors during degranulation which further induced apoptosis in DC (Figure 3D). In order to identify these factors, we performed bead-based multiplex assays and thereby revealed the release of interferon-g (IFN-g), granzyme B, perforin and granulysin (Figure 3E).

Invariant natural killer T cells induce preferential apoptosis of blood conventional dendritic cells in healthy donors and graft-versus-host disease patients Our previous observations are based on ex vivo cultured mo-DC. In order to support our findings, we additionally performed MLR and apoptosis assays using blood DC isolated from PBMC of healthy donors and GvHD patients following allogeneic HCT. Blood DC are mainly composed of cDC and pDC with the latter expressing lower levels of CD1d (Online Supplementary Figure S4A). Also, human blood DC of healthy volunteers induce activation and proliferation of MHC-mismatched T cells that can be diminished through the addition of iNKT cells (Figure 4A). Given that iNKT cells interact with DC through CD1d and CD1d engagement contributes to efficient lysis of target cells, we aimed to determine how human blood pDC and cDC are susceptible to iNKT-cell apoptosis induction. For this purpose, we isolated HLA-DR+ pDC (CD303+) and cDC (CD1c+) by FACS, co-cultured them separately with iNKT cells for 4 h and stained with annexin V and PI. We observed preferential apoptosis induction of cDC, while pDC were less affected by the addition of iNKT cells (Figure 4B). Further, we wondered whether preferential apoptosis of cDC by iNKT cells would also affect Tcell alloreactivity using fresh human blood DC as stimulators. We observed that only allogeneic cDC in contrast to pDC could induce significant T-cell activation and proliferation (Online Supplementary Figure S4B). Consequently, co-culture of these distinct blood DC subsets with allogeneic T cells and iNKT cells revealed that iNKT cells were also able to suppress activation and proliferation of alloreactive T cells induced by cDC (Figure 4C). Finally, we tested whether our findings also apply to patients with acute GvHD having received grafts from HLA-matched donors. Therefore, we isolated blood DC from PBMC obtained from patients with clinical manifestations of acute GvHD grade ≥2 prior to induction of systemic treatment with steroids. At the time point of blood collection patients had complete donor chimerism in their peripheral blood. cDC from GvHD patients also showed higher expression levels of CD1d (Online Supplementary Figure S4C) and were more susceptible to iNKT-cellinduced apoptosis than pDC (Figure 4D), similarly as demonstrated in our previous experiments with cells from healthy donors. Next, T cells derived from donors prior to transplantation were co-cultured with blood DC from GvHD patients. Importantly, adding iNKT cells from haematologica | 2022; 107(2)

iNKT cells promote tolerance by cDC apoptosis

haematologica | 2022; 107(2)

Figure 4. Preferential apoptosis induction of blood conventional dendritic cells by invariant natural killer T cells. (A) Representative dot plots and pooled data of activated (CD69, day 1; CD25, day 3) and proliferating (CFSE, day 7) major histocompatibility complex (MHC)-matched T cells after coculture with blood dendritic cells (DC) from healthy volunteers in presence or absence of invariant natural killer T (iNKT) cells. (B) Representative dot plots showing increased blood DC apoptosis (upper row) and the frequency of plasmacytoid (pDC, CD303+) and conventional DC (cDC, CD1c+) among living blood DC (annexin V-/propidium iodide [PI]-, lower row) after co-culture with iNKT cells. (C) Representative dot plots showing early and late T-cell activation (CD69, day 1; CD25, day 3) and proliferating T cells (carboxyfluorescein succinimidyl ester [CFSE], day 7) after co-culture with sorted pDC and cDC from healthy volunteers. (D) Representative dot plots showing apoptosis of blood DC from patients with graft-versus-host disease (GvHD) after co-culture (4 hours) with culture-expanded third-party donor iNKT cells and the frequency of cDC and pDC among living DC (annexin V-/PI-). (E) Representative dot plots showing early and late activation (CD25, day 3) and proliferation (CFSE, day 7) of MHC-matched donor T cells after co-culture with blood DC from GvHD patients in presence or absence of third-party donor iNKT cells. Histograms show the mean of three independent experiments. Error bars indicate standard error of the mean. *P<0.05.

433

H. Schmid et al.

third-party donors also inhibited alloreactive T-cell activation and proliferation (Figure 4E).

Discussion Allogeneic HCT is an established therapeutic option for the treatment of advanced and high-risk hematologic malignancies. Efforts to optimize donor selection, tailored preparative conditioning regimes and advanced supportive care have significantly contributed to improved outcomes and enabled long-term survival even in aged and comorbid patient populations. Nevertheless, GvHD and relapse still represent the most important reasons for significant morbidity and mortality after allogeneic HCT.1,2 Various strategies have been applied to prevent or treat GvHD such as immunosuppressive medications and in vivo or ex vivo donor T-cell depletion. However, these approaches are suboptimal since they also inhibit immune reconstitution, pathogen control and beneficial GvL effects, leading to higher relapse rates.23,34 Therefore, strategies that prevent GvHD while preserving the capacity of the graft to promote GvL effects are urgently needed. A convincing body of evidence has demonstrated the potential of iNKT cells as a promising alternative for the prevention of GvHD in both mice and humans. Early murine studies demonstrated that the reinfusion of NK1.1+ T cells after transplantation resulted in GvHD suppression.25 In particular, low doses of CD4+ iNKT cells prevented GvHD lethality in mice by promoting the expansion of Tregs while maintaining GvL effects.14 We also showed previously that third-party iNKT cells are equipotent due to the highly conserved invariant TCR of iNKT cells.15,26 Based on these findings, we used iNKT cells from third-party donors in our present study. In humans, several groups have shown that high numbers of iNKT cells were associated with a decreased incidence of GvHD.17,18,27 Malard et al. also showed in a study of 80 patients that high iNKT-cell numbers in the graft correlated with an increased GvHD-free, relapse-free survival; however, the frequency of Tregs did not seem to correlate with iNKT-cell numbers.19 Moreover, Cheng et al. analyzed Treg expansion after infusion of a-galactosylceramide (a-GalCer), a potent iNKT-cell stimulator, but expansion of Tregs could only be observed in a subset of patients.28 Thus, the role of Treg expansion as mediator of therapeutically used iNKT cells is not well established in humans and suggests further mechanisms that contribute to the immunoregulatory properties of iNKT cells. GvHD can be characterized as a response of donor T cells to host antigens presented by MHC molecules through APC1: first, host APC become activated and present allo-antigens to donor T cells, which are stimulated and expand. Consequently, cellular effectors promote cell damage and apoptosis.1,8,29 In this context, several studies have emphasized the role of DC as potent APC in the pathogenesis of GvHD and therefore, they represent an interesting target for prophylactic and therapeutic strategies against GvHD.30,31 In the present study, we therefore focused on the cellular and humoral interplay of human culture-expanded iNKT cells with DC. Thereby, we could show that iNKT cells induce DC apoptosis and consequently, impair alloreactive T-cell activation and proliferation. Liu and Coman reported similar findings previously 434

hypothesizing a relevant mechanism for the modulation of immune responses and GvHD suppression.32,33 We add significant knowledge by showing that preferential apoptosis induction of cDC leads to a relative expansion of beneficial pDC. In contrast, we did not find significant functional differences regarding distinct iNKT-cell subsets. iNKT cells are activated upon recognition of glycolipids presented by the MHC-I-like molecule CD1d, which is highly expressed on DC.11,34 Hence, T-cell receptor-CD1d engagement induces cytokine release by iNKT cells, which confers immunoregulatory properties and the ability to orchestrate immune responses of several cell types. For instance, the release of cytokines such as IFN-g, tumor necrosis factor-a (TNF-a), interleukin-2 (IL-2), IL-4, IL-17 and IL-21 has been noted.13,22 Beyond immunoregulatory properties, iNKT cells exert potent direct cytotoxic effects using different pathways.13,20 In this context, stimulation via CD95 (Fas)35,36 and TRAIL pathways36,37 has been demonstrated, resulting in a classical lymphocytotoxic response against tumor cells. Moreover, several studies have shown that perforin/granzyme B is involved in iNKT-cell tumor cytotoxicity.38-40 Using different blocking reagents and specific antibodies, we could demonstrate that DC apoptosis induced by iNKT cells relies on degranulation of perforin, granzyme B and granulysin and partially on the interaction of the invariant T-cell receptor with CD1d . DC originate from either myeloid or lymphoid hematopoietic stem cell progenitors in the bone marrow.41,42 They constitute a heterogeneous cell group of different subsets playing distinct roles in regulating immune responses.43 DC have been categorized in cDC, pDC and mo-DC, considering their lineage and expression of transcription factors such as IFN regulatory factors 8 and 4.44 In humans, cDC are potent producers of IL-12 and harbor excellent cross-priming properties. In the context of GvHD, cDC turned out to be important stimulators of alloreactive T-cell responses.45 Also, Markley et al. demonstrated that donor cDC are critical for allo-antigen presentation and consequently potentiate GvHD.46 Besides, cDC are most likely responsible for the replenishment of tissuespecific DC such as migratory Langerhans cells of the skin after inflammation and therefore might contribute to the occurrence and perpetuation of skin GvHD.47,48 In contrast, the functional hallmark of pDC is the release of high quantities of type I and type III interferon (IFN) in response to viral antigen recognition.49,50 Interestingly, precursor and fully differentiated pDC are associated with an improved outcome after allogeneic HCT due to a decreased incidence of GvHD and optimized GvL effects.51,52 Thus, the modulation of cDC and pDC by iNKT cells could represent a useful approach to reduce the incidence of GvHD. In this study, we focused on human blood DC, which are mainly composed of cDC and pDC and can be easily obtained from healthy volunteers and GvHD patients. Our results suggest an additional mechanism of how human culture-expanded iNKT cells prevent GvHD: preferential apoptosis of cDC leads to a relative expansion of beneficial pDC. This bias results in decreased activation and proliferation of alloreactive T cells from healthy volunteers and GvHD patients. However, we could also observe a minor direct impact of iNKT cells on T- cell activation and proliferation when higher numbers of iNKT cells were used. Given the high plasticity and functional haematologica | 2022; 107(2)

iNKT cells promote tolerance by cDC apoptosis

diversity of iNKT cells we assume that several mechanisms, that are not mutually exclusive, are generally involved in tolerance induction: modulation of DC function, expansion of FoxP3 regulatory T cells, induction of a Th2 bias of T-helper cells and decreased expansion of alloreactive donor T cells. Indeed, it has been observed that distinct iNKT-cell subsets are associated with certain functional properties which might explain different findings from other groups in humans and mice. Also, culture conditions might affect the function of iNKT cells after expansion. In conclusion, we postulate an additional mechanism by which iNKT cells prevent GvHD in humans, focusing on their interaction with different DC subsets. iNKT cells promote selective cDC apoptosis through the release of effector molecules such as perforin and granzyme B in a cell-contact-dependent manner, which could consequently prevent GvHD. However, pDC are spared and may still convey beneficial immune responses leading to efficient GvL effects and pathogen control resulting in improved survival after allogenic HCT. Disclosures No conflicts of interest to disclose Contributions HS, EMR and DS designed and performed the research and analyzed data; K-AS SD-S, HK, RD, TM, KS-O, SH and

References 1. Ferrara JLM, Levine JE, Reddy P, Holler E. Graft-versus-host disease. Lancet. 2009;373(9674):1550-1561. 2. Anasetti C, Logan BR, Lee SJ, et al. Peripheral-blood stem cells versus bone marrow from unrelated donors. N Engl J Med. 2012;367(16):1487-1496. 3. Shlomchik WD, Couzens MS, Tang CB, et al. Prevention of graft versus host disease by inactivation of host antigen-presenting cells. Science. 1999;285(5426):412-415. 4. Wang X, Li H, Matte-Martone C, et al. Mechanisms of antigen presentation to T cells in murine graft-versus-host disease: cross-presentation and the appearance of cross-presentation. Blood. 2011; 118(24):6426-6437. 5. Stenger EO, Turnquist HR, Mapara MY, Thomson AW. Dendritic cells and regulation of graft-versus-host disease and graftversus-leukemia activity. Blood. 2012;119(22):5088-5103. 6. Lau J, Sartor M, Bradstock KF, Vuckovic S, Munster DJ, Hart DN. Activated circulating dendritic cells after hematopoietic stem cell transplantation predict acute graft-versushost disease. Transplantation. 2007; 83(7):839-846. 7. Matte CC, Liu J, Cormier J, et al. Donor APCs are required for maximal GVHD but not for GVL. Nat Med. 2004;10(9):987-992. 8. Ferrara JLM, Levy R, Chao NJ. Pathophysiologic mechanisms of acute graft-vs.-host disease. Biol Blood Marrow Transplant. 1999;5(6):347-356. 9. Bourque J, Hawiger D. Immunomodulatory bonds of the partnership between dendritic cells and T cells. Crit Rev Immunol. 2018;38(5):379-401.

haematologica | 2022; 107(2)

CS performed the research and analyzed data; EMR and DS wrote the manuscript. All authors edited the manuscript for content. Acknowledgments We would like to thank the Flow Cytometry Core Facility of the University Hospital Tuebingen for their excellent technical support. Furthermore, we thank Stella Autenrieth for sharing her expertise about dendritic cell biology and Kirsten Lauber for many fruitful discussions about apoptosis. Funding This study was supported by a Max Eder Research fellowship of the German Cancer Aid (Deutsche Krebshilfe, 70112548), a Junior Research Group Grant of the Interdisciplinary Center for Clinical Research (IZKF, 2316-0-0) and the Clinician Scientist Program of the Faculty of Medicine Tuebingen. HS received a grant from the Ludwig Hiermaier Foundation. CS was funded by a Junior Research Group Grant of the Interdisciplinary Center for Clinical Research (IZKF, 2383-0-0), the Clinician Scientist Program of the Faculty of Medicine Tuebingen and the Wuerttemberg Cancer Award (Wuerttembergischer Krebspreis). The National Institutes of Health Tetramer Core Facility kindly provided CD1d tetramer reagents. Data sharing statement Raw data and detailed protocols of the used methods used can be obtained upon direct request to the corresponding author.

10. Lantz O, Bendelac A. An invariant T cell receptor alpha chain is used by a unique subset of major histocompatibility complex class I-specific CD4+ and CD4-8- T cells in mice and humans. J Exp Med. 1994; 180(3):1097-1106. 11. Bendelac A, Lantz O, Quimby ME, Yewdell JW, Bennink JR, Brutkiewicz RR. CD1 recognition by mouse NK1+ T lymphocytes. Science. 1995;268(5212):863-865. 12. Metelitsa LS, Naidenko OV, Kant A, et al. Human NKT cells mediate antitumor cytotoxicity directly by recognizing target cell CD1d with bound ligand or indirectly by producing IL-2 to activate NK cells. J Immunol. 2001;167(6):3114-3122. 13. Schmid H, Schneidawind C, Jahnke S, et al. Culture-expanded human invariant natural killer T cells suppress T-cell alloreactivity and eradicate leukemia. Front Immunol. 2018;9:1817. 14. Schneidawind D, Pierini A, Alvarez M, et al. CD4+ invariant natural killer T cells protect from murine GVHD lethality through expansion of donor CD4+CD25+FoxP3+ regulatory T cells. Blood. 2014; 124(22):3320-3328. 15. Schneidawind D, Baker J, Pierini A, et al. Third-party CD4+ invariant natural killer T cells protect from murine GVHD lethality. Blood. 2015;125(22):3491-3500. 16. Du J, Paz K, Thangavelu G, et al. Invariant natural killer T cells ameliorate murine chronic GVHD by expanding donor regulatory T cells. Blood. 2017;129(23):31213125. 17. Rubio MT, Moreira-Teixeira L, Bachy E, et al. Early posttransplantation donor-derived invariant natural killer T-cell recovery predicts the occurrence of acute graft-versushost disease and overall survival. Blood. 2012;120(10):2144-2154.

18. Chaidos A, Patterson S, Szydlo R, et al. Graft invariant natural killer T-cell dose predicts risk of acute graft-versus-host disease in allogeneic hematopoietic stem cell transplantation. Blood. 2012;119(21):50305036. 19. Malard F, Labopin M, Chevallier P, et al. Larger number of invariant natural killer T cells in PBSC allografts correlates with improved GVHD-free and progression-free survival. Blood. 2016;127(14):1828-1835. 20. Jahnke S, Schmid H, Secker KA, et al. Invariant NKT cells from donor lymphocyte infusions (DLI-iNKTs) promote ex vivo lysis of leukemic blasts in a CD1ddependent manner. Front Immunol. 2019; 10:1542. 21. Riccardi C, Nicoletti I. Analysis of apoptosis by propidium iodide staining and flow cytometry. Nat Protoc. 2006;1(3):14581461. 22. Coquet JM, Chakravarti S, Kyparissoudis K, et al. Diverse cytokine production by NKT cell subsets and identification of an IL-17-producing CD4-NK1.1- NKT cell population. Proc Natl Acad Sci U S A. 2008;105(32):11287-11292. 23. Martinez C, Urbano-Ispizua A. Graft-versus-host disease therapy: something else beyond glucocorticoids? Haematologica. 2011;96(9):1249-1251. 24. Horowitz MM, Gale RP, Sondel PM, et al. Graft-versus-leukemia reactions after bone marrow transplantation. Blood. 1990; 75(3):555-562. 25. Zeng D, Lewis D, Dejbakhsh-Jones S, et al. Bone marrow NK1.1(-) and NK1.1(+) T cells reciprocally regulate acute graft versus host disease. J Exp Med. 1999;189(7):10731081. 26. Brossay L, Kronenberg M. Highly conserved antigen-presenting function of

435

H. Schmid et al. CD1d molecules. Immunogenetics. 1999; 50(3-4):146-151. 27. Haraguchi K, Takahashi T, Hiruma K, et al. Recovery of Valpha24+ NKT cells after hematopoietic stem cell transplantation. Bone Marrow Transplant. 2004;34(7):595602. 28. Chen YB, Efebera YA, Johnston L, et al. Increased Foxp3(+)Helios(+) regulatory T cells and decreased acute graft-versus-host disease after allogeneic bone marrow transplantation in patients receiving Sirolimus and RGI-2001, an activator of invariant natural killer T cells. Biol Blood Marrow Transplant. 2017;23(4):625-634. 29. Reddy P. Pathophysiology of acute graftversus-host disease. Hematol Oncol. 2003; 21(4):149-161. 30. Duffner UA, Maeda Y, Cooke KR, et al. Host dendritic cells alone are sufficient to initiate acute graft-versus-host disease. J Immunol. 2004;172(12):7393-7398. 31. Zhang Y, Louboutin JP, Zhu J, Rivera AJ, Emerson SG. Preterminal host dendritic cells in irradiated mice prime CD8+ T cellmediated acute graft-versus-host disease. J Clin Invest. 2002;109(10):1335-1344. 32. Coman T, Rossignol J, D'Aveni M, et al. Human CD4- invariant NKT lymphocytes regulate graft versus host disease. Oncoimmunology. 2018;7(11):e1470735. 33. Liu TY, Uemura Y, Suzuki M, et al. Distinct subsets of human invariant NKT cells differentially regulate T helper responses via dendritic cells. Eur J Immunol. 2008; 38(4):1012-1023. 34. Exley M, Garcia J, Balk SP, Porcelli S. Requirements for CD1d recognition by human invariant Valpha24+ CD4-CD8- T cells. J Exp Med. 1997;186(1):109-120. 35. Wingender G, Krebs P, Beutler B, Kronenberg M. Antigen-specific cytotoxicity by invariant NKT cells in vivo is

436

CD95/CD178-dependent and is correlated with antigenic potency. J Immunol. 2010; 185(5):2721-2729. 36. Mattarollo SR, Kenna T, Nieda M, Nicol AJ. Chemotherapy pretreatment sensitizes solid tumor-derived cell lines to V alpha 24+ NKT cell-mediated cytotoxicity. Int J Cancer. 2006;119(7):1630-1637. 37. Nieda M, Nicol A, Koezuka Y, et al. TRAIL expression by activated human CD4(+)V alpha 24NKT cells induces in vitro and in vivo apoptosis of human acute myeloid leukemia cells. Blood. 2001;97(7):2067-2074. 38. Kawano T, Cui J, Koezuka Y, et al. Natural killer-like nonspecific tumor cell lysis mediated by specific ligand-activated Valpha14 NKT cells. Proc Natl Acad Sci U S A. 1998;95(10):5690-5693. 39. Nicol A, Nieda M, Koezuka Y, et al. Human invariant valpha24+ natural killer T cells activated by alpha-galactosylceramide (KRN7000) have cytotoxic anti-tumour activity through mechanisms distinct from T cells and natural killer cells. Immunology. 2000;99(2):229-234. 40. Beilke JN, Kuhl NR, Van Kaer L, Gill RG. NK cells promote islet allograft tolerance via a perforin-dependent mechanism. Nat Med. 2005;11(10):1059-1065. 41. Watowich SS, Liu YJ. Mechanisms regulating dendritic cell specification and development. Immunol Rev. 2010;238(1):76-92. 42. Manz MG, Traver D, Miyamoto T, Weissman IL, Akashi K. Dendritic cell potentials of early lymphoid and myeloid progenitors. Blood. 2001;97(11):3333-3341. 43. Yu H, Tian Y, Wang Y, Mineishi S, Zhang Y. Dendritic cell regulation of graft-vs.-host disease: immunostimulation and tolerance. Front Immunol. 2019;10:93. 44. Collin M, Bigley V. Human dendritic cell subsets: an update. Immunology. 2018;154(1):3-20.

45. Koyama M, Hashimoto D, Aoyama K, et al. Plasmacytoid dendritic cells prime alloreactive T cells to mediate graft-versushost disease as antigen-presenting cells. Blood. 2009;113(9):2088-2095. 46. Markey KA, Banovic T, Kuns RD, et al. Conventional dendritic cells are the critical donor APC presenting alloantigen after experimental bone marrow transplantation. Blood. 2009;113(22):5644-5649. 47. Martinez-Cingolani C, Grandclaudon M, Jeanmougin M, Jouve M, Zollinger R, Soumelis V. Human blood BDCA-1 dendritic cells differentiate into Langerhanslike cells with thymic stromal lymphopoietin and TGF-beta. Blood. 2014; 124(15): 2411-2420. 48. Ito T, Inaba M, Inaba K, et al. A CD1a(+)/CD11c(+) subset of human blood dendritic cells is a direct precursor of Langerhans cells. J Immunol. 1999;163(3): 1409-1419. 49. Siegal FP, Kadowaki N, Shodell M, et al. The nature of the principal type 1 interferon-producing cells in human blood. Science. 1999;284(5421):1835-1837. 50. Cella M, Jarrossay D, Facchetti F, et al. Plasmacytoid monocytes migrate to inflamed lymph nodes and produce large amounts of type I interferon. Nat Med. 1999;5(8):919-923. 51. Lu Y, Giver CR, Sharma A, et al. IFNgamma and indoleamine 2,3-dioxygenase signaling between donor dendritic cells and T cells regulates graft versus host and graft versus leukemia activity. Blood. 2012; 119(4):1075-1085. 52. Hassan M, Ulezko Antonova A, Li JM, et al. Flt3L Treatment of bone marrow donors increases graft plasmacytoid dendritic cell content and improves allogeneic transplantation outcomes. Biol Blood Marrow Transplant. 2019;25(6):1075-1084.

haematologica | 2022; 107(2)

ARTICLE

Cell Therapy & Immunotherapy

CD38 knockout natural killer cells expressing an affinity optimized CD38 chimeric antigen receptor successfully target acute myeloid leukemia with reduced effector cell fratricide Mark Gurney,1 Arwen Stikvoort,2 Emma Nolan,1 Lucy Kirkham-McCarthy,1 Stanislav Khoruzhenko,3 Rama Shivakumar,3 Sonja Zweegman,2 Niels W.C.J. van de Donk,2 Tuna Mutis,2 Eva Szegezdi,1 Subhashis Sarkar1# and Michael O’Dwyer1# National University of Ireland Galway, Galway, Ireland; 2Cancer Center Amsterdam, VU University Medical Center, Amsterdam, the Netherlands and 3MaxCyte, Inc., Gaithersburg, MD, USA 1

Ferrata Storti Foundation

Haematologica 2022 Volume 107(2):437-445

SS and MOD contributed equally as co-senior authors.

ABSTRACT

here is a strong biological rationale for the augmentation of allogeneic natural killer (NK) cell therapies with a chimeric antigen receptor (CAR) to enhance acute myeloid leukemia (AML) targeting. CD38 is an established immunotherapeutic target in multiple myeloma and under investigation as a target antigen in AML. CD38 expression on NK cells and its further induction during ex vivo NK cell expansion represent barriers to the development of a CD38 CAR-NK cell therapy. We set out to develop a CD38 CAR-NK cell therapy for AML, first by using an NK cell line which has low baseline CD38 expression and subsequently NK cells expanded from healthy donors. To overcome anticipated fratricide due to NK cell CD38 expression when using primary expanded NK cells, we applied CRISPR/Cas9 genome editing to disrupt the CD38 gene during expansion, achieving a mean knockdown efficiency of 84%. The resulting CD38 knockdown expanded NK cells, after expression of an affinity optimized CD38 CAR, showed reduced NK-cell fratricide and an enhanced ability to target primary AML blasts. Furthermore, the cytotoxic potential of CD38 CAR-NK cells was augmented by pretreatment of the AML cells with all-trans retinoic acid which drove enhanced CD38 expression, offering a rational combination therapy. These findings support the further investigation of CD38 knockdown - CD38 CAR-NK cells as a viable immunotherapeutic approach to the treatment of AML.

Correspondence: MICHAEL O’DWYER michael.odwyer@nuigalway.ie Received: September 11, 2020. Accepted: December 22, 2020. Pre-published: December 30, 2020.

Introduction Acute myeloid leukemia (AML) is the most common acute leukemia in adults, accounting for approximately 2% of all cancer deaths.1 Curative treatment approaches remain chemotherapy-based, with allogeneic stem cell transplant consolidation for selected patients. The introduction of molecularly targeted therapies has provided important incremental improvements for specific AML subtypes.2-4 Relapsed disease, mediated by the persistence of chemotherapy-resistant leukemic stem cells (LSC) is particularly difficult to treat, and accounts for much of the mortality burden associated with AML. For many older patients, treatment options that are both tolerable and efficacious do not yet exist. Anti-CD19 chimeric antigen receptor (CAR) T-cell therapies have provided a ground-breaking approach to cancer immunotherapy in B-cell acute lymphoblastic leukemia and Bcell non-Hodgkin lymphomas.5,6 While there is considerable interest in applying the principle of CAR technology in other diseases, progress in AML has been limited to date by the absence of an ideal antigenic target, concerns about ‘on-target off-tumor’ toxicity including that to normal hematopoietic stem cells, and blast cell heterogeneity which exists both within and between patients.7-9.

haematologica | 2022; 107(2)

https://doi.org/10.3324/haematol.2020.271908

437

M. Gurney et al.

The multifunctional cell surface glycoprotein CD38, a breakthrough immunotherapeutic target in multiple myeloma, is also considered a potential target antigen in AML. In contrast to the uniformly high CD38 expression on malignant plasma cells, blast cell CD38 expression is heterogeneous although frequently exceeds that of normal cell populations.10 The CD38 monoclonal antibody daratumumab has been investigated in AML and has shown promising pre-clinical activity.10 CD38 CAR-T cells have been evaluated mainly for their activity in multiple myeloma and cytotoxicity against primary AML samples has also been confirmed.11 However, there remains concern about a potent myelosuppressive effect with a constituently expressed high-affinity anti-CD38 CAR due to CD38 expression on both mature myeloid cells and their precursors.11,12 To circumvent this problem, an affinity-optimized CD38 CAR has been developed to minimize the targeting of positive, but low-expressing normal cell populations.13 There is a strong biological rationale for natural killer (NK) cell-based approaches to adoptive cell transfer immunotherapy for AML. NK cells confer a component of the graft-versus-leukemia effect of allogeneic stem cell transplant and infusions of purified alloreactive NK cells have proven therapeutic potential.14-16 CAR-NK cell therapies are emerging as a complementary approach to CART cells, with potential advantages including allogeneic cell sources and innate antigen independent anti-leukemic activity. An early clinical report of a cord-blood derived CD19 CAR-NK cell therapy has shown promising safety and efficacy in B-cell malignancies.17 We set out to develop and evaluate an affinity optimized CD38 CAR-NK cell therapy for AML. We first used the NK cell line KHYG-1, which has naturally low levels of CD38 expression. While allogeneic expanded NK (eNK) cell approaches are more suited to clinical translation, ex vivo NK cell expansion has been shown to lead to upregulation of CD38, which we also encountered using a feeder-free, interleukin-2-based expansion protocol.18 To reduce the anticipated NK cell fratricide that would occur using eNK cells, we applied CRISPR/Cas9 to disrupt the CD38 gene during NK cell expansion, creating fratricide-resistant NK cells prior to CD38 CAR expression. Both KHYG-1 and CD38 knockdown (KD) eNK approaches lead to efficient targeting of AML blasts upon CD38 CAR expression, with the degree of cytotoxicity correlating with CD38 expression. Finally, we confirm a rational combination approach utilizing all-trans retinoic acid (ATRA) to enhance CD38 expression on the AML cells. Collectively, our data support the potential of CD38 as a therapeutic target in AML and help to define a CD38 CAR-NK cell approach suited to clinical development.

obtained from the American Type Culture Collection and their identities confirmed by short tandem repeat profiling (Eurofins Genomics™). CD38 CAR and mock KHYG-1 cells were generated by retroviral transduction with genomic integration confirmed by the inclusion of DsRed fluorescent protein. The development of the second-generation CD28-CD3ζ, optimized-affinity CD38 CAR was reported previously.13 Primary NK cells were isolated from healthy donor peripheral blood mononuclear cells after Ficoll-Paque density gradient centrifugation and negative immunomagnetic selection (NK Isolation Kit, Miltenyi Biotec™). NK cells were expanded in NK MACS medium (Miltenyi Biotec™) containing NK MACS supplement, 5% heat-inactivated human AB serum and 100 U/mL. interleukin-2 (PeproTech™). Cultures were pre-treated for 48 h with ATRA (Sigma-Aldrich™) or dimethyl sulfoxide, in the relevant experiments.

CRISPR/Cas9 gene editing Five days after isolation, 5x105 NK cells were electroporated with sgRNA-Cas9 complexes targeting multiple sites within the CD38 gene (Gene Knockout Kit V2, Synthego™) or control electroporated (MaxCyte™ GT flow transfection system). CD38-edited and control electroporated cells were expanded at a target density of 1x106 cells/mL. On day 13-15 of expansion, CD38 expression was assessed by flow cytometry. Knockdown efficiency was calculated as (% CD38-positive cells [mock electroporated] - % CD38-positive cells [CRISPR/Cas9 edited]).

CD38 chimeric antigen receptor mRNA electroporation CD38 CAR mRNA was synthesized (Trilink Biotechnologies™) and CD38 CAR expression in primary CD38 KD and control eNK cells was achieved by electroporation (100 mg/mL mRNA, Maxcyte™ GT Flow Transfection System). CAR expression was confirmed by flow cytometry using anti-IgG H+L specific goat anti-human antibody (JacksonImmuno research™) and biotinylated protein L stain (ACRO Biosystems™).

Cytotoxicity assays

Methods

Co-culture experiments involved 10,000 target cells (cell lines), or 20,000-50,000 bone marrow mononuclear cells from AML patients’ samples. NK cell numbers were determined by the desired effector to target (E:T) cell ratio. After co-culture for 18-24 h, target cell lines or bone marrow mononuclear cells were identified by flow cytometry, using a cell-tracking dye: Tag-IT BV™ proliferation and cell tracking dye (Biolegend™) or VioletTraceTM (Thermo Fisher). Primary blast cell populations were identified as CD45int/SSClow (CD45 APC), supported by additional markers chosen based on clinical immunophenotyping data. Cell death was determined using propidium iodide (PI) or LIVE/DEAD Fixable Near-IR (Life Technologies L10119) staining and reported as ‘% specific (blast) cytotoxicity’ ([sample cytotoxicity – background cytotoxicity]/[100 – background cytotoxicity] x 100%) or ‘% blast cell cytotoxicity’ as indicated.

Ethical statement

Statistical analysis

Healthy donor blood and AML patients’ bone marrow samples were collected with written informed consent and approval from the institutional review boards at each institution (ref: CA2219). Cryopreserved samples were obtained from the biobank of Blood Cancer Network Ireland.

GraphPad Prism 8 software (San Diego, CA, USA) was used for statistical analysis. Comparisons were conducted using multiple two-sided t-tests for cytotoxicity assays at each E:T ratio or one-way analysis of variance for cell expression data with statistical significance indicated by asterisks (*P<0.05, **P<0.01, ***P<0.001 and ****P<0.0001). Flow cytometry data were acquired on a BD FACS Canto II and analyzed using Flow Jo V10 software and Infinicyt (Cytognos™).

Cells and reagents The cell lines THP-1, KG1a, U937 and KHYG-1 were

438

haematologica | 2022; 107(2)

CD38 CAR-NK cells targeting AML

Results CD38 chimeric antigen receptor expression enhances KHYG-1 acute myeloid leukemia targeting To assess the feasibility of targeting CD38 with a CARNK cell approach in AML, we first defined the CD38 expression profile of AML cell lines. We classified THP-1 and U937 as CD38-positive, and KG1a as CD38-negative for further experiments (Figure 1A). We also confirmed a low level of CD38 expression on the KHYG-1 cell line, previously shown to be a NK cell line with significant

cytotoxic potential, expressing a high concentration of active perforin and signaling kinases.19 The low CD38 expression of KHYG-1 is in contrast to that of NK-92 cells, another NK cell line which has been investigated clinically as an adoptive cell therapy, but is strongly CD38-positive.18 CD38 CAR-KHYG-1 cells were generated by retroviral transduction using an ‘affinity-optimized’, second-generation anti-CD38 CAR (CD3ζ-CD28).13 CD38 CARKHYG-1 cells displayed similar characteristics and kinetics in cell culture as those of mock-transduced KHYG-1

Figure 1. Affinity-optimized CD38 CAR-KHYG-1 cell cytotoxicity against acute myeloid leukemia cell lines. (A) Histograms depict CD38 expression of the KHYG-1 NK cell line and acute myeloid leukemia (AML) cell lines THP-1 (CD38-positive), U937 (CD38-positive) and KG1a (CD38-negative) with mean fluorescence intensity (MFI) indicated. Bars represent relative CD38 expression of KHYG-1 and AML cell lines (n=3 individual repetitions). Mean values for AML cell lines compared by an unpaired t-test. (B) Bar chart depicting the specific cytotoxicity of mock-transduced and CD38 CAR-KHYG-1 cells at varying effector to target (E:T) ratios against the CD38-positive cell lines THP1 and U937. Comparisons of four independent experiments made by an unpaired t-test at each effector to target (E:T) ratio. (C) Representative histograms depicting CD38 upregulation and MFI values in KG1a cells after 48 h of treatment with all-trans retinoic acid (ATRA) at 10 nM and 20 nM concentrations, compared to dimethylsulfoxide (DMSO) control treated KG1a cells. The bar chart summarize data from four independent experiments with comparisons by one-way analysis of variance. (D) Bar chart depicting specific cytotoxicity of mock-transduced and CD38 CAR-KHYG-1 cells in co-culture with 48 h, 10 nM ATRA-pretreated KG1a cells at varying E:T ratios (summary of 4 experiments). Error bars indicate standard error of mean (SEM). Statistical significance is defined as *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001.

haematologica | 2022; 107(2)

439

M. Gurney et al.

cells. To assess whether CD38 CAR targeting has an additional cytotoxic effect, we tested CD38 CAR-KHYG-1 cells and mock-transduced KHYG-1 cells against CD38positive AML cell lines (Figure 1B). CD38 CAR-KHYG-1 cells demonstrated greater cytotoxic effects against CD38-positive cell lines relative to mock-transduced KHYG-1 cells at all E:T ratios tested, with relatively greater increases seen against the strongly CD38-positive THP-1 cells (specific cytotoxicity 58% vs. 28% for THP1, 10:1 E:T ratio; P<0.0001). ATRA has been shown to upregulate CD38 expression across all AML subtypes, mediated by a retinoic acid response element in the first intron of the CD38 gene.20 Pretreatment with ATRA at 10 nM for 48 h led to marked induction of CD38 expression on KG1a cells, which do not express detectable levels of CD38 in resting conditions (Figure 1C). CD38 CAR-KHYG-1 cells were cytotoxic to ATRA-pretreated KG1a cells, while mock-transduced KHYG-1 cells showed little cytotoxicity despite ATRA pretreatment (Figure 1D). To better mimic the CD38 expression profile encountered in AML, we tested the efficacy of CD38 CARKHYG-1 cells against primary bone marrow mononuclear cells from AML patients (Figure 2A, B). CD38 CARKHYG-1 cells displayed greater specific cytotoxicity against AML blasts relative to mock-transduced KHYG-1 cells across a range of blast cell CD38 expression, with the degree of specific cytotoxicity correlating with blast cell CD38 expression (Figure 2C).

CRISPR/Cas9 gene editing of CD38 in primary NK cells reduces NK cell fratricide upon CD38 chimeric antigen receptor expression While alloreactive NK cell approaches have shown some success in treating AML, we hypothesized that increased expression of CD38 during ex vivo NK cell expansion could be sufficient to trigger effector cell fratricide after expression of a CD38 CAR, despite affinity optimization. Indeed, we observed a consistent, mean 4fold increase in CD38 expression during feeder-free expansion of NK cells in interlueukin-2-containing media. Increases in CD38 from baseline (mean fluorescence intensity [MFI] 11,903) were detectable by day 5 (MFI 40,948) and persisted to at least day 13 (MFI 38,600) (Figure 3A). Extrapolating from our previous work on THP-1 cells (MFI 31,866), we concluded that this degree of CD38 expression would lead to a fratricidal effect upon CD38 CAR expression thus limiting the cytotoxic capacity of ex vivo-expanded CD38 CAR-NK cells. We, therefore, set out to use CRISPR/Cas9 gene editing technology to disrupt the CD38 gene in primary NK cells. We used a multi-sgRNA format, introducing sgRNA-Cas9 complexes using a high-efficiency, electroporation-based approach on a platform scalable to Good Manufacturing Practice (GMP) grade development. CD38 KD and mock-electroporated cells were further expanded for use in functional assays. A consistent KD effect was achieved across all NK cell donors (mean 84%; range, 75-92%) (Figure 3B). CD38 KD was detectable 48 h after CRISPR/Cas9 gene editing, peaked by day 3-7 after electroporation and was stable across the duration of expansion suggesting minimal differences in the growth potential of CD38 KD and mock-electroporated NK cells in this expansion system (Online Supplementary Figure S1). 440

To confirm that CD38 KD eNK cells showed greater resistance to fratricide than wild-type eNK cells, we introduced mRNA coding for an affinity-optimized CD38 CAR. CAR expression was confirmed by complementary staining techniques – an anti-human IgG with light chain specificity, and biotinylated protein L, with control (background) and CAR staining depicted in Figure 3C. CD38 KD eNK cells displayed significantly less cell death than wild-type eNK cells, measured 18 h after CD38 CAR mRNA electroporation in the absence of target cells (18% vs. 37%, P=0.002) (Figure 3D), confirming a greater resistance to fratricide. Furthermore, the biphasic CD38 expression pattern (representing the small residual CD38positive NK cell population after CRISPR/Cas9 gene editing) was lost in the CD38 KD population after CD38 CAR mRNA transfection, but not after non-specific (CD16) mRNA electroporation (Figure 3E). This emphasized the tendency of the CD38 CAR-NK cells to target CD38-positive eNK cells despite affinity-optimization of the CD38 CAR binding domain.

CD38 knockdown - CD38 chimeric antigen receptor-NK cells efficiently target primary acute myeloid leukemia blasts To confirm that CD38 CAR expression in CD38 KD eNK cells enhances the activity of alloreactive NK cells against AML, CD38 KD eNK cells were electroporated with CD38 CAR mRNA or mock-electroporated prior to co-culture with bone marrow mononuclear cells from AML patients with a variety of molecular AML subtypes (Online Supplementary Table S1). CD38 KD - CD38 CARNK cells showed enhanced cytotoxicity relative to mockelectroporated CD38 KD cells, with the effect being most prominent at the highest E:T ratios tested (Figure 4A, B). Enhanced cytotoxicity was observed for all AML patients and cytotoxicity at the 5:1 E:T ratio correlated with blast cell CD38 expression (R2=0.81) (Figure 4C). We investigated the potential of ATRA pretreatment as a means of modulating CD38 expression and potentiating the effects of CD38 CAR targeting using CD38 KD CD38 CAR-NK cells. ATRA pretreatment induced a mean 5-fold upregulation of surface CD38 expression in blast cells (Figure 4D). The increased CD38 expression was associated with greater sensitivity to CD38 KD - CD38 CAR-NK cells compared to dimethylsulfoxide-treated bone marrow mononuclear cell samples tested at the 2:1 and 5:1 E:T ratios (Figure 4E).

Discussion We set out to augment the potential of NK cell adoptive transfer strategies in AML through expression of an affinity-optimized CD38 CAR. We demonstrated two potential approaches to CD38 CAR-NK cell therapy in this setting. We confirmed that CD38 KD eNK cells show reduced fratricide after CD38 CAR expression, allowing effective targeting of primary AML blasts. As an alternative approach we modified the NK cell line KHYG-1 to express a CD38 CAR, successfully targeting AML cell lines and primary samples. Both approaches could be enhanced by induction of CD38 expression using ATRA. We chose a NK cell line with naturally low CD38 expression to ensure viability after introducing a CD38 CAR. KHYG-1 cells have previously been shown to maintain haematologica | 2022; 107(2)

CD38 CAR-NK cells targeting AML

cytotoxicity after irradiation and could be applied clinically in a similar manner to the NK-92 cell line.21 However, irradiation limits the potential for in vivo expansion and persistence - important variables in determining the clinical efficacy of cellular therapies. This requirement for irradiation may be avoided by using donor-derived, eNK cells, although this approach is further complicated by robust CD38 upregulation encountered during ex vivo expansion. Our CRISPR/Cas9 CD38 KD eNK cells reduce effector cell fratricide, representing an approach that could be explored clinically. CD38 was a breakthrough immunotherapeutic target in multiple myeloma. While there is greater variability in CD38 expression in AML, CD38 is a potential target antigen in this disease. Daratumumab was shown to be active in an in vivo model of AML, while isatuximab has recently been examined in a large-scale, in vitro study.10,22

The expression pattern of CD38 in AML, in which there is often overlap with normal cell populations including myeloid and monocytic populations, raises concerns about considerable ‘on-target, off-tumor’ toxicity when a potent effector cell is directed toward CD38. High-affinity CD38 CAR strategies may maximize the proportion of patients for whom a CD38-directed therapy is likely to have activity, at the expense of considerable myelosuppressive effects. It is important to consider that not all offtumor effects are undesirable: in the case of CD38, elimination of CD38-positive immunoregulatory cell subsets may lead to a beneficial therapeutic effect.23,24 The efficacy of lower-affinity CD38 CAR strategies is likely to be limited to cases with strong expression or pharmacological upregulation of CD38. Herein we investigated an approach to CD38 CAR targeting in AML which aims to strike the balance of efficacy, applicability, and off-tumor

Figure 2. CD38 CAR-KHYG-1 activity against primary acute myeloid leukemia samples. (A) Histograms depict unstained controls (blue) and anti-CD38 stained blast cells (red), from a range of acute myeloid leukemia (AML) patients chosen to represent a spectrum of CD38 expression. Relative mean fluorescence intensity (MFI) figures for stained samples are reported. (B) Graphs represent specific blast cytotoxicity after co-culture assays with CD38 CAR transduced KHYG-1 (blue) and mocktransduced KHYG-1 (black) at specified effector to target (E:T) ratios for each corresponding patient’s sample in Figure 2A. (C) The correlation plot and linear regression line depicts specific blast cell cytotoxicity at the E:T ratio of 3:1, versus CD38 expression (relative MFI) of primary AML samples from all co-culture experiments carried out in 2A, (n=8 experiments).

haematologica | 2022; 107(2)

441

M. Gurney et al.

effects. The established anti-leukemic activity of alloreactive NK cells in AML provides a rationale for the development of CAR-NK cell approaches.14,15 Alloreactive NK cells can be expected to retain their innate anti-leukemic activity, with enhancement against CD38-positive cells conferred by an anti-CD38 CAR. Toxicity against normal cell populations can be minimized through the use of an

optimized-affinity CD38 CAR variant.11 While many target antigens are being considered in AML, CD38 is also unique in the availability of a licensed and well-tolerated oral agent capable of modulating target antigen expression, ATRA.20 Furthermore, it has been shown that malignant blast cells are particularly sensitive to the CD38-inducing effect of ATRA, acting directly

Figure 3. CRISPR/Cas9 gene editing of CD38 in primary expanded natural killer cells to reduce natural killer cell fratricide upon CD38 CAR expression. (A) CD38 expression in freshly isolated (day 0), and expanded natural killer (eNK) cells at day 5 and day 13, as measured by flow cytometry and presented as a representative histogram and summary bar chart of three unique expansions. (B) Residual CD38 expression measured on day 8-10 after CRISPR/Cas9 gene editing of CD38. Dot plots from representative donor for mock-electroporated and CD38 knockdown (KD) conditions. Bar chart represents summary data for four donors. (C) Confirmatory CAR staining performed 18 h after CD38 CAR mRNA electroporation. Pseudo-colored plots depict results from one representative experiment. (D) NK cell death after 18 h of culture after CD38 CAR mRNA electroporation comparing CD38 KD and control eNK cells across three individual experiments and NK cell donors. Comparison of mean cell death by an unpaired t-test. (E) Histogram depicting a representative residual CD38 expression profile of viable CD38 KD eNK cells demonstrating loss of residual CD38-positive eNK cell population after CD38 CAR electroporation but not after ‘dummy’ mRNA (CD16) electroporation. EP: electroporated; H + L: heavy and light chain specific. Error bars indicate standard error of mean (SEM). Statistical significance is defined as *P≤0.05, **P≤0.01.

442

haematologica | 2022; 107(2)

CD38 CAR-NK cells targeting AML

through a retinoic acid response element within the CD38 gene.25 The vitamin D receptor agonist inecalcitol represents another investigational approach to CD38 modulation which could enhance the efficacy of CD38directed therapies in AML through a similar principle.26 The timing of a clinically applied combination therapy using ATRA and a CD38 CAR-NK cell would require

careful planning. Evidence suggests that NK cell exposure to ATRA may have a net inhibitory effect on NK cell function, suggesting that the preferred approach may be ATRA prior to adoptive cell transfer.27 CD38 targeted therapies are complicated by NK cell CD38 expression, observed clinically with the NK celldepleting effects of daratumumab seen during the treat-

Figure 4. CD38 KD - CD38 CAR-NK cells show enhanced cytotoxicity against primary acute myeloid leukemia samples. (A) CD38 knockdown (KD) expanded natural killer (NK) cells were electroporated with CD38 CAR mRNA or ‘dummy’ mRNA (CD16) prior to co-culture with bone marrow mononuclear cells from acute myeloid leukemia (AML) patients. Blast cell cytotoxicity was measured by percentage of propidium iodide (PI)-positive cells (representative dot-plots are shown). (B) Summary data of co-culture assays as described in (A), for seven AML patients compared using unpaired t-tests for each effector to target (E:T) cell ratio. (C) The CD38 expression level of the blast population was correlated with the cytotoxic effect observed at an E:T ratio of 5:1 for experiments conducted in (B), and a linear regression model fitted using GraphPad Prism. (D) Bone marrow mononuclear cells from n=4 donors were treated with 10 nM all-trans retinoic acid (ATRA) or dimethyl sulfoxide (DMSO) for 48 h prior to anti-CD38 staining. A representative histogram is displayed and summary data of four pooled donors were compared using an unpaired ttest. (E) ATRA or DMSO pretreated cells were co-cultured with CD38 KD-CD38 CAR-NK cells. Summary data from three experiments. Analysis by unpaired t-test for each E:T ratio. Statistical significance is defined as *P≤0.05, **P≤0.01, ***P≤0.001.

haematologica | 2022; 107(2)

443

M. Gurney et al.

ment of multiple myeloma.28 Indeed, overcoming the fratricidal effect of daratumumab through combination with ex vivo eNK cells is actively under investigation in multiple myeloma. In keeping with prior reports, we observed CD38 upregulation during NK cell expansion, which was sufficient to lead to a fratricidal effect despite the use of an optimized-affinity CD38 CAR design.18 While it has been considered difficult to apply genetic engineering approaches to primary NK cells, we achieved a consistent, and high-efficiency disruption of the CD38 gene using a multi-sgRNA approach coupled with a flow transfection system. Our findings are comparable to recent descriptions of CRISPR/Cas9 editing in primary NK cells but using a different sgRNA design and expansion approach.27,29,30 The resulting CD38 KD eNK cells continued to expand and displayed reduced fratricide after CD38 CAR expression. With the availability of CRISPR/Cas9 and the relative ease of application to primary NK cells using clinically adaptable platforms now demonstrated by multiple groups, there are vast possibilities for this technology across NK cell therapeutics. One potential limitation to CD38 targeting in AML is the limited capacity to target LSC populations, questioning the ‘curative’ potential of the therapies. LSC in AML are well-established, and while our understanding has evolved to include the existence of some CD38-positive LSC populations, it is likely that many LSC do reside within the traditional CD34-positive, CD38-negative compartment.31 Considering this feature of AML LSC, a CAR-NK cell targeting CD38 could be expected to have greater LSC targeting potential than a CAR-T cell, because of the presence of the innate activating pathways of NK cells above and beyond the CD38-specific CAR. Indeed the potential for long-term disease control, and thus LSC targeting capabilities can be inferred from data establishing the importance of NK cell KIR-ligand mismatch in the efficacy of allogeneic stem cell transplantation.16 Furthermore, a tandem CAR approach including a LSC-specific antigen and/or a variant of TRAIL (tumor necrosis factor related apoptosis inducing ligand) could be incorporated to augment LSC targeting.32 Tailored approaches using CAR modified NK cells targeting combinations based on the specific identified LSC immunophenotype in each case may ultimately be required given the absence of an identified universal LSC marker. This approach is becoming feasible with current and emerging technologies. Antibody- and protein-based approaches have been considered previously in attempts to overcome fratricide in a CD38-directed CAR-T cell platform.33 CRISPR/Cas9generated, CD38 KD eNK cells have recently and successfully been applied to reducing the NK cell fratricidal effects of daratumumab with a focus on multiple myeloma.27 Interestingly, while a magnetic separation step was utilized to enhance the purity of the KD population in this innovative study, our data suggest that expression of a CD38 CAR combined with a highly efficient CRISPR/ Cas9 KD will likely lead to a self-selecting KD population without additional processing. While not the focus of our experiments, Kararoudi et al.27 also explored the cellular bioenergetic benefit of deletion of CD38 in eNK cells. CD38 converts nicotinamide adenine dinucleotide (NAD+) to cyclic adenosine diphosphate-ribose through an enzymatic function. Additional NAD+ availability due to loss of CD38 supplies an important co-factor favoring oxidative 444

phosphorylation within NK cells. FT538, a NK cell product derived from induced pluripotent stem cells being developed by FATE Therapeutics, incorporates a CD38 deletion to overcome fratricide when combined with daratumumab. The group also demonstrated greater resistance to oxidative stress conferred by deletion of CD38, a characteristic likely to be favorable within the tumor microenvironment.34 These enhancements to NK cell biology suggest a broad range of applications for CD38 KD eNK cells beyond CD38 targeting and fratricide concerns. Simple and consistent approaches to their generation will likely be of clinical utility. In conclusion, we present two viable approaches to CD38 CAR-NK cell therapies applied to AML. Both our CD38 CAR-KHYG-1 cells and CD38 KD eNK cell platforms overcome effector cell fratricide relating to NK cell CD38 expression. Furthermore, we report an efficient approach to CRISPR/Cas9 genome editing adapted to primary eNK cells and suitable for GMP expansion. Disclosures MG has received educational funding from Janssen Pharmaceuticals and Takeda. AS and SS have received research funding from ONK Therapeutics Limited. LKM is an employee of ONK Therapeutics Limited. SK and RS are employees of Maxcyte Inc. SZ has received research funding from Takeda, Celgene, and Janssen and is a member of the board of directors or an advisory committee for Takeda, Celgene, and Janssen. NWCJvdD has received research funding from Janssen Pharmaceuticals, Amgen, Celgene, Novartis, and BMS and has participated in advisory boards for Janssen Pharmaceuticals, Amgen, Celgene, BMS, Takeda, Roche, Novartis, Bayer, and Servier. TM has received research funding from Gilead, Celgene, Novartis, ONK Therapeutics Limited, Genmab, Janssen and has been a member of an advisory board for Janssen. ES has collaborated in research projects with Janssen, Roche, Celgene, and Takeda. MOD has received research funding from ONK Therapeutics Limited, BMS, Celgene, and Glycomimetics; is a member of the board of directors or an advisory committee for Janssen, Abbvie, and ONK Therapeutics Limited; and owns equity in ONK Therapeutics Limited. EN has no potential conflicts of interest to disclose. Contributions SS and MOD conceived the research. EN, MG, AS and LKM performed functional assays. SK and RS contributed to electroporation optimization. ES contributed to acquisition of patients’ samples and the primary AML assay design. TM, SZ and NVD developed CD38 CAR-KHYG1 cells and associated functional assays. MG, SS and AS wrote the manuscript and prepared the figures. All authors contributed to editing and reviewing the final manuscript prior to submission. Acknowledgments Reagents for electroporation were contributed by Maxcyte inc. Funding This research was supported by Irish Clinical Academic Training (ICAT) Programme fellowship funding to MG. ICAT is supported by the Wellcome Trust and the Health Research Board (grant number 203930/B/16/Z), the Health Service Executive National Doctors Training and Planning and the Health and Social Care, Research and Development Division, Northern Ireland. Work performed in collaboration with Blood Cancer Network Ireland was funded by Science Foundation Ireland and the Irish Cancer Society (Blood Cancer Network Ireland, 14/ICS/B3042). haematologica | 2022; 107(2)

CD38 CAR-NK cells targeting AML

References 1. Surveillance, Epidemiology, and End Results (SEER) Program (1969-2018). U.S. Population Data with Other Software: (SEER Web site: www.seer.cancer.gov/popdata), National Cancer Institute, DCCPS, Surveillance Research Program, released December 2019. 2. DiNardo CD, Stein EM, de Botton S, et al. Durable remissions with ivosidenib in IDH1-mutated relapsed or refractory AML. N Engl J Med. 2018;378(25):2386-2398. 3. Stein EM, DiNardo CD, Pollyea DA, et al. Enasidenib in mutant IDH2 relapsed or refractory acute myeloid leukemia. Blood. 2017;130(6):722-731. 4. Stone RM, Mandrekar SJ, Sanford BL, et al. Midostaurin plus chemotherapy for acute myeloid leukemia with a FLT3 mutation. N Engl J Med. 2017;377(5):454-464. 5. Maude SL, Laetsch TW, Buechner J, et al. Tisagenlecleucel in children and young adults with B-cell lymphoblastic leukemia. N Engl J Med. 2018;378(5):439-448. 6. 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. 7. Cummins KD, Gill S. Chimeric antigen receptor T-cell therapy for acute myeloid leukemia: how close to reality? Haematologica. 2019;104(7):1302-1308. 8. Kenderian SS, Ruella M, Shestova O, et al. Targeting CLEC12A with chimeric antigen receptor T cells can overcome the chemotherapy refractoriness of leukemia stem cells. Biol Blood Marrow Transplant. 2017;23(3):S247-S248. 9. Petrov JC, Wada M, Pinz KG, et al. Compound CAR T-cells as a doublepronged approach for treating acute myeloid leukemia. Leukemia. 2018;32(6):1317-1326. 10. Naik J, Themeli M, de Jong-Korlaar R, et al. CD38 as a therapeutic target for adult acute myeloid leukemia and T-cell acute lymphoblastic leukemia. Haematologica. 2019; 104(3):e100-e103. 11. Drent E, Groen RWJ, Noort WA, et al. Preclinical evaluation of CD38 chimeric antigen receptor engineered T cells for the treatment of multiple myeloma. Haematologica. 2016;101(5):616-625. 12. Mihara K, Yanagihara K, Takigahira M, et al. Activated T-cell-mediated immunotherapy with a chimeric receptor against CD38 in Bcell non-Hodgkin lymphoma. J Immunother.

haematologica | 2022; 107(2)

2009;32(7):737-743. 13. Drent E, Themeli M, Poels R, et al. A rational strategy for reducing on-target off-tumor effects of CD38-chimeric antigen receptors by affinity optimization. Mol Ther. 2017;25(8):1946-1958. 14. Björklund AT, Carlsten M, Sohlberg E, et al. Complete remission with reduction of highrisk clones following haploidentical NK-cell therapy against MDS and AML. Clin Cancer Res. 2018;24(8):1834-1844. 15. Nguyen R, Wu H, Pounds S, et al. A phase II clinical trial of adoptive transfer of haploidentical natural killer cells for consolidation therapy of pediatric acute myeloid leukemia. J Immunother Cancer. 2019;7 (1):81. 16. Ruggeri L, Capanni M, Urbani E, et al. Effectiveness of donor natural killer cell alloreactivity in mismatched hematopoietic transplants. Science. 2002;295(5562):20972100. 17. Liu E, Marin D, Banerjee P, et al. Use of CAR-transduced natural killer cells in CD19positive lymphoid tumors. N Engl J Med. 2020;382(6):545-553. 18. Wang Y, Zhang Y, Hughes T, et al. Fratricide of NK cells in daratumumab therapy for multiple myeloma overcome by ex vivoexpanded autologous NK cells. Clin Cancer Res. 2018;24(16):4006-4023. 19. Suck G, Branch DR, Smyth MJ, et al. KHYG1, a model for the study of enhanced natural killer cell cytotoxicity. Exp Hematol. 2005;33(10):1160-1171. 20. Uruno A, Noguchi N, Matsuda K, et al. Alltrans retinoic acid and a novel synthetic retinoid tamibarotene (Am80) differentially regulate CD38 expression in human leukemia HL-60 cells: possible involvement of protein kinase C-δ. J Leukoc Biol. 2011;90(2):235-247. 21. Tang X, Yang L, Li Z, et al. First-in-man clinical trial of CAR NK-92 cells: safety test of CD33-CAR NK-92 cells in patients with relapsed and refractory acute myeloid leukemia. Am J Cancer Res. 2018;8(6):10831089. 22. Zabaleta A, Tomas J, Simoes C, et al. The mode of action of the anti-CD38 monoclonal antibody (MAB) isatuximab in elderly acute myeloid leukaemia (AML). Hemasphere. 2020;4(S1):Abstract Book EP467. 23. Krejcik J, Casneuf T, Nijhof IS, et al. Daratumumab depletes CD38+ immune regulatory cells, promotes T-cell expansion,

and skews T-cell repertoire in multiple myeloma. Blood.2016;128(3):384-394. 24. Zhao C, Jia B, Wang M, et al. Multi-dimensional analysis identifies an immune signature predicting response to decitabine treatment in elderly patients with AML. Br J Haematol. 2020;188(5):674-684. 25. Farber M, Arnold L, Chen Y, Möllmann M, Duehrsen U, Hanoun M. Inhibition of CD38 shows anti-leukemic activity in acute myeloid leukemia. Blood. 2018;132(Suppl 1):1456-1456. 26. Mouly E, Planquette C, Rousseau E, Delansorne R. Inecalcitol respectively induces or increases CD38 expression at the surface of CD38- or CD38+ AML cell lines representative of all 9 FAB subtypes except M6. Cancer Res. 2018;78(13 Suppl):1890. 27. Naeimi Kararoudi M, Nagai Y, Elmas E, et al. CD38 deletion of human primary NK cells eliminates daratumumab-induced fratricide and boosts their effector activity. Blood. 2020;136(21):2416-2427. 28. Casneuf T, Xu XS, Adams HC, et al. Effects of daratumumab on natural killer cells and impact on clinical outcomes in relapsed or refractory multiple myeloma. Blood Adv. 2017;1(23):2105-2114. 29. Pomeroy EJ, Hunzeker JT, Kluesner MG, et al. A genetically engineered primary human natural killer cell platform for cancer immunotherapy. Mol Ther. 2019;28(1):5263. 30. Kararoudi MN, Dolatshad H, Trikha P, et al. Generation of knock-out primary and expanded human NK cells using Cas9 ribonucleoproteins. J Vis Exp. 2018;2018 (136):58237. 31. Taussig DC, Miraki-Moud F, Anjos-Afonso F, et al. Anti-CD38 antibody-mediated clearance of human repopulating cells masks the heterogeneity of leukemia-initiating cells. Blood. 2008;112(3):568-575. 32. Szegezdi E, Reis CR, Sloot AM van der, et al. Targeting AML through DR4 with a novel variant of rhTRAIL. J Cell Mol Med. 2011;15(10):2216-2231. 33. Gao Z, Tong C, Wang Y, Chen D, Wu Z, Han W. Blocking CD38-driven fratricide among T cells enables effective antitumor activity by CD38-specific chimeric antigen receptor T cells. J Genet Genomics. 2019;46(8):367-377. 34. Cichocki F, Woan K, Wu C-Y, et al. NK cells lacking CD38 are resistant to oxidative stress-induced death. Blood. 2019;134(Supp 1):3215-3215.

445

ARTICLE Ferrata Storti Foundation

Cell Therapy & Immunotherapy

Successful gene therapy of Diamond-Blackfan anemia in a mouse model and human CD34+ cord blood hematopoietic stem cells using a clinically applicable lentiviral vector Yang Liu,1 Maria Dahl,1 Shubhranshu Debnath,1 Michael Rothe,2 Emma M. Smith,1 Tan Hooi Min Grahn,1 Sarah Warsi,1 Jun Chen,1 Johan Flygare,1 Axel Schambach2,3 and Stefan Karlsson1

Haematologica 2022 Volume 107(2):446-456

Molecular Medicine and Gene Therapy, Lund Stem Cell Center, Lund University, Lund, Sweden; 2Institute of Experimental Hematology, Hannover Medical School, Hannover, Germany; and 3Division of Hematology/Oncology, Boston Children’s Hospital, Harvard Medical School, Boston, MA, USA

ABSTRACT

Correspondence: STEFAN KARLSSON stefan.karlsson@med.lu.se YANG LIU yang.liu@med.lu.se Received: August 27, 2020. Accepted: December 23, 2020. Pre-published: January 14, 2021. https://doi.org/10.3324/haematol.2020.269142

446

iamond-Blackfan anemia (DBA) is an inherited bone marrow failure disorder in which pure red blood cell aplasia is associated with physical malformations and a predisposition to cancer. Twentyfive percent of patients with DBA have mutations in a gene encoding ribosomal protein S19 (RPS19). Our previous proof-of-concept studies demonstrated that DBA phenotype could be successfully treated using lentiviral vectors in Rps19-deficient DBA mice. In our present study, we developed a clinically applicable single gene, self-inactivating lentiviral vector, containing the human RPS19 cDNA driven by the human elongation factor 1a short promoter, which can be used for clinical gene therapy development for RPS19-deficient DBA. We examined the efficacy and safety of the vector in a Rps19-deficient DBA mouse model and in human primary RPS19deficient CD34+ cord blood cells. We observed that transduced Rps19-deficient bone marrow cells could reconstitute mice long-term and rescue the bone marrow failure and severe anemia observed in Rps19-deficient mice, with a low risk of mutagenesis and a highly polyclonal insertion site pattern. More importantly, the vector can also rescue impaired erythroid differentiation in human primary RPS19-deficient CD34+ cord blood hematopoietic stem cells. Collectively, our results demonstrate the efficacy and safety of using a clinically applicable lentiviral vector for the successful treatment of Rps19-deficient DBA in a mouse model and in human primary CD34+ cord blood cells. These findings show that this vector can be used to develop clinical gene therapy for RPS19-deficient DBA patients.

Introduction Diamond-Blackfan anemia (DBA) is a congenital bone marrow (BM) failure disorder with erythroid hypoplasia that presents early in infancy.1-3 The classic hematologic profile of DBA consists of macrocytic anemia with reticulocytopenia, normal or decreased levels of neutrophils, and a variable platelet count.3,4 Additionally, patients with DBA can also manifest with non-hematologic symptoms such as physical abnormalities and a predisposition to cancer.2,5 The majority of DBA cases (60-70%) are caused by heterozygous loss-of-function mutations in genes coding for ribosomal proteins (RP), resulting in functional RP haploinsufficiency.2 Recent studies have also identified mutations in erythroid transcriptional factors GATA1 and TSR2 (a direct binding partner of RPS26) as a cause of the DBA phenotype.6-8 Mutations in RPS19 are the most common alterations among patients with putative causal mutations, contributing to over 25% of cases.2 The main therapeutic option for DBA patients is corticosteroids, with more than 80% of the subjects responding well during early stages of treatment. However, half of these patients become non-responsive to corticosteroid therapy over prolonged treatment and have to be given blood transfusions.9 Importantly,

haematologica | 2022; 107(2)

Gene therapy of DBA

none of the currently available treatments is curative, and the treatments are often accompanied by serious complications.1,2,10 Hematopoietic stem cell transplantation is currently the sole curative option for the treatment of DBA. This treatment is, however, limited by the availability of suitable donors and the potential for serious immunological complications.11 Gene therapy, using gene-corrected hematopoietic stem cells, would be a potential alternative therapeutic strategy, as highlighted in our previous studies.12-16 For clinical applications using this approach, the efficiency of transgene expression and safety aspects, including potential insertional mutagenesis, should be addressed.17-19 Our group recently demonstrated that correction of Rps19-deficient BM cells using lentiviral vectors containing a clinically relevant promoter could rescue BM failure and defects in erythroid development, while exhibiting limited risk of insertional mutagenesis.12 In our previous study, we utilized a lentiviral vector containing the RPS19 gene as well as a GFP marker. We have subsequently designed a clinically applicable single gene lentiviral self-inactivating (SIN) vector for the clinical development of gene therapy for Rps19-deficient DBA patients. This vector harbors a codon-optimized human RPS19 cDNA driven by the short human elongation factor 1a promoter and lacks a fluorescent marker. In this study, we demonstrate that this vector can rescue the anemia and lethal BM failure observed in mouse models of Rps19-deficient DBA, with a low-risk insertion profile and no evidence of clonal expansion associated with vector integration near cancer-associated genes. We also observed the rescue of impaired erythroid differentiation in human RPS19-deficient CD34+ cord blood cells treated with this vector. Our results demonstrate the feasibility and preclinical efficacy for treatment of RPS19-deficient DBA using a clinically applicable SIN lentiviral vector, which opens the possibility for the development of clinical gene therapy for RPS19-deficient DBA patients.

mented with penicillin/streptomycin (Gibco), murine stem cell factor (100 ng/mL; PeproTech), and human thrombopoietin (50 ng/mL; PeproTech) in six-well plates at the concentration of 0.5x106 cells/mL. For transduction, retronectin-coated (20 ng/mL; Takara) 12-well plates were preloaded with the viral vectors (multiplicity of infection [MOI]=5-10), followed by seeding of 0.5x106 cells into each well filled with 1 mL pre-stimulation medium.

Flow cytometry A complete description of all antibodies used is listed in the Online Supplementary Methods.

Human primary cord blood cells and erythroid differentiation Human cord blood samples were obtained from the maternity wards of Helsingborg General Hospital and Skåne University Hospital in Lund and Malmö, Sweden, after informed, written consent according to guidelines approved by the regional ethical committee. Mononuclear cells were separated through densitygradient centrifugation. CD34+ cells were magnetically isolated according to the manufacturer’s description (Milteny Biotec, cat. n. 130-046-702). Cells were cultured in serum-free expansion medium (Stem Cell Technologies), supplemented with human stem cell factor, thrombopoietin, and FLT3-ligand at 100 ng/mL from Peprotech. Full descriptions of transduction and erythroid differentiation are provided in the Online Supplementary Methods.

Other experimental details Full descriptions of the quantitative reverse transcriptase polymerase chain reaction (qRT-PCR), determination of transduction efficiency and vector copy number measurements are provided in the Online Supplementary Methods.

Insertion site analysis

Methods

Whole BM cells were isolated at 16 weeks after transplantation. Genomic DNA was isolated from the BM of flushed femora using the DNA Blood & Tissue kit (Qiagen). The vectorgenome junction was amplified using the INtegration Site PIpeline for paIRed-End reaDs (INSPIIRED) workflow as described by Sherman and colleagues.24

Lentiviral vector constructs

Statistical analysis

The SIN lentiviral vector is derived from the pRRL.PPT.PGK. vector backbone.20 A codon-optimized human RPS19 cDNA was designed as described previously21 and inserted downstream of the EFS promoter. Lentiviral vectors were produced by the Vector Unit at Lund University as previously described.12

t-tests and one-way analysis of variance with the Tukey multiple-comparison test were used to determine statistical significance. Computations were performed using GraphPad Prism (version 6; GraphPad Software).

Mice and transplantations

Results

Mice were maintained at the Lund University animal facility and all animal experiments were approved by the Lund University animal ethics committee. The homozygous doxycycline-inducible Rps19-deficient mouse model used in the study was established as previously described.15 A detailed description of transplantations is provided in the Online Supplementary Methods.

Transduction c-kit+ or lineage negative (Lin–) cells isolated from BM of transgenic mice were enriched by using CD117 or Lin– microbeads and magnetic-activated cell sorting separation columns (all from Miltenyi Biotec) according to the manufacturer’s protocol. After enrichment, cells were pre-stimulated for 24 h in StemSpan serum-free expansion medium (Stem Cell Technologies), supple-

haematologica | 2022; 107(2)

High transduction efficiency of the EFS-RPS19 vector We first studied the transduction efficiency of the vector in BM progenitor cells isolated from our established Rps19-deficient DBA mouse model.15 This model contains the Rps19-targeting shRNA expressed under a doxycycline-responsive promoter located downstream of the collagen A1 gene (Figure 1A). Experimental animals were bred to be either heterozygous (D/+) or homozygous (D/D) for the shRNA to generate two models with intermediate or severe Rps19 deficiency (Figure 1B). Rps19 mRNA expression was reduced by approximately 50%, and a trend toward more efficient knockdown in D/D mice compared to D/+ mice was seen, as shown in our previous studies.15 Upon induction with doxycycline, 447

Y. Liu et al.

transplanted recipients receiving D/D BM cells developed an acute and lethal BM failure, while recipients of D/+ BM cells developed a mild chronic anemia.15 As D/D mice develop lethal BM failure shortly after doxycycline induction, the more severely affected D/D mice were used in the present study to investigate whether the lethal phenotype could be fully rescued with the vector. The clinically applicable single gene lentiviral vector was developed using a SIN lentiviral vector design harboring the codon-optimized human RPS19 cDNA driven by an internal EFS promoter (named EFS-RPS19) (Figure 1C). Compared to the human codon-optimized RPS19 cDNA, there are six mismatches in the shRNA construct for generating the mouse model. Because of this, gene expression derived from the human codon-optimized RPS19 cDNA is not affected by the shRNA. We first examined the transduction efficiency of the vector. In our previous study, we transduced cells at a MOI of 10-20 and demonstrated the rescue of the anemia and BM failure with a reduced risk of insertional mutagenesis.12 To investigate the therapeutic effects of a lower average vector copy number per cell, we decided to use an MOI of 5-10. As shown in Online Supplementary Figure S1, the transduction efficiency was 75% on average in ckit+ cells isolated from D/D mice. We next examined both endogenous Rps19 and vector-derived RPS19 mRNA expression in c--kit+ cells isolated from D/D and +/+ mice at a MOI of 5. As shown in Online Supplementary Figure S2, the endogenous Rps19 mRNA expression levels were significantly decreased in the cells isolated from D/D mice compared to the levels in cells from +/+ mice after doxycycline administration. Cells transduced with EFSRPS19 exhibited a 2.5-fold higher level of expression of human RPS19 mRNA compared to the endogenous Rps19 expression in the D/D group. Interestingly, transduced cells isolated from +/+ mice showed a significantly lower level of human RPS19 mRNA expression than the transduced cells isolated from D/D mice at the same MOI, indicating the internal physiological regulation of excess RPS19 production as reported by others.29,30 The overall results indicated that cells transduced with the EFS-RPS19 vector could successfully express the human RPS19 transgene

Gene-corrected bone marrow cells can rescue the Diamond-Blackfan anemia phenotype in vivo We next assessed the function of gene-corrected BM cells using the EFS-RSP19 vector in vivo. As shown in Figure 2A, uninduced (no doxycycline) c-kit+ BM cells from D/D mice (CD45.2) were transduced with the EFSRPS19 vector (MOI=5-10), and then transplanted into lethally irradiated wild-type B6SJL recipient mice (CD45.1/CD45.2, named the EFS-RPS19 group). Mice receiving uninduced c-kit+ BM cells without vector transduction were regarded as the mock group (negative control). Following engraftment and stable donor-derived reconstitution of the hematopoietic system, doxycycline was administrated to all recipients to induce the DBA phenotype. To determine whether the vector-treated cells could achieve a full correction, age-matched B6SJL wildtype (WT) mice receiving no irradiation and no transplantation but the same doxycycline administration were used as the control group. Before doxycycline administration, both the mock and EFS-RPS19 groups showed high overall donor reconstitution, indicating minimal to absent recipient-derived hematopoiesis (Online Supplementary Figure S3). After induction with doxycycline for 2 weeks, recipients in the mock group showed a dramatic decrease in red blood cell counts, mean corpuscular volume (MCV), and white blood cell and platelets counts, indicating that the mice developed BM failure shortly after doxycycline administration (Figure 2B-F). In contrast, recipients in the EFSRPS19 vector-treated group showed normal blood cellularity compared to the WT group. To assess the long-term therapeutic effects, recipients were administered doxycycline for 16 weeks (Figure 3A). As shown in Figure 3B, most of the recipients in the mock group died (9 out of 16) due to severe anemia or BM failure (data not shown) at 2-3 weeks after doxycycline administration. The few remaining recipients exhibited a macrocytic anemia phenotype with significantly reduced red blood cell counts and increased MCV at 16 weeks. The hemoglobin levels and platelet counts were also decreased compared to those in the WT group (Figure 3C-G). As expected, there was a significantly decreased expression of endogenous Rps19 in donor-derived BM cells from both the mock and

Figure 1. The inducible Rps19-deficient mouse model and structure of the EFS-RPS19 self-inactivating lentiviral vector. (A) Overview of modified loci. Black arrowheads indicate the transcriptional start sites. (B) Breeding strategy to adjust the level of Rps19 downregulation. Homozygous mice (D/D mice) are used in the project. (C) The self-inactivating lentiviral vector harboring a codon-optimized human RPS19 cDNA driven by human elongation factor 1a short (EFS) promoter. LTR: long terminal repeat; pA: polyadenylation signal; PPT: polypurine tract; RRE: Rev response element; SA: splice acceptor.

448

haematologica | 2022; 107(2)

Gene therapy of DBA

EFS-RPS19 groups after 16 weeks of doxycycline administration (Online Supplementary Figure S4). Strikingly, all the mice in the EFS-RPS19 group survived without any signs of anemia and with normal BM cellularity compared to the WT group. These results indicate that the lethal BM failure can be prevented by the vector. The vector copy number at 16 weeks after transplantation was 5.2±1.6 and 4.7±1.0 on average in gene-corrected cells isolated from peripheral blood and BM, respectively (Figure 4A, B). We also analyzed the fraction of myeloiderythroid compartments by flow cytometry (Online Supplementary Figure S5). Donor-derived hematopoiesis was observed in the EFS-RPS19 group, and the mean percentage of donor cells (CD45.2) in every progenitor population was significantly higher in the EFS-RPS19 group than in the mock group (Figure 4C-I). Unlike in the EFSRPS19 group, the transplanted cells in the mock group had limited reconstituting ability. In addition, we observed a significantly higher reconstitution of resident

recipient cells (CD45.1/CD45.2) in the few surviving mice in the mock group than in the EFS-RPS19 group, perhaps explaining why these animals did not develop severe BM failure.

EFS-RPS19 vector-treated Rps19-deficient bone marrow cells provide long-term reconstitution We next investigated whether the EFS-RPS19 vectortreated Rps19-deficient BM cells could generate long-term engraftment and reconstitution in doxycycline-induced lethally irradiated WT recipients (Figure 5A). To this end, D/D mice were induced with doxycycline for 1 week and red blood cell counts and hemoglobin levels were measured to confirm the DBA phenotype (Online Supplementary Figure S6). The isolated Lin– BM cells from doxycyclineinduced mice were transduced with the vector and transplanted into the doxycycline-induced lethally irradiated WT recipients (EFS-RPS19 group). Recipients receiving untransduced Rps19-deficient Lin– BM cells from doxycy-

Figure 2. Effective correction of anemia by the EFS-RPS19 vector at 2 weeks after induction of the Diamond-BLackfan phenotype. (A) The scheme of the uninduced genecorrected cell transplantation model and plan for examining short-term therapeutic effects. (B-F) Blood cellularity at 2 weeks after doxycycline induction (n=13-16, error bars represent the standard deviation, *P<0.05, **P<0.01, ***P<0.005 ****P<0.001 by one-way analysis of variance). BM: bone marrow; MOI: multiplicity of infection; WT: wild-type; RBC: red blood cells; MCV: mean corpuscular volume; WBC: white blood cells.

haematologica | 2022; 107(2)

449

Y. Liu et al.

cline-induced D/D mice were regarded as the mock group. The setting of the WT group was the same as described above. Doxycycline was administrated to all the recipients directly after transplantation. After induction with doxycycline for 2-3 weeks, the majority of mice in the mock group died (13 out of 16 animals) due to severe anemia or BM failure (Figure 5B, Online Supplementary Figure S7). The few surviving mice exhibited significantly decreased levels of endogenous Rps19 expression in donor-derived BM cells at 16 weeks after doxycycline administration (Online Supplementary Figure S8) with the concomitant development of a severe anemia phenotype in the mock group (Figure 5C-G). In contrast, all recipients in the EFS-RPS19 group survived with normal blood cellularity compared to the WT group. The vector copy number was on average 8.3±4.0 and 10.9±3.9 in gene-corrected Rps19-deficient cells isolated from peripheral blood and BM, respectively (Figure 6A, B). By analyzing the fraction of myeloid-erythroid compartments at 16 weeks, we observed almost complete donor-derived hematopoiesis in the EFS-RPS19

group, which was significantly higher than in the mock group (Figure 6C-I). Taken together, our results demonstrate that the EFS-RPS19 vector-treated group obtained full correction of anemia and BM failure.

Gene-corrected bone marrow cells showed polyclonal hematopoiesis and had a typical lentiviral insertion profile The risk of insertional mutagenesis is a major concern for future applications of gene therapy in the clinic. To assess the safety of the EFS-RPS19 vector integration profile, as well as the clonal dynamics of the transduced cells, insertion site analysis was performed using the INSPIIRED workflow.24 BM cells from uninduced donors (Figure 3A) of cohort 1 (animals 5-8) and cohort 2 (animals 13-16), or those from doxycycline-induced donors (Figure 5A) of cohorts 3 (animals 7-11) and cohort 4 (animals 17-20) were isolated 16 weeks after disease induction in the recipients. Detailed information on pool size estimation and sequence diversity in each sample is

Figure 3. Effective long-term correction of the anemia and bone marrow failure in mice treated with the EFS-RPS19 vector. (A) The scheme of the induced genecorrected cell transplantation model and the plan for examining long-term therapeutic effects. (B) Surviral rate analysis. (C-G) Blood cellularity at 16 weeks after doxycycline induction (n=13-16, error bars represent the standard deviation, *P<0.05, **P<0.01, ***P<0.005 by one-way analysis of variance). BM: bone marrow; MOI: multiplicity of infection; WT: wild-type; RBC: red blood cells; MCV: mean corpuscular volume; WBC: white blood cells.

450

haematologica | 2022; 107(2)

Gene therapy of DBA

shown in Online Supplementary Tables S2 and S3. Particularly, the top ten integrations with the highest sequence contribution in each sample are depicted in Figure 7A and Online Supplementary Figure S9A. Overall, EFS-RPS19-transduced cells showed a highly polyclonal insertion site pattern, reflecting the overall integration preferences of lentiviral vectors. No integration site contributed with more than 5.37% in uninduced gene-corrected cells (Figure 7A) or 4.4% in gene-corrected Rps19deficient cells (Online Supplementary Figure S9A) to the overall sequence pool, and there were no integrations in or close to known high-risk proto-oncogenes (Lmo2, Ccnd2 or Hmga2). However, one integration at 16.5 kb upstream of the high-risk locus Ikzf1 was detected (accounting for 1.65%), and another integration at 20.7 kb upstream of the high-risk gene Mecom was found (accounting for 0.04%) in recipients of gene-corrected Rps19-deficient BM cells. For the analysis of overlaps between integrations in or near the same genes among BM samples, we observed common lentiviral integration sites identified in previous integration site analysis (Online Supplementary Tables S4 and S5).31,32 We identified 11 integrations in or near the same refSeq genes between cohorts 1+2 and cohorts 3+4 (Online Supplementary Figure S10). Four of these shared common insertion sites in or near Hgf, Kdm6a, Lnpep and Mef2c also listed as protooncogenes in the All Onco database.25 The sole occurrence of integration sites in or near high-risk loci was not an indication of a higher risk of insertional mutagenesis if no dominant clones were detected. The detected integrations might simply reflect that lentiviral vectors were capable of integrating at these genomic sites. We performed analysis of insertion site profile including parameters of the integration site preferences close to CpG islands, GC-rich regions, in or near transcription units, the transcriptional start site of genes, gene boundaries or

proto-oncogenes (Figure 7B-D, Online Supplementary Figure S9B-D). More integrations were detected in a distance of 100 kb relative to CpG islands, marking actively transcribed regions but not in the direct vicinity of CpG islands (1-10 kb), hence not close to the promoter region. Our data showed that GC rich regions (marking promoter regions of genes) and long intergenic regions were generally disfavored by the vector. The integrations inside transcriptional units and in or close to proto-oncogenes (within a 100 kb window = onco.100k) are displayed relative to the matched random controls (Figure 7B, Online Supplementary Figure S9B).

The EFS-RPS19 vector rescued impaired erythroid differentiation of human RPS19-deficient CD34+ cord blood cells We next examined the therapeutic effects of the EFSRPS19 vector using human primary CD34+ cord blood cells. Since primary CD34+ cells from DBA patients are difficult to obtain, we utilized previously validated lentiviral shRNA vectors that silence RPS19 expression in human CD34+ cells to induce a DBA phenotype.23 Two lentiviral vectors expressing shRNA (shRNA1 and shRNA2) targeting different regions of the human RPS19 mRNA sequence were used to induce the DBA phenotype, and a vector expressing a scrambled shRNA sequence (Scr) was used as a healthy control. To ensure that the human construct in the vector would not be degraded by the targeting shRNA, alignments were performed and there were two mismatches (out of 19 nucleotides) for shRNA1 and four mismatches (out of 19 nucleotides) for shRNA2. Thus, it is very unlikely that the gene expression derived from the human codon-optimized RPS19 cDNA would be affected by either shRNA. Since the shRNA vectors also contain a GFP marker gene, transduced GFP+ cells were sorted for further examination.

Figure 4. Gene-corrected bone marrow cells show a competitive advantage in contributing to long-term hematopoiesis in vivo. (A, B) Vector copy number in peripheral blood (A) and bone marrow (B). (C, D) Donor reconstitution in peripheral blood (C) and bone marrow (D). (E–I) The percentage of transduced cells in hematopoietic stem cells (E), megakaryocyte progenitors (F), pre-granulocyte-macrophage and granulocyte-macrophage progenitors (G), pre-megakaryocyte-erythroid (H), and pre-colony-forming unit erythroid and colony-forming unit erythroid (I) (n=13-16, error bars represent the standard deviation, black asterisks indicate the statistical significance of the comparison of recipient-derived cells between the mock and EFS-RPS19 groups, orange asterisks indicate the statistical significance of the comparison of donor-derived cells between the mock and EFS-RPS19 groups. *P<0.05, **P<0.01, ***P<0.005, ****P<0.001 by one-way analysis of variance). VCN: vector copy number; PB: peripheral blood; BM: bone marrow; HSC: hematopoietic stem cells; MkP: megakaryocyte progenitors; pre-GM/GMP: pre-granulocyte macrophage and granulocyte macrophage progenitors; preMegE: pre-megakaryocyte-erythroid; preCFU-E/CFU-E: pre-colony-forming unit–erythroid (CFU-E)/CFU-E.

haematologica | 2022; 107(2)

451

Y. Liu et al.

and GFP– cells (Online Supplementary Figure S12A-B). By further analysis of cell distribution, significantly decreased outputs of CD71+CD235A– cells on day 6 and CD71– CD235A+ cells on day 10 were observed in GFPhigh populations of RPS19-deficient groups (Online Supplementary Figure S12C, D). The impaired differentiation was rescued by EFS-RPS19, with significantly increased GFPhigh populations (1.6-fold for shRNA1 and 1.8-fold for shRNA2) (Figure 8B) and progenitor populations (CD71+CD235A–) (Online Supplementary Figure S12C) compared to the RPS19-deficient groups on day 6. Particularly, during terminal erythropoiesis on day 16, cells in the RPS19-deficient groups showed reduced red blood cell production (especially in the shRNA2 group) and few GFPhigh cells (<1%) could be detected (Figure 8C-E). However, EFSRSP19 vector-treated groups produced more red blood cells and maintained significantly higher GFPhigh populations (~4%) than the RPS19-deficient groups. These results collectively demonstrate that the EFS-RPS19 vector can rescue the impaired erythroid differentiation in human primary RPS19-deficient CD34+ cord blood cells.

As shown in Figure 8A, both shRNA1 and shRNA2 significantly decreased RPS19 mRNA expression, with slightly more efficient knockdown being obtained with shRNA1 than shRNA2, which resembles our previous observations in D/+ and D/D mouse models.15 In order to examine the function of EFS-RPS19, we transduced sorted GFP+ cells and cultured them in erythroid differentiation medium for 48 h after transduction. The PCR results indicated successful integration of the vector into human cells, as shown in Online Supplementary Figure S11. It is a well-known phenomenon that lentiviral vector-mediated transgene expression is silenced in a fraction of CD34+ cells, and that this fraction increases during differentiation of cells, likely due to changes in chromatin accessibility. As shown in Figure 8B, transgene silencing was evident in a fraction of the Scr-transduced CD34+ cells in which the fraction of GFPhigh shRNA expressing cells decreased from 100% in the sorted cells to around 90% on day 6 and day 10, with a further reduction to around 25% at day 16. However, in the RPS19-deficient groups, the loss of GFPhigh cells was evident from day 6, with an increased fraction of GFPlow

Figure 5. Amelioration of disease phenotype in Rps19-deficient animals transplanted with gene-corrected cells. (A) Scheme of the gene-corrected Rps19-deficient cell transplantation model and plan for examining short-term and long-term therapeutic effects. (B) Survival rate analysis. (C-G) Blood cellularity at 4 and 16 weeks after doxycycline induction (n=14-16, error bars represent the standard deviation, *P<0.05, **P<0.01, ***P<0.005 ****P<0.001 by one-way analysis of variance). WT: wild-type; RBC: red blood cells; MCV: mean corpuscular volume; WBC: white blood cells.

452

haematologica | 2022; 107(2)

Gene therapy of DBA

Discussion Currently, hematopoietic stem cell transplantation is the sole curative option for DBA patients, but suitable donors are often unavailable and there can be serious immunological complications.3,33 Gene therapy using gene-corrected hematopoietic stem cells has been shown to be a promising therapeutic strategy for genetic blood disorders in recent years.19,34,35 Our previous proof-of-concept studies also demonstrated the feasibility of applying gene therapy to cure DBA.12-16 In the present study, a clinically applicable lentiviral vector was used to investigate the efficacy and safety for treating anemia and lethal BM failure in Rps19-deficient mice and for ameliorating the impaired erythroid differentiation in human primary RPS19-deficient CD34+ cord blood cells. The next step towards clinical gene therapy will be to perform toxicology and biodistribution analyses, and thereafter proceed with submitting an application to regulatory authorities in order to initiate a phase I/II clinical trial with 6-12 patients focusing on safety. For successful development of clinical gene therapy, vector efficacy, generating long-term therapeutic effects, is crucial. To examine the therapeutic effects of the vector with lower copies per cell, we decreased the MOI to 5-10 in the present study and demonstrated that the EFS-RPS19 vector has robust therapeutic effects with no evidence of clonal expansion associated with vector integration near cancer-associated genes, as we showed in our previous study.12 It has already been shown that ribosomal proteins are produced in excess of the needs of the ribosome assembly, and that the excess protein is subjected to proteasomal degradation.22,29,30 Similarly, in the present study, our transduced Rps19-deficient cells had physiological levels of expression of RPS19. Hence, it is unlikely that the ectopic expression of RPS19 would promote uncontrolled

growth. Since single-copy insertion of the therapeutic gene in the target cells is suggested to avoid the risk of genotoxicity in clinical gene therapy manipulation,19 we plan to examine the therapeutic effects using transduced cells with a lower MOI (e.g. MOI=1) in future studies. Our results also demonstrated that the vector could rescue the impaired erythroid differentiation of RPS19-deficient cord blood cells by increasing red blood cell production. Overall, we showed that the EFS-RPS19 vector could rescue the anemia and BM failure of RPS19-deficient DBA. Apart from efficacy, vector safety is the other essential factor to assess when applying gene therapy. The risk of insertional mutagenesis is a concern for future applications of gene therapy in the clinic. To prevent this risk, we utilized a third-generation SIN lentiviral vector that lacks potent enhancers in the long terminal repeat regions, since such vectors were shown to exhibit a safer integration profile in previous clinical trials19,34,36 and also in our previous animal studies.12,15 By using the state-of-the-art INSPIIRED workflow, which can provide better quantification of clonal abundance compared to a linear amplificationmediated PCR approach, we found that gene-corrected BM cells in both models exhibited a low risk of mutagenesis with no evidence of clonal expansion associated with vector integration near cancer-associated genes. In our study, no hematologic abnormalities were observed due to enforced expression of RPS19. The results collectively demonstrate the safety of the EFS-RPS19 vector for clinical gene therapy development. The bioinformatic pipeline of the INSPIIRED workflow is a more automated approach, making it well suited for monitoring patients in gene therapy trials in the future. Increased MCV, due to macrocytic anemia, is a classic clinical observation in patients with DBA. It is, at least in part, caused by the stabilization of p53 and activation of p53 targets (e.g., p21, Bax), which are responsible for cell

Figure 6. EFS-RPS19 vector-treated Rps19-deficient cells show a competitive advantage in contributing to long-term hematopoiesis in vivo. (A, B) Vector copy number in peripheral blood (A) and bone marrow (B). (C, D) Donor reconstitution in peripheral blood (C) and bone marrow (D). (E–I) The percentage of transduced cells in hematopoietic stem cells (E), megakaryocyte progenitors (F), pre-granulocyte-macrophage and granulocyte-macrophage progenitors (G), pre-megakaryocyte-erythroid (H), and pre-colony-forming unit erythroid and colony-forming unit erythroid (I) (n=14-16, error bars represent the standard deviation, black asterisks indicate the statistical significance of the comparison of recipient-derived cells between the mock and EFS-RPS19 groups, orange asterisks indicate the statistical significance of the comparison of donor-derived cells between the mock and EFS-RPS19 groups. *P<0.05, **P<0.01, ***P<0.005, ****P<0.001 by one-way analysis of variance). VCN: vector copy number; PB: peripheral blood; BM: bone marrow; HSC: hematopoietic stem cells; MkP: megakaryocyte progenitors; pre-GM/GMP: pregranulocyte macrophage and granulocyte macrophage progenitors; preMegE: pre-megakaryocyte-erythroid; preCFU-E/CFU-E: pre-colony-forming unit –erythroid (CFUE)/CFU-E.

haematologica | 2022; 107(2)

453

Y. Liu et al.

cycle arrest in G0/G1 phases leading to larger cell size.9,37 We also observed a significantly increased MCV in the mock group after induction with doxycycline for more than 4 weeks (Figures 3E and 5E). In particular, during the initial phase after doxycycline induction, the majority of red blood cells in the circulation were produced before the DBA phenotype was induced. Since old red blood cells are smaller than the newly produced cells, the MCV is decreased compared to normal for a short period during the first 2 weeks after doxycycline induction (Figure 2E). In addition to the severe anemia phenotype after induction of Rps19 deficiency, we observed decreased white blood cell and platelet counts in the mock group. Mild thrombocytopenia and neutropenia (low levels of neutrophilic granulocytes) have also been observed in about 25% of DBA patients during the course of the disease.4,37,38 In addition, several patients with RPS19-deficient DBA have developed myelodysplastic syndrome with multilineage cytopenia, which suggests a multilineage defect.37 This observation correlates with a reduction in the

absolute numbers of hematopoietic stem and progenitor cells in BM due to Rps19 deficiency,15 which led to the lethal BM failure we observed in untreated animals shortly after doxycycline administration. This is also supported by the limited reconstitution ability of progenitor compartments in the mock group (Figure 4C-I) and impaired erythroid differentiation of human primary RPS19-deficient cord blood cells. RPS19-deficient patients who develop thrombocytopenia and neutropenia also experience similar progressive phenotypes of hypocellularity in the BM.38,39 Moreover, DBA is a very heterogeneous disease. It is unknown why family members with the same genetic mutation in RPS19 may have very different phenotypes, ranging from no anemia to severe anemia with progression to multilineage BM failure.40 The current understanding of phenotype-genotype correlations is far from comprehensive and needs to be studied further. Although the majority of mice in the mock group died, a few mice did survive until the planned endpoint at 16 weeks. One of the possible reasons for this may be the

Figure 7. Gene-corrected bone marrow cells show a vector integration pattern that indicates low risk of mutagenesis and a highly polyclonal insertion site pattern. (A) The top ten integration sites in each sample (*indicates that the integration was within a transcription unit, ~ indicates that the insertion was within 50 kb of a cancer-related gene). (B, C) Percent of all integrations inside transcriptional units (B) and percent of integrations within 100 kb of proto-oncogenes compared to matched random control sites (C). (D) Genomic heatmap analysis of the insertion site profile. mrc: matched randon control. ***P<0.001 by an unpaired t-test.

454

haematologica | 2022; 107(2)

Gene therapy of DBA

Figure 8. Impaired erythroid differentiation of RPS19-deficient CD34+ cord blood cells can be rescued by the EFS-RPS19 vector. (A) RPS19 mRNA expression in CD34+ cord blood cells transduced with shRNA. (B) Percentage of GFPhigh population in RPS19-deficient CD34+ cord blood cells treated or not with EFS-RPS19 during erythroid differentiation from stage I to stage III. (C) Fluorescnce activated cell sorting analysis of erythroid differentiation of RPS19-deficient cells treated or not with EFS-RPS19 on day 16. (D) Percentage of indicated cell outputs of GFPhigh populations on day 16. (E) Red blood cell pellets at the end of stage III initiated with equal numbers of CD34+ cord blood cells (data shown as mean ± standard deviation, ^P<0.05 compared to the shRNA1 group, #P<0.05 compared to the shRNA2 group, *P<0.05, **P<0.01, ***P<0.005 by a t-test, 3 independent experiments).

emergence of resident hematopoietic stem cells derived from the recipients, which contribute to the reconstitution of hematopoiesis as a protective mechanism against stress-induced exhaustion in the BM.41-43 In support of this concept, recipient-derived hematopoietic stem cells were observed in BM of the mock group at 16 weeks after transplantation, even though full lethal irradiation was performed prior to transplantation. Other unknown reasons may also contribute to this observation and the underlying mechanism is unknown. As mentioned before, the majority of DBA patients have mutations in genes coding for ribosomal proteins, and 25% of them are RPS19-deficient (mostly because of point mutations or small deletions). Our findings indicate the possibility of developing SIN lentiviral vectors also tailored for other DBA mutations (e.g., RPL5) in the future. In conclusion, our data show the safety and efficacy of a clinically applicable SIN lentiviral vector for the successful treatment of Rps19-deficient DBA in our mouse model and in human primary CD34+ cord blood cells. We did not observe any hematologic abnormalities in vivo due to enforced expression of RPS19. Our present study suggests haematologica | 2022; 107(2)

that the clinically applicable SIN lentiviral vector, EFSRPS19, has the potential to be employed in a clinical gene therapy strategy for RPS19-deficient DBA patients. Disclosures No conflicts of interest to disclose. Contributions SK and YL conceived the project and directed the research; YL, MD, MR, ES, THMG, SW, and JC performed the experiments; YL, MR, AS, and SK analyzed the data; and YL and SK wrote the manuscript. Other co-authors provided feedback on the manuscript. Acknowledgments The authors thank Beata Lindqvist and Xiaojie Xian for lentivirus production, Zhi Ma for technical assistance, and Alexander Doyle for English language editing. Funding This work was supported by a Hemato-Linne grant from the Swedish Research Council Linnaeus, project grants from the 455

Y. Liu et al.

Swedish Research Council, the Swedish Cancer Society and the Swedish Children’s Cancer Society (to SK), the Tobias Prize awarded by the Royal Swedish Academy of Sciences financed by the Tobias Foundation, a clinical research grant from Lund

References 1. Lipton JM, Ellis SR. Diamond-Blackfan anemia: diagnosis, treatment, and molecular pathogenesis. Hematol Oncol Clin North Am. 2009;23(2):261-282. 2. Ulirsch JC, Verboon JM, Kazerounian S, et al. The genetic landscape of DiamondBlackfan anemia. Am J Hum Genet. 2018; 103(6):930-947. 3. Vlachos A, Muir E. How I treat DiamondBlackfan anemia. Blood. 2010;116(19):37153723. 4. Willig TN, Niemeyer CM, Leblanc T, et al. Identification of new prognosis factors from the clinical and epidemiologic analysis of a registry of 229 Diamond-Blackfan anemia patients. DBA group of Societe d'Hematologie et d'Immunologie Pediatrique (SHIP), Gesellshaft fur Padiatrische Onkologie und Hamatologie (GPOH), and the European Society for Pediatric Hematology and Immunology (ESPHI). Pediatr Res. 1999;46(5):553-561. 5. Narla A, Vlachos A, Nathan DG. Diamond Blackfan anemia treatment: past, present, and future. Semin Hematol. 2011;48(2):117123. 6. Gripp KW, Curry C, Olney AH, et al. Diamond-Blackfan anemia with mandibulofacial dystostosis is heterogeneous, including the novel DBA genes TSR2 and RPS28. Am J Med Genet A. 2014;164A(9):2240-2249. 7. Ludwig LS, Gazda HT, Eng JC, et al. Altered translation of GATA1 in Diamond-Blackfan anemia. Nat Med. 2014;20(7):748-753. 8. Sankaran VG, Ghazvinian R, Do R, et al. Exome sequencing identifies GATA1 mutations resulting in Diamond-Blackfan anemia. J Clin Invest. 2012;122(7):2439-2443. 9. Vlachos A, Ball S, Dahl N, et al. Diagnosing and treating Diamond Blackfan anaemia: results of an international clinical consensus conference. Br J Haematol. 2008;142(6):859876. 10. Boria I, Garelli E, Gazda HT, et al. The ribosomal basis of Diamond-Blackfan anemia: mutation and database update. Hum Mutat. 2010;31(12):1269-1279. 11. Myers KC, Davies SM. Hematopoietic stem cell transplantation for bone marrow failure syndromes in children. Biol Blood Marrow Transplant. 2009;15(3):279-292. 12. Debnath S, Jaako P, Siva K, et al. Lentiviral vectors with cellular promoters correct anemia and lethal bone marrow failure in a mouse model for Diamond-Blackfan anemia. Mol Ther. 2017;25(8):1805-1814. 13. Flygare J, Olsson K, Richter J, Karlsson S. Gene therapy of Diamond Blackfan anemia CD34(+) cells leads to improved erythroid development and engraftment following transplantation. Exp Hematol. 2008;36(11): 1428-1435. 14. Hamaguchi I, Ooka A, Brun A, Richter J, Dahl N, Karlsson S. Gene transfer improves

456

University Hospital (to SK), European Union project grants STEMEXPAND and PERSIST (to SK), a grant from The Royal Physiographic Society of Lund, Sweden (to YL), and a grant from Stiftelsen Lars Hiertas Minne (to YL).

erythroid development in ribosomal protein S19-deficient Diamond-Blackfan anemia. Blood. 2002;100(8):2724-2731. 15. Jaako P, Flygare J, Olsson K, et al. Mice with ribosomal protein S19 deficiency develop bone marrow failure and symptoms like patients with Diamond-Blackfan anemia. Blood. 2011;118(23):6087-6096. 16. Naldini L. Ex vivo gene transfer and correction for cell-based therapies. Nat Rev Genet. 2011;12(5):301-315. 17. Anguela XM, High KA. Entering the modern era of gene therapy. Annu Rev Med. 2019;70:273-288. 18. Biffi A, Bartolomae CC, Cesana D, et al. Lentiviral vector common integration sites in preclinical models and a clinical trial reflect a benign integration bias and not oncogenic selection. Blood. 2011;117(20): 5332-5339. 19. Cavazzana M, Bushman FD, Miccio A, Andre-Schmutz I, Six E. Gene therapy targeting haematopoietic stem cells for inherited diseases: progress and challenges. Nat Rev Drug Discov. 2019;18(6):447-462. 20. Dull T, Zufferey R, Kelly M, et al. A thirdgeneration lentivirus vector with a conditional packaging system. J Virol. 1998;72(11):8463-8471. 21. Moreno-Carranza B, Gentsch M, Stein S, et al. Transgene optimization significantly improves SIN vector titers, gp91phox expression and reconstitution of superoxide production in X-CGD cells. Gene Ther. 2009;16(1):111-118. 22. Jaako P, Debnath S, Olsson K, et al. Gene therapy cures the anemia and lethal bone marrow failure in a mouse model of RPS19deficient Diamond-Blackfan anemia. Haematologica. 2014;99(12):1792-1798. 23. Flygare J, Kiefer T, Miyake K, et al. Deficiency of ribosomal protein S19 in CD34+ cells generated by siRNA blocks erythroid development and mimics defects seen in Diamond-Blackfan anemia. Blood. 2005;105(12):4627-4634. 24. Sherman E, Nobles C, Berry CC, et al. INSPIIRED: a pipeline for quantitative analysis of sites of new DNA integration in cellular genomes. Mol Ther Methods Clin Dev. 2017;4:39-49. 25. Berry CC, Nobles C, Six E, et al. INSPIIRED: quantification and visualization tools for analyzing integration site distributions. Mol Ther Methods Clin Dev. 2017;4:17-26. 26. Chao A, Chazdon RL, Colwell RK, Shen TJ. Abundance-based similarity indices and their estimation when there are unseen species in samples. Biometrics. 2006;62(2): 361-371. 27. Haemmerle R, Phaltane R, Rothe M, et al. Clonal dominance with retroviral vector insertions near the ANGPT1 and ANGPT2 genes in a human xenotransplant mouse model. Mol Ther Nucleic Acids. 2014;3: e200. 28. DeLury BD. On the estimation of biological

populations. Biometrics. 1947;3(4):145-167. 29. Lam YW, Lamond AI, Mann M, Andersen JS. Analysis of nucleolar protein dynamics reveals the nuclear degradation of ribosomal proteins. Curr Biol. 2007;17(9):749-760. 30. Devlin EE, Dacosta L, Mohandas N, Elliott G, Bodine DM. A transgenic mouse model demonstrates a dominant negative effect of a point mutation in the RPS19 gene associated with Diamond-Blackfan anemia. Blood. 2010;116(15):2826-2835. 31. Cesana D, Ranzani M, Volpin M, et al. Uncovering and dissecting the genotoxicity of self-inactivating lentiviral vectors in vivo. Mol Ther. 2014;22(4):774-785. 32. Deichmann A, Brugman MH, Bartholomae CC, et al. Insertion sites in engrafted cells cluster within a limited repertoire of genomic areas after gammaretroviral vector gene therapy. Mol Ther. 2011;19(11):2031-2039. 33. Da Costa L, O'Donohue MF, van Dooijeweert B, et al. Molecular approaches to diagnose Diamond-Blackfan anemia: The EuroDBA experience. Eur J Med Genet. 2018;61(11):664-673. 34. High KA, Roncarolo MG. Gene therapy. N Engl J Med. 2019;381(5):455-464. 35. Dunbar CE, High KA, Joung JK, Kohn DB, Ozawa K, Sadelain M. Gene therapy comes of age. Science. 2018;359(6372):eaan4672. 36. Marktel S, Scaramuzza S, Cicalese MP, et al. Intrabone hematopoietic stem cell gene therapy for adult and pediatric patients affected by transfusion-dependent ss-thalassemia. Nat Med. 2019;25(2):234-241. 37. Da Costa L, Leblanc T, Mohandas N. Diamond-Blackfan anemia. Blood. 2020;136 (11):1262-1273. 38. Giri N, Kang E, Tisdale JF, et al. Clinical and laboratory evidence for a trilineage haematopoietic defect in patients with refractory Diamond-Blackfan anaemia. Br J Haematol. 2000;108(1):167-175. 39. Casadevall N, Croisille L, Auffray I, Tchernia G, Coulombel L. Age-related alterations in erythroid and granulopoietic progenitors in Diamond-Blackfan anaemia. Br J Haematol. 1994;87(2):369-375. 40. Engidaye G, Melku M, Enawgaw B. Diamond Blackfan anemia: genetics, pathogenesis, diagnosis and treatment. EJIFCC. 2019;30(1):67-81. 41. Shi W, Vu T, Boucher D, et al. Ssb1 and Ssb2 cooperate to regulate mouse hematopoietic stem and progenitor cells by resolving replicative stress. Blood. 2017;129 (18):24792492. 42. Baumgartner C, Toifl S, Farlik M, et al. An ERK-dependent feedback mechanism prevents hematopoietic stem cell exhaustion. Cell Stem Cell. 2018;22(6):879-892.e6. 43. Singh SK, Singh S, Gadomski S, et al. Id1 ablation protects hematopoietic stem cells from stress-induced exhaustion and aging. Cell Stem Cell. 2018;23(2):252-265.e8.

haematologica | 2022; 107(2)

ARTICLE

Complications in Hematology

Long term follow-up of pediatric-onset Evans syndrome: broad immunopathological manifestations and high treatment burden Thomas Pincez,1,2 Helder Fernandes,1,3 Thierry Leblanc,4 Gérard Michel,5 Vincent Barlogis,5 Yves Bertrand,6 Bénédicte Neven,7,8,9 Wadih Abou Chahla,10 Marlène Pasquet,11 Corinne Guitton,12 Aude Marie-Cardine,13 Isabelle Pellier,14 Corinne Armari-Alla,15 Joy Benadiba,16 Pascale Blouin,17 Eric Jeziorski,18 Frédéric Millot,19 Catherine Paillard,20 Caroline Thomas,21 Nathalie Cheikh,22 Sophie Bayart,23 Fanny Fouyssac,24 Christophe Piguet,25 Marianna Deparis,26 Claire Briandet,27 Eric Doré,28 Capucine Picard,9,29 Frédéric Rieux-Laucat,8,9 Judith Landman-Parker,30 Guy Leverger30 and Nathalie Aladjidi1,3 on the behalf of members of the French Reference Center for Pediatric Autoimmune Cytopenia (CEREVANCE) and of collaborators from the French Reference Center for Adult Autoimmune Cytopenia (CERECAI).

Ferrata Storti Foundation

Haematologica 2022 Volume 107(2):457-466

Centre de Référence National des Cytopénies Auto-immunes de l’Enfant (CEREVANCE), Bordeaux, France; 2Division of Pediatric Hematology-Oncology, Charles-Bruneau Cancer Center, Department of Pediatrics, Sainte-Justine University Hospital, Université de Montréal, Montréal, Québec, Canada; 3Pediatric Oncology Hematology Unit, University Hospital, Plurithématique CIC (CICP), Centre d’Investigation Clinique (CIC) 1401, INSERM, Bordeaux, France; 4Pediatric Hematology Unit, Robert Debré University Hospital, AP-HP, Paris, France; 5Department of Pediatric Hematology, La Timone Hospital, Marseille University Hospital, Marseille, France; 6Institute of Pediatric Hematology and Oncology, Lyon University Hospital, Lyon, France; 7Pediatric ImmunoHematology and Rheumatology Department, Necker-Enfants Malades University Hospital, AP-HP, Paris, France; 8Laboratory of Immunogenetics of Pediatric Autoimmune Diseases, Paris, France; 9Imagine Institute, UMR 1163 INSERM, University of Paris, Paris, France; 10Department of Pediatric Hematology, Jeanne de Flandre Hospital, Lille University Hospital, Lille, France; 11Pediatric Oncology Immunology Hematology Unit, Children’s University Hospital, Toulouse, France; 12Department of Pediatrics, Bicêtre University Hospital, AP-HP, Le Kremlin-Bicêtre, France; 13Department of Pediatric Hematology and Oncology, Rouen University Hospital, Rouen, France; 14Pediatric Unit, Angers University Hospital, Angers, France; 15Pediatric Oncology Hematology Unit, Grenoble University Hospital, Grenoble, France; 16Department of Hemato-Oncology Pediatric, Nice University Hospital, Nice, France; 17Department of Pediatric HematologyOncology, Clocheville Hospital, Tours University Hospital, Tours, France; 18Pediatric Oncology Hematology Unit, Arnaud de Villeneuve University Hospital, Montpellier, France; 19Department of Pediatric Hematology, Poitiers University Hospital, Poitiers, France; 20Department of Pediatric Hematology and Oncology, Hautepierre University Hospital, Strasbourg, France; 21Pediatric Hematology Unit, Nantes University Hospital, Nantes, France; 22Department of Pediatric Hematology-Oncology, Besançon University Hospital, Besançon, France; 23Pediatric Hematology Unit, Rennes University Hospital, Rennes, France; 24Pediatric Hematology Unit, Nancy University Hospital, Nancy, France; 25 Pediatric Oncology Hematology Unit, Limoges University Hospital, Limoges, France; 26 Pediatric Oncology-Hematology Unit, Caen University Hospital, Caen, France; 27 Department of Pediatrics, Dijon University Hospital, Dijon, France; 28Pediatric Unit, Clermont-Ferrand University Hospital, Clermont-Ferrand, France; 29Study Center for Primary Immunodeficiencies, Necker-Enfants Malades University Hospital, AP-HP, Paris, France and 30Pediatric Oncology Immunology Hematology Unit, Armand-Trousseau University Hospital, AP-HP, Paris, France

Correspondence: NATHALIE ALADJIDI nathalie.aladjidi@chu-bordeaux.fr Received: August 31, 2020. Accepted: December 22, 2020. Pre-published: January 14, 2021. https://doi.org/10.3324/haematol.2020.271106

ediatric-onset Evans syndrome (pES) is defined by both immune thrombocytopenic purpura (ITP) and autoimmune hemolytic anemia (AIHA) before the age of 18 years. There have been no comprehensive long-term studies of this rare disease, which can be associated to various immunopathological manifestations (IM). We report outcomes of the 151 patients with pES and more than 5 years of follow-up from the nationwide French prospective OBS’CEREVANCE cohort. Median age at final follow-up was 18.5 years (range, 6.8–50.0 years) and the median follow-up period was 11.3 years (range, 5.1–38.0 years). At 10 years, ITP and AIHA were in sustained complete

haematologica | 2022; 107(2)

457

T. Pincez et al. remission in 54.5% and 78.4% of patients, respectively. The frequency and number of clinical and biological IM increased with age: at the age of 20 years, 74% had at least one clinical IM (cIM). A wide range of cIM occurred, mainly lymphoproliferation, dermatological, gastrointestinal/hepatic and pneumological IM. The number of cIM was associated with a subsequent increase in the number of second-line treatments received (other than steroids and immunoglobulins; hazard ratio 1.4, 95% Confidence Interval: 1.15–1.60, P=0.0002, Cox proportional hazards method). Survival at 15 years after diagnosis was 84%. Death occurred at a median age of 18 years (range, 1.7–31.5 years), and the most frequent cause was infection. The number of second-line treatments and severe/recurrent infections were independently associated with mortality. In conclusion, long-term outcomes of pES showed remission of cytopenias but frequent IM linked to high second-line treatment burden. Mortality was associated to drugs and/or underlying immunodeficiencies, and adolescents-young adults are a high-risk subgroup.

Introduction

Methods

The presence of both immune thrombocytopenic purpura (ITP) and autoimmune hemolytic anemia (AIHA) defines Evans syndrome (ES). Pediatric-onset ES (pES) is a rare disease, and approximately ten new cases are diagnosed every year in France, which has a population of 66 million.1 Since its first description in 1951 by Robert Evans,2 our understanding of pES has been based on small retrospective cohorts with limited follow-up.3-7 In 2004, the French Rare Disease Center CEREVANCE set up the prospective national cohort OBS’CEREVANCE, which includes children with AIHA, chronic ITP persisting for more than 12 months (cITP), and pES.8 Preliminary reports from this cohort and previously published studies showed that pES is a chronic disease with a high rate of relapse for both types of cytopenias.1,3,4,7,9 Mortality rates across studies have ranged from 7–36%.1,3-7 In addition to cytopenias, immunopathological manifestations (IM) such as autoimmune/autoinflammatory organ diseases, lymphoproliferation, and hypogammaglobulinemia have been reported in 70–80% of patients with pES.4,5,8,10 In an undetermined number of cases, pES is thought to be “secondary” and caused by an underlying disease, classically systemic lupus erythematosus (SLE) or autoimmune lymphoproliferative syndrome (ALPS).1,11,12 Recently, genetic analyses found a heterogeneous genetic background in up to 65% of a subset of 80 patients from the OBS’CEREVANCE cohort. These patients carried variants in genes that are linked to primary immunodeficiencies (PID) or involved in immune responses.13 Overall, outcomes and the long-term course of pES are poorly understood. There have been no comprehensive longitudinal studies including both cytopenia and IM. In addition, the transition to adulthood is often particularly challenging for patients with chronic pediatric diseases.14 Adolescents–young adults (AYA) outcomes have not been investigated in patients with pES, and whether the disease improves with age is unknown. In a clinical setting, the possibility to identify high-risk patients would be extremely helpful in the management of this complex disease. Here, we describe the long-term course of hematological IM and treatments received throughout childhood into adulthood in patients with pES from the OBS’CEREVANCE cohort. We aimed to identify clinically relevant factors associated to the occurrence of IM, the number of second-line treatments received and mortality. Particularly, we investigated the impact of the number second-line treatment received and splenectomy on mortality.

OBS’CEREVANCE prospective national cohort

458

Inclusion and exclusion criteria are shown in the Online Supplementary Table S1.1,8,15 Data collected have been previously detailed.1,8 Patients were included if <18 years old at first cytopenia diagnosis. The coordinating center gathered and analyzed all data from the medical team in charge in real time, enabling prospective follow-up even after the pediatric-to-adult transition. The CEREVANCE group recommends scheduling clinical and biological follow-up at least every 6-12 months. Some patients underwent genetic analyses, as previously described.13 Written informed consent was obtained from parents and eligible patients. The cohort study was approved by the Institutional Ethics Committee (CPPRB-A; Bordeaux, France) and the database was registered with the national data protection authority (CNIL, 1396823V0).

Patient selection Patients with pES, defined as the simultaneous (within 1 month) or sequential association of ITP and AIHA, were included if at least 5 years of follow-up data were available after the first cytopenia diagnosis. In order to provide a complete mortality report, all patients, including those with less than 5 years of follow-up data, were included in the survival analyses. The data were extracted on 21 June 2019.

Definitions Initial cytopenia refers to the onset of ITP or AIHA (whichever occurred first) and does not take autoimmune neutropenia (AIN) into account. The IM categories were separated in clinical (cIM) and biological (bIM). pES was defined as secondary if a diagnosis of SLE or PID was made during the follow-up period. SLE diagnosis was made according to the Systemic Lupus International Collaborating Clinics Classification criteria (SLICC).16 ALPS diagnosis was based on international criteria.17 Second-line treatments were all immunomodulatory or immunosuppressive treatments, including splenectomy but excluding steroids and therapeutic intravenous immunoglobulins (IVIG). Sustained complete remission (CR) was defined as remission persisting until final follow-up, regardless of ongoing treatments. For analyses by age, patients were assessed annually from birth (for IM and treatments) or from cytopenia onset (for AIHA and ITP), until final follow-up. Occasional treatments (e.g., splenectomy and rituximab) were considered as ongoing if these occurred during the previous year. Further details are stated in the Online Supplementary Table S1.

Statistical analyses Continuous and categorical variables were compared using Wilcoxon–Mann–Whitney non-parametric test and Fisher’s exact test, respectively. Correlations were tested using the Pearson cor-

haematologica | 2022; 107(2)

Pediatric-onset Evans syndrome outcomes

relation coefficient. Survival and cumulative incidence estimates were based on the Kaplan–Meier method and compared using log-rank test. Patients of OBS’CEREVANCE cohort with isolated cITP or AIHA were used for comparison in survival analyses (unpublished data). The Cox proportional hazards method was used to analyze factors associated with time-dependent variables (i.e., time to CR, AIN, cIM, and second-line treatment, as well as survival). The potential cumulative and/or time-dependent nature of variables was taken in account. Proportionality of hazard was assessed for each variable. Logistic regression was used to analyze factors associated with severe or recurrent infections. Variables that were statistically significant in the univariate analyses were included in the multivariate analyses. We investigated associations with the following characteristics and events: sex, consanguinity, cIM/cancer in a first-degree relative, age at first and second cytopenia, sequence of cytopenia, AIN, hypogammaglobulinemia, time to ITP/AIHA CR, severe/recurrent infections, number of cIM, number of second-line treatments. The 95% Confidence Intervals (CI) for hazard ratios (HR) and odds ratios (OR) were not adjusted for multiple testing and should not be used to infer definitive effects. All tests were two-sided and a P-value <0.05 was considered statistically significant. Statistical analyses were performed using R (ver. 4.0; R Development Core Team) and GraphPad Prism (ver. 8; GraphPad Software, Inc., San Diego, CA, USA) software.

Results Population Of the 216 patients with pES, 151 were included in this study (Online Supplementary Figure S1). They were followed from 25 different centers. Patient characteristics are shown in Table 1. The median (min–max) follow-up time from the first cytopenia diagnosis was 11.3 years (range, 5.1–38.0 years). Median age at final follow-up was 18.5 years (range, 6.8–50.0 years). In 20 cases (15%), follow-up was discontinued because the patient was considered cured (n=11) or lost to follow-up (n=9). Median age at loss to follow-up was 18.4 years (range, 6.8-25.1 years).

Hematological outcomes AIN developed in 43 patients (28.5%). It was diagnosed within 1 month before or after first cytopenia onset in 23 of 43 cases (53.5%), more than 1 month before in two cases (4.7%), and more than 1 month after in 18 cases (41.9%; maximal delay, 12.4 years). In all cases, the diagnosis was made before the age of 18 years (median age, 6.8 years; range, 0.6–16.2 years). ITP and AIHA flare rates at 5 years of follow-up were calculated for the 61 alive patients who did not receive a second-line treatment during this period. Forty-eight patients (79%) had experienced an ITP flare and seven (11%) an AIHA flare. The proportion of patients achieving sustained CR for ITP and AIHA steadily increased after cytopenia onset (Figure 1A). At 5 and 10 years, ITP was in sustained CR in 40.5% and 62.3% of patients (P=0.02) and AIHA was in sustained CR in 54.5% and 74.1% of patients (P=0.001), respectively. Sustained CR was achieved earlier for AIHA than ITP (median time to CR, 4.0 years vs. 7.0 years; P=0.01). At the final follow-up of the 135 surviving patients, the numbers of patients in CR, partial remission, and no remission were 126 (83%), five (3%), and one (1%) for AIHA and 119 (79%), eight (5%), and five (3%) for ITP, respectively (missing data in three cases). Forty-six haematologica | 2022; 107(2)

Table 1. Patient characteristics.

Number of patients Sex ratio (male/female) Consanguinity, n (%) cIM/cancer in first-degree relative, n (%) Median age (years) at First cytopenia (min-max) ITP diagnosis (min-max) AIHA diagnosis (min-max) ES diagnosis (min-max) Sequence of ES Simultaneous, n (%) ITP then AIHA, n (%) AIHA then ITP, n (%) Time between first and second cytopenia (years) Median (min-max)** Direct antiglobulin test at AIHA diagnosis IgG, n (%) IgG + C3, n (%) Unspecified, n (%) C3, n (%) IgA + C3, n (%) IgM then IgG, n (%) Duration of follow-up after first cytopenia (years) Median (min-max) Mean ± SD Age at last follow-up (years) Median (min-max) Mean ± SD

151 1.40 (88/63) 12 (7.9) 43 (28) 5.4 (0.2-16.0) 6.7 (0.2-17.1)* 7.8 (0.2-21.5)* 8.9 (0.2-21.5) 52 (34.4) 62 (41.1) 37 (24.5) 2.5 (0.1-15.8) 73 (48.3) 61 (40.4) 11 (7.3) 4 (2.6) 1 (0.7) 1 (0.7) 11.3 (5.1-38.0) 12.5 ± 6.0 18.5 (6.8-50.0) 19.1 ± 6.8

*P=0.0076. **Considering the 99 patients with sequential cytopenias. CIM: clinical immunopathological manifestation; Ig: immunoglobulin; SD: standard deviation; ITP: immune thrombocytopenic purpura; AIHA: autoimmune hemolytic anemia; ES: Evans syndrome.

patients (34%) had no treatment ongoing at last followup. No particular characteristic was associated with AIHA or ITP CR, including cIM and bIM. Over the first three decades, the proportions of patients achieving sustained CR increased with age (Figure 1B). ITP and AIHA were in CR in 26% and 30% of cases at 10 years compared to 50% and 72% at 20 years, respectively (P<0.001 for both comparisons).

Immunopathological manifestations A total of 122 of 151 patients (81%) had at least one IM. The data for each category and specific diagnosis are shown in the Online Supplementary Table S2. cIM developed in 100 of 151 patients (66%). A total of 47 patients (31%) had two or more IM and 22 (15%) patients had three or more IM (Online Supplementary Figure S2A). Patients with no cIM had shorter median follow-up times (9.7 years vs. 13 years; P=0.0002) and were younger when data were collected (15 years vs. 20 years; P<0.0001). A cIM was diagnosed before the first cytopenia in 21 of 100, simultaneously in 13, and after in 66 cases (median delay, 3.7 years [range, 0.2–20.5 years]; Figure 2A). Among the 185 cIM, 29 (16%) were diagnosed before any second-line treatment. No cIM category had a statistically significant difference in frequency before and after first second-line treatment. The number of cIM increased 459

T. Pincez et al.

with age. At 10 compared to 20 years old, 37% and 74% of patients had at least one cIM and 9% and 34% of patients had at least two cIM, respectively (P<0.001 for both comparisons; Figure 2B). The most common cIM categories were lymphoproliferation (n=71), dermatological (n= 26), gastrointestinal/hepatic (n=23) and pneumological manifestations (n=16, Figure 3; Online Supplementary Figure S2B). The most frequent cIM diagnosis are shown in Table 2. Thirteen patients developed a neurological manifestation as previously described.10 Four patients had a hematological malignancy (age at diagnosis): Hodgkin lymphoma (16 years), juvenile myelomonocytic leukemia (20 years), large granular lymphocytic leukemia (21 years) and angioimmunoblastic Tcell lymphoma (29 years). Older age at ES diagnosis (HR 1.09; 95% CI: 1.01–1.17; P=0.02), cIM/cancer in a firstdegree relative (HR 1.64; 95% CI: 1.1–2.4; P=0.006), and the presence of AIN were independently associated with the number of cIM (HR 2.41; 95% CI: 1.5–3.8; P=0.0002). Biological IM (bIM) were diagnosed in 101 of 151 patients (67%), and the frequency of bIM also increased with the age (Figure 2C). Hypogammaglobulinemia was

the most frequently diagnosed bIM (n=54), including 44 cases diagnosed prior to any anti-CD20 treatment. Among those 54 patients, 25 (46%) received immunoglobulin replacement therapy. SLE and ALPS biomarkers were present (regardless of whether patients met the diagnostic criteria) in 42 and 24 patients, respectively. At 10 and 20 years of age, 39% and 75% of patients had at least one bIM, respectively (P<0.001). Patients with bIM were more likely to have cIM (79% vs. 40%; P<0.001), and patients with cIM were more likely to have bIM (80% vs. 41%; P<0.001) but the correlation between the number of bIM and cIM was low (r=0.27; P<0.001).

Secondary pediatric-onset Evans syndrome In 37 patients (24.5%), pES eventually revealed a SLE or a PID unknown at cytopenia onset. Eleven patients (7.3%) eventually met the SLE SLICC diagnostic criteria.16 These patients were older at first cytopenia (median age 13 years vs. 5 years; P=0.007) and almost exclusively female (one of 88 males [1%] and ten of 63 females [16%]); P<0.001).

Figure 1. Hematological outcomes. (A) Cumulative incidence of patients achieving a sustained complete remission (CR). Among the 23 patients without sustained CR for autoimmune hemolytic anemia (AIHA), four (17%) had achieved sustained CR for immune thrombocytopenic purpura (ITP). Conversely, among the 32 patients without sustained CR for ITP, 13 (40%) had achieved sustained CR for AIHA. (B) Percentage of patients with a sustained complete remission according to age.

460

haematologica | 2022; 107(2)

Pediatric-onset Evans syndrome outcomes

Seven patients (4.6%) met the diagnostic criteria for ALPS after pES onset which prompted targeted genetic analysis.17 Overall, 66 of 151 patients (44%) underwent genetic analyses as previously described.13 Among them, 26 (39%) patients were considered to have a PID (including the seven with ALPS). They carried a heterozygous

pathogenic variant in CTLA4 (n=7), TNFRSF6 (germline n=6, somatic n=1), STAT3 (n=5), PIK3CD (n=1), CBL (n=1), and KRAS (somatic n=1) or a homozygous/compound heterozygous pathogenic variants in LRBA (n=3) and RAG1 (n=1). Compared to the 40 other patients, the 26 with a PID had more cIM (2 [range, 1-5] vs. 1 [range,

C Figure 2. Immunopathological manifestations. (A) Age at first clinical immunopathological manifestation (cIM) diagnosis and at first cytopenia. Pearson correlation coefficient r=0.42, P<0.0001. There was no difference in the median age at first cIM and at first cytopenia in terms of the number of cIM (data not shown). (B) Cumulative incidence of cIM according to age. Half of the patients had developed a cIM by the age of 13.5 years and a second IM by the age of 27 years. (C) Cumulative incidence of any biological IM (bIM), as well as each category. Half of the patients had at least one bIM diagnosed by 13.2 years of age. The biological workup was not standardized and was made at the clinician’s discretion. SLE: systemic lupus erythematosus; ALPS: autoimmune lymphoproliferative syndrome.

haematologica | 2022; 107(2)

461

T. Pincez et al.

Figure 3. Immunopathological manifestations and other associated manifestations. Individual occurrence of autoimmune neutropenia, clinical immunopathological manifestations (cIM), biological IM (bIM), atopy, severe or recurrent infections, and malignancies. Each column represents a patient. The patients are ordered according to their cIM, from the most (lymphoproliferation) to the least (hematological, other) frequent. Hypoγ: hypogammaglobulinemia; SLE: systemic lupus erythematosus; ALPS: autoimmune lymphoproliferative syndrome.

Table 2. Most frequent clinical immunopathological manifestations diagnosis.

cIM Superficial (palpable) adenopathies Splenomegaly Deep (abdominal or thoracic) adenopathies Granulomatous–lymphocytic interstitial lung disease Cutaneous lupus erythematosus involvement Autoimmune hepatitis Subtentorial inflammatory lesions

n (%)

cIM

n (%)

61 (40) 49 (33) 16 (11) 16 (11) 8 (5) 7 (5) 7 (5)

Lymphoid enteropathy Chronic gastritis Polyarthritis Vitiligo Eczema Keratitis Uveitis

5 (3) 5 (3) 5 (3) 4 (3) 4 (3) 4 (3) 4 (3)

Diagnosis present in at least four patients are shown and ordered by frequency. Complete diagnosis list is provided in the Online Supplementary Table S2. cIM: clinical immunopathological manifestation.

0-4], P=0.008) and a trend toward more bIM as shown by Wilcoxon–Mann–Whitney sum ranks comparison but same medians (1 [range, 0-3] vs. 1 [range, 0-2], P=0.029). There was no statistically significant difference in the median time to ITP CR (4.7 years vs. 8.0 years, P=0.26) and to AIHA CR (5.5 years vs. 5.5 years, P>0.9), the number of second-line treatment received (3 [range, 0-9] vs. 2 [range, 0-6]; P=0.057) and mortality (two of 26 [7.7%] vs. three of 40 [7.5%]; P>0.9).

Treatments All except two patients (98.6%) had received at least one first-line treatment course. Second-line treatments (regardless of the hematological and/or extra-hematological indication) were required in 117 of 151 (77%) patients (Online Supplementary Figure 3A). Patients who did not receive any second-line treatment had shorter median follow-up times (10.5 years vs. 12.3 years; P=0.017). The median number of second-line treatments received was two (range, 0–9). The number of second-line treatments received increased with the time elapsed since first cytopenia without reaching a plateau (Online Supplementary Figure 3B). After a sustained CR for both ITP and AIHA achieved, the number of treatments received had continued to increase: at 5 years after CR of both cytopenias, 67% of patients who achieved CR for both ITP and AIHA had received a 462

new first and/or second-line treatments and 31% had received a new second-line treatment (Online Supplementary Figure S3C). The number of second-line treatments received increased with age, particularly after the first decade (Figure 4A). At 10 and 20 years, 47% and 88% of patients had received a second-line treatment, respectively (P<0.001). The number of patients receiving ongoing treatments also increased with age (Figure 4B). At 10 and 20 years, 27% and 69% of patients had received an active second-line treatment, respectively (P<0.001). At the final follow-up, patients with a cIM had received more secondline treatments (median, 3 vs. 1; P<0.0001). The most frequently used second-line treatments were rituximab (n=79; 52%), azathioprine (n=55; 36%), splenectomy (n=36; 24%), and mycophenolate (n=29; 19%; Online Supplementary Table S3). The number of cIM was associated with a subsequent increase in the number of second-line treatments received (HR 1.4; 95% CI: 1.15–1.60; P=0.0002). On the contrary, the number of second-line treatment was not associated to a subsequent increase in the number of cIM in univariate analysis (HR 1.09; 95% CI: 0.98–1.22; P=0.11).

Infections In total, 53 (35%) patients had severe or recurrent infections (Online Supplementary Table S4). The most frequent haematologica | 2022; 107(2)

Pediatric-onset Evans syndrome outcomes

Figure 4. Second-line treatments. (A) Total number of second-line treatments received according to the age. (B) Number of second-line treatments ongoing according to age.

were herpes zoster (n=17), sinusitis/otitis media (n=15), pneumopathy (n=12), and bronchiectasis (n=11). Patients with infections had more cIM (median, 2 vs. 1; P<0.0001), a higher incidence of hypogammaglobulinemia (53% vs. 28%; P=0.003), and received more second-line treatments (median 3 vs. 1; P<0.0001). Among the 16 patients with severe infection, nine (63%) were receiving an active treatment at infection time. Severe/recurrent infections were independently associated with hypogammaglobulinemia (OR 2.4; 95% CI: 1.10–5.33; P=0.03) and the number of second-line treatments (OR 1.34; 95% CI: 1.13–1.71; P=0.002).

Mortality Sixteen of the 151 patients followed for more than 5 years (10.6%) died, and seven other patients died before the fifth year of follow-up (23 deaths in total, 22 with available data). Patient survival at 5, 10, and 15 years after the first cytopenia was 97%, 92%, and 84%, respectively (Figure 5A). Mortality rates in patients with pES were haematologica | 2022; 107(2)

higher than those in patients with cITP or AIHA alone (P<0.0001 for both comparisons). Deaths occurred regularly throughout the follow-up period (median delay after first cytopenia diagnosis, 8.9 years [range, 0.1–24.3 years]) and at a median age of 18.0 years (range, 1.7–31.5 years) (Figure 5B). In the majority of these patients, cytopenia was under control at the time of death: 15 (65%) and 19 (83%) patients had CR or partial remission from ITP and AIHA, respectively (Figure 5C). Mortality was linked to the disease, the treatment, or both in eight (36%), two (9%), and twelve (55%) cases, respectively. The most frequent cause of death was infections (n=12 [52%]; Online Supplementary Table S5). Four patients (18%) died of a hemorrhage, and all were less than 13 years of age. The patients who died from a hemorrhage were younger than those who died from an infection (median 10 years vs. 18 years; P=0.03). All of these patients, except for one who died in the first month of a cerebral hemorrhage, had at least one cIM. Eight of the patients (36%) had hypogammaglobulinemia. 463

T. Pincez et al. A Figure 5. Long-term survival. (A) Survival estimate of patients in terms of time from first cytopenia. At 10-year follow-up, survival rates among patients with chronic immune thrombocytopenic purpura (ITP) alone, autoimmune hemolytic anemia (AIHA) alone and pediatric-onset Evans syndrome (pES) were 100%, 99% and 92%, respectively. (B) Mortality is shown in terms of time from first cytopenia, as well as age. Individual values are shown as dots with medians and interquartile ranges shown as lines. (C) Hematological status at death. CR: complete remission; PR: partial remission; NR: no remission.

The patients who died had received more second-line treatments than the others in the cohort (median 3 vs. 2; P=0.02), including splenectomy, which was more common in this subgroup (56% vs. 20%; P=0.003). Patients who had received more than two second-line treatments had a three-fold increase in the risk of death compared to those who had received two or less (11 of 65 [16.9%] vs. five of 86 [5.8%], P=0.03). At death, 81% of patients were receiving ongoing second-line treatment. The number of second-line treatments (HR 1.3; 95% CI: 1.1–1.6; P=0.004) and severe/recurrent infections (HR 3.4; 95% CI: 1.2–9.7; P=0.02) were independently associated with a higher risk of mortality after 5 years of follow-up.

Discussion This large follow-up study of pES patients included more than 1,900 patient-years. Over the long term, AIHA and ITP were sustainably controlled in the majority of 464

patients. Conversely, clinical and biological IM increased in frequency and number with increasing patient age, finally affecting almost all adult patients. The number of cIM was associated with a subsequent increase in the number of second-line treatments received. Mortality was high, frequently occurred while cytopenias were in remission, and most deaths concerned AYA. Two characteristics were associated to mortality: severe or recurrent infections and the number of second-line treatments received. Overall, the age-related clinical picture showed similar trends for all patients, shifting from cytopenia to increased IM, a greater treatment burden, and an increased risk of mortality. In setting up a nationwide cohort, the CEREVANCE group tried to ensure unbiased patient inclusion in this study. Omitting patients with less than 5 years of followup data limited any bias due to short-term follow-up, which probably accounts for many of the discrepancies between previous studies. Indeed, our median follow-up period was more than twice as long as in previous studies haematologica | 2022; 107(2)

Pediatric-onset Evans syndrome outcomes

(median 4.8 years [range, 3–7 years]).1,3–7 However, although the trends reported here are clear, some factors may also influence the estimates. The loss to follow-up mainly concerned AYA and few patients were followed after the age of 20 years. As well, the CEREVANCE group recommends clinical and biological follow-up at least every 6-12 months but local practice or patients’ phenotype (such as the presence of cIM) may have influenced biological testing. Sustained CR was eventually achieved for both types of cytopenia in the vast majority of patients, although this often took many years, especially for ITP (>10 years for one-third of our patients). Because active treatments are used to treat most AYA (notably because of cIM), hematological CR may be drug induced and it is impossible to determine whether an underlying hematological autoimmunity is still present. The higher rate of sustained CR in ITP among patients with pES compared to patients with cITP alone may be due to more patients with pES receiving treatment.18 One of the most striking findings in this study was the progressive increase in the frequency and number of IM. A range of cIM, affecting almost every organ, were identified and developed independently of cytopenias. These findings clearly show that pES is a marker for a more general tendency toward immunodeficiencies while we cannot exclude a contribution of the second-line treatments received to some IM. The underlying etiology is not completely understood and may vary among patients, with both genetic and environmental factors being important. Consequently, pES may be considered a composite syndrome with several overlapping subgroups of secondary pES. One of these subgroups includes patients with PID. Classically, ALPS has been associated with pES.12 In this study, only 4% of patients were diagnosed with ALPS based on well-defined criteria, despite evocative biological “ALPS-like” abnormalities in a larger proportion of patients.17 This observation is consistent with our previous study,13 which showed that more immune-response genes are potentially involved in pES than initially suspected.13,19–21 However, pES rarely comports as a Mendelian disease,1 and some of these variants may be predisposing rather than disease-causing alleles. Even in patients carrying a variant in a monogenic PID gene (e.g., TNFRFS6 or CTLA4),13,22 the altered genes show incomplete penetrance.23,24 We were unable to evaluate the proportion of patients who met common variable immunodeficiency disorders diagnostic criteria,25 as vaccine responses were not evaluable in all cases due to secondline treatments received. A second subgroup includes patients with SLE, although the prevalence of this subgroup is controversial.11,26,27 Our cohort suggests that SLE eventually occurs almost exclusively within the known atrisk population of female adolescents and is frequent in this subgroup, as it developed in seven of 15 (47%) of the females >12 years old.26 Despite its heterogeneity, the course of pES, in terms of age-related changes and trends, was similar for the majority of patients. The spectrum of IM described here is probably influenced by the underlying etiology, and further analyses are needed to understand the determinant of IM. The long-term follow-up of the present study confirms that the subgroup of patients with identified PID had more cIM.13 Most patients required second-line treatments. These treatments reflect local practices and we cannot draw conhaematologica | 2022; 107(2)

clusions regarding their efficacy. We were unable to investigate the risk associated to specific treatments given the high heterogeneity in second-line treatment combinations and duration as well as the changes in management practices since the cohort onset in 2004. The rapid initial increase in second-line treatments is partly due to the high rate of early relapse and the current practice of administering steroid-sparing agents to treat pES.28 However, the presence of cytopenia is not the only reason for using these drugs and first- and second-line treatments were also used after CR of both cytopenias. cIM are important in determining the number of second-line treatments used, but bIM may also play a role, particularly in patients with SLE biomarkers, who are frequently given hydroxychloroquine. Nevertheless, second-line treatments are rarely selected based on a single factor. Patients with pES often have bIM and cIM, and the whole clinical picture needs to be assessed before selecting a treatment strategy. As previously reported,13 approximately one-third of patients may carry alterations in genes that are potentially accessible to targeted therapy.29–31 Given the high burden of second-line treatments and their association with infections and mortality, the CEREVANCE network has proposed implementing genetic analyses for all patients with pES to limit the use of immunosuppressive and toxic drugs. Comprehensively, the pES clinical picture changes as patients age. From 10 to 20 years of age, cytopenia tends to be controlled but IM are more prevalent, and active second-line treatments are used in more than two-thirds of patients during the pediatric-to-adult transition. Overall, as patients age, the illness becomes more severe and the risk of mortality increases. Both IM and treatment burden contribute to the infection-related mortality peak observed at the end of the second decade. The patients who died had received more second-line treatments, including splenectomy. Because these two parameters are correlated (r=0.60; P<0.0001), the number of deaths was too low to determine whether splenectomy alone was a risk factor of mortality per se or a marker of severity. In conclusion, pES must now be considered a complex multi-systemic disease in which cytopenias frequently present fewer challenges than IM and infections in longterm follow-up. Adult patients with pES form a specific subgroup, distinct from older adults with ES.32 Multidisciplinary follow-up of patients with pES is needed and must focus on IM screening, genetic diagnosis, infections prevention, patient-tailored drugs development, and AYA management. Specifically, the infection burden may be reduced by ensuring up-to-date vaccinations, eradicating chronic infections, and using adequate antimicrobial prophylaxis or immunoglobulin replacements. As in several chronic pediatric diseases,33 dedicated child-to-adult transition programs are warranted to improve outcomes in patients with pES. Disclosures No conflicts of interest to disclose. Contributions TP, HF, TL, GL and NA designed the study, analyzed the data and drafted the paper; TP and HF performed statistical analyses; CP and FR-L performed genetic analyses. All of the authors participated to prospective data collection and interpretation and revised the manuscript for critical content. 465

T. Pincez et al.

Acknowledgments The list of collaborators is given in the Online Supplementary Appendix. The authors would like to thank all of the patients, families, medical and para-medical teams involved in the CEREVANCE prospective cohort study from 2004 onwards. Funding This work was supported from 2004 by the French Ministry of Health (Programme Hospitalier de Recherche Clinique [PHRC] 2005, Rare Disease Plan 2007 and 2017), the Association

References 1. Aladjidi N, Fernandes H, Leblanc T, et al. Evans syndrome in children: long-term outcome in a prospective French national observational cohort. Front Pediatr. 2015;3:79. 2. Evans RS, Takahashi K, Duane RT, Payne R, Liu C. Primary thrombocytopenic purpura and acquired hemolytic anemia; evidence for a common etiology. AMA Arch Intern Med. 1951;87(1):48-65. 3. Mathew P, Chen G, Wang W. Evans syndrome: results of a national survey. J Pediatr. Hematol Oncol. 1997;19(5):433437. 4. Pui CH, Wilimas J, Wang W. Evans syndrome in childhood. J. Pediatr. 1980;97(5):754–758. 5. Savaşan S, Warrier I, Ravindranath Y. The spectrum of Evans’ syndrome. Arch Dis Child. 1997;77(3):245-248. 6. Wang WC. Evans syndrome in childhood: pathophysiology, clinical course, and treatment. Am J Pediatr Hematol Oncol. 1988;10(4):330-338. 7. Blouin P, Auvrignon A, Pagnier A, et al. Syndrome d’Evans : étude rétrospective de la société d’hématologie et d’immunologie pédiatrique (36 cas). Arch Pédiatrie. 2005;12(11):1600-1607. 8. Aladjidi N, Leverger G, Leblanc T, et al. New insights into childhood autoimmune hemolytic anemia: a French national observational study of 265 children. Haematologica. 2011;96(5):655-663. 9. Mannering N, Hansen DL, Frederiksen H. Evans syndrome in children below 13 years of age – a nationwide population-based cohort study. PoS One. 2020; 15(4):e0231284. 10. Pincez T, Neven B, Le Pointe HD, et al. Neurological involvement in childhood Evans syndrome. J Clin Immunol. 2019; 39(2):171-181. 11. Costallat GL, Appenzeller S, Costallat LTL. Evans syndrome and systemic lupus erythematosus: clinical presentation and outcome. Joint Bone Spine. 2012;79(4):362364. 12. Teachey DT, Manno CS, Axsom KM, et al. Unmasking Evans syndrome: T-cell phenotype and apoptotic response reveal autoimmune lymphoproliferative syndrome

466

Bordelaise pour l’Avancement des Sciences Pédiatriques (ABASP) research charity, the Association pour la Recherche et les Maladies Hématologiques de l’Enfant (RMHE) research charity, the Association Française du Syndrome d’Evans (AFSE), the O-CYTO patients’ association, and partially by GlaxoSmithKline, AMGEN and Novartis. TP is a recipient of a Charles Bruneau fellowship award. Data sharing statement Data are available on request to corresponding author.

(ALPS). Blood. 2005;105(6):2443-2448. 13. Hadjadj J, Aladjidi N, Fernandes H, et al. Pediatric Evans syndrome is associated with a high frequency of potentially damaging variants in immune genes. Blood. 2019;134(1):9-21. 14. White PH, Cooley WC, Group TCRA, et al. Supporting the health care transition from adolescence to adulthood in the medical home. Pediatrics. 2018;142(5):e20182587. 15. 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. 16. Petri M, Orbai A-M, Alarcón GS, et al. Derivation and validation of Systemic Lupus International Collaborating Clinics Classification Criteria for systemic lupus erythematosus. Arthritis Rheum. 2012;64(8):2677-2686. 17. Oliveira JB, Bleesing JJ, Dianzani U, et al. Revised diagnostic criteria and classification for the autoimmune lymphoproliferative syndrome (ALPS): report from the 2009 NIH International Workshop. Blood. 2010;116(14):e35-40. 18. Ducassou S, Gourdonneau A, Fernandes H, et al. Second-line treatment trends and long-term outcomes of 392 children with chronic immune thrombocytopenic purpura: the French experience over the past 25 years. Br J Haematol. 2020;189(5):931-942. 19. Rieux-Laucat F, Le Deist F, Fischer A. Autoimmune lymphoproliferative syndromes: genetic defects of apoptosis pathways. Cell Death Differ. 2003;10(1):124133. 20. Schubert D, Bode C, Kenefeck R, et al. Autosomal dominant immune dysregulation syndrome in humans with CTLA4 mutations. Nat Med. 2014;20(12):14101416. 21. Flanagan SE, Haapaniemi E, Russell MA, et al. Activating germline mutations in STAT3 cause early-onset multi-organ autoimmune disease. Nat Genet. 2014;46(8):812-814. 22. Besnard C, Levy E, Aladjidi N, et al. Pediatric-onset Evans syndrome: heterogeneous presentation and high frequency of monogenic disorders including LRBA and CTLA4 mutations. Clin Immunol. 2018;188:52-57.

23. Schwab C, Gabrysch A, Olbrich P, et al. Phenotype, penetrance, and treatment of 133 CTLA-4-insufficient individuals. J. Allergy Clin Immunol. 2018;142(6):19321946. 24. Neven B, Magerus-Chatinet A, Florkin B, et al. A survey of 90 patients with autoimmune lymphoproliferative syndrome related to TNFRSF6 mutation. Blood. 2011; 118(18):4798-4807. 25. Bonilla FA, Barlan I, Chapel H, et al. International Consensus Document (ICON): common variable immunodeficiency disorders. J. Allergy Clin Immunol Pract. 2016;4(1):38-59. 26. Tarvin SE, O’Neil KM. Systemic lupus erythematosus, Sjögren syndrome, and mixed connective tissue disease in children and adolescents. Pediatr Clin North Am. 2018; 65(4):711-737. 27. Lube GE, Ferriani MPL, Campos LMA, et al. Evans syndrome at childhood-onset systemic lupus erythematosus diagnosis: a large multicenter study. Pediatr Blood Cancer. 2016;63(7):1238-1243. 28. Miano M. How I manage Evans Syndrome and AIHA cases in children. Br J Haematol. 2016;172(4):524-534. 29. Lee S, Moon JS, Lee C-R, et al. Abatacept alleviates severe autoimmune symptoms in a patient carrying a de novo variant in CTLA-4. J. Allergy Clin Immunol. 2016;137(1):327-330. 30. Lo B, Zhang K, Lu W, et al. Patients with LRBA deficiency show CTLA4 loss and immune dysregulation responsive to abatacept therapy. Science. 2015;349(6246):436440. 31. Klemann C, Esquivel M, Magerus-Chatinet A, et al. Evolution of disease activity and biomarkers on and off rapamycin in 28 patients with autoimmune lymphoproliferative syndrome. Haematologica. 2017;102(2):e52-e56. 32. Michel M, Chanet V, Dechartres A, et al. The spectrum of Evans syndrome in adults: new insight into the disease based on the analysis of 68 cases. Blood. 2009; 114(15):3167-3172. 33. Gabriel P, McManus M, Rogers K, White P. Outcome evidence for structured pediatric to adult health care transition interventions: a systematic review. J Pediatr. 2017; 188:263-269.

haematologica | 2022; 107(2)

ARTICLE

Iron Metabolism & its Disorders

Risk factors for endocrine complications in transfusion-dependent thalassemia patients on chelation therapy with deferasirox: a risk assessment study from a multi-center nation-wide cohort Maddalena Casale,1 Gian Luca Forni,2 Elena Cassinerio,3 Daniela Pasquali,4 Raffaella Origa,5 Marilena Serra,6 Saveria Campisi,7 Angelo Peluso,8 Roberta Renni,9 Alessandro Cattoni,10 Elisa De Michele,11 Massimo Allò,12 Maurizio Poggi,13 Francesca Ferrara,14 Rosanna Di Concilio,15 Filomena Sportelli,16 Antonella Quarta,17 Maria Caterina Putti,18 Lucia Dora Notarangelo,19 Antonella Sau,20 Saverio Ladogana,21 Immacolata Tartaglione,1 Stefania Picariello,1 Alessia Marcon,3 Patrizia Sturiale,22 Domenico Roberti,1 Antonio Ivan Lazzarino23 and Silverio Perrotta1 Department of Woman, Child and General and Specialized Surgery, University of Campania Luigi Vanvitelli, Naples, Italy; 2Center of Microcitemia and Congenital Anemias, Galliera Hospital, Genoa, Italy; 3Rare Diseases Center, General Medicine Unit, IRCCS Ca’ Granda Ospedale Maggiore Policlinico Section co, Milan, Italy; 4Endocrinology Unit, Department of Advanced Medical and Surgical Sciences, University “ Luigi Vanvitelli”, Naples, Italy; 5Thalassemia Center, Pediatric Hospital A CAO, AOG Brotzu, Cagliari, Italy; 6Thalassemia Center, Department of Internal Medicine, Hospital "V. Fazzi", Lecce, Italy; 7Thalassemia Center, Hospital Umberto I, Siracusa, Italy; 8Center of Microcitemia, POC SS.Annunziata - ASL TA, Taranto, Italy; 9Thalassemia Center, Department of Internal Medicine, Hospital F.Ferrari, Casarano, Italy; 10Department of Pediatrics, Università degli Studi di Milano Bicocca, Fondazione Monza e Brianza per il Bambino e la sua Mamma, Azienda Ospedaliera San Gerardo, Monza, Italy; 11 Immunotransfusion Medicine Unit, AOU OO.RR. S. Giovanni di Dio e Ruggi d'Aragona, Salerno, Italy; 12Center of Microcitemia, Hospital ASL 5, Crotone, Italy; 13Department of Endocrinology, Sant'Andrea Hospital, Rome, Italy; 14Department of Internal Medicine, Policlinico Hospital of Modena, Modena, Italy; 15Department of Pediatrics, Hospital Umberto I, Nocera, Italy; 16Immunotransfusion Unit, Hospital Riuniti, Foggia, Italy; 17 Center for Microcythemia, Iron Metabolism Disorders, Gaucher Disease - Hematology and Transplantation Unit, "A. Perrino" Hospital, Brindisi, Italy; 18Department of Women's and Child's Health (DSDB), University Hospital, Padova, Italy; 19Hematology Oncology Unit, Children's Hospital, ASST Spedali Civili, Brescia, Italy; 20Department of Pediatric Hematology and Oncology, Hospital “Spirito Santo”, Pescara, Italy; 21Pediatric Oncohematology Unit, “Casa Sollievo della Sofferenza” Hospital, IRCCS, San Giovanni Rotondo, Italy; 22SSD Microcitemia Center, G.O.M Reggio Calabria, Reggio Calabria, Italy and 23EPISTATA – Agency for Clinical Research and Medical Statistics, London, UK

Ferrata Storti Foundation

Haematologica 2022 Volume 107(2):467-477

ABSTRACT

ransfusion-dependent patients typically develop iron-induced cardiomyopathy, liver disease, and endocrine complications. We aimed to estimate the incidence of endocrine disorders in transfusiondependent thalassemia (TDT) patients during long-term iron-chelation therapy with deferasirox (DFX). We developed a multi-center follow-up study of 426 TDT patients treated with once-daily DFX for a median duration of 8 years, up to 18.5 years. At baseline, 118, 121, and 187 patients had 0, 1, or ≥2 endocrine diseases respectively. 104 additional endocrine diseases were developed during the follow-up. The overall risk of developing a new endocrine complication within 5 years was 9.7% (95% Confidence Interval [CI]: 6.3–13.1). Multiple Cox regression analysis identified three key predictors: age showed a positive log-linear effect (adjusted hazard ratio [HR] for 50% increase 1.2, 95% CI: 1.1–1.3, P=0.005), the serum concentration of thyrotropin showed a positive linear effect (adjusted HR for 1 mIU/L increase 1.3, 95% CI: 1.1–1.4, P<0.001) regardless the kind of disease incident, while the number of previous endocrine diseases showed a negative linear effect: the higher the number of diseases at baseline the lower the chance of developing further diseasess (adjusted HR for unit haematologica | 2022; 107(2)

Correspondence: MADDALENA CASALE maddalena.casale@unicampania.it Received: September 18, 2020. Accepted: December 22, 2020. Pre-published: January 7, 2021. https://doi.org/10.3324/haematol.2020.272419

467

M. Casale et al.

increase 0.5, 95% CI: 0.4–0.7, P<0.001). Age and thyrotropin had similar effect sizes across the categories of baseline diseases. The administration of levothyroxine as a covariate did not change the estimates. Although in DFX-treated TDT patients the risk of developing an endocrine complication is generally lower than the previously reported risk, there is considerable risk variation and the burden of these complications remains high. We developed a simple risk score chart enabling clinicians to estimate their patients’ risk. Future research will look at increasing the amount of variation explained from our model and testing further clinical and laboratory predictors, including the assessment of direct endocrine magnetic resonance imaging.

Introduction Transfusion-induced iron overload in thalassemia patients typically results in iron-induced cardiomyopathy, liver disease, and endocrine complications. However, those three phenomena have been studied to different extents. In transfusion-dependent thalassemia (TDT) patients, mortality due to cardiovascular and hepatic complications has markedly declined during the last decades.1–3 The development of magnetic resonance imaging techniques (MRI), specifically designed to quantify myocardial and hepatic iron concentration, measuring heart T2* and liver iron concentration (LIC), has enabled the design of clinical trials evaluating the efficacy of iron chelators in targeting specific iron overload.4 Moreover, new anti-hepatitis C drugs have remarkably reduced the complications linked to hepatitis C infection, which used to dramatically deteriorate liver iron overload.5 However, in spite of the outstanding advances in the care of cardiovascular and hepatic complications due to blood transfusions, the management of endocrine complications has been left behind and, nowadays, they are the most frequent and the most resource-draining complications in TDT patients.3 In addition, serological testing fails to identify high-risk groups and, once occurred, these complications are often irreversible. While MRI imaging of endocrine glands is promising in detecting preclinical disease, it has not reached the level of validation required for routine clinical use.6 The once-daily oral iron chelator deferasirox (DFX) was shown to be effective in chelating iron from the heart and the liver, with preservation of the heart function,7–9 and with reversal of the hepatic fibrosis.10 While the effective control of heart and liver siderosis remains the primary goal in the management of TDT patients, observational data suggest that iron loading in endocrine organs may precede myocardial involvement and there is now substantial evidence on the role of iron overload in endocrine morbidity.11–14 While there have been small studies on endocrine disorders in TDT patients during chelation therapy with DFX,15,16 the data are still scarce, even though DFX is nowadays the most prescribed drug for iron chelation in TDT patients.17 The aim of this study was to assess the incidence of endocrine diseases including hypothyroidism, hypoparathyroidism, glucose metabolisms disorders, hypogonadism, and metabolic bone disease in patients suffering from TDT who are on treatment with the drug DFX.

Methods In this multi-center study, TDT patients from 21 hospitals located in 21 cities and 19 regions of Italy were assessed for eligibility to be recruited in the cohort.

468

We considered the following endocrine conditions: 1. Hypothyroidism (overt: thyrotropin [TSH] >10 mU/mL and low free thyroxine [FT4]; subclinical: TSH 5–10 mU/mL and normal FT4). 2. Hypoparathyroidism (low parathyroid hormone [PTH] and calcium and high phosphorus). 3. Hypogonadism (hypogonadotropic hypogonadism, in adult female: amenorrhea, low estradiol levels and low or normal luteinizing hormone and follicle stimulating hormone [LH/FSH] levels; in adult male: low testosterone levels, clinical signs or symptoms consistent with hypogonadism and low/normal LH/FSH. Testosterone reference ranges vary according to patients’ age at the time of biochemical assessment. In general, they were regarded as normal if >3.5 ng/mL and unequivocally pathological <2.3 ng/mL. Additional data [clinical features, free testosterone] were taken into account for values between 2.3 and 3.5 ng/mL. Hypergonadotropic hypogonadism, in adult female: amenorrhea and raised FSH [>30 U/L] with undetectable estradiol; in adult male: raised gonadotropins with low total testosterone and clinical signs consistent with hypogonadism). 4. Pubertal disturbances (delayed puberty: lack of breast budding [Tanner stage 2] in girls by the age of 13 and testicular volume <4 mL in boys by the of 14; arrested puberty: lack of pubertal progression over a year or more). 5. Disorders of glucose metabolism (diabetes: fasting plasma glucose ≥126 mg/dL or 2-hour plasma glucose [2-h PG] value during a 75-g oral glucose tolerance test [OGTT] >200 mg/dL; impaired fasting glucose [IFG]: fasting glucose between 100 and 125 mg/dL; impaired glucose tolerance: 2-h PG during 75-g OGTT levels between 140 and 199 mg/dL). 6. Bone metabolism disorder (BMD) (osteoporosis: bone mineral density T score ≤-2.5 and Z score value ≤-2; osteopenia: T score value 1.01/-2.5 and Z score value 1.01/-2. In childhood, osteoporosis was defined by either the association of at least two pathological fractures by the age of 10 years/ three by the age of 19 and Z score ≤-2 or by the finding of at least one vertebral crush, in the absence of high-energy trauma or local disease, irrespectively of the BMD recorded; low bone mineral density was defined as the finding of BMD Z score ≤-2, in the absence of the above-mentioned additional criteria for osteoporosis). According to standardized protocols,18 laboratory tests for detection of endocrine disorders were performed every year in patients with no endocrine complications and more frequently (every 3-6 months) in patients with endocrine disorders, as per consolidated clinical practice and according to the endocrinologists’ prescription. Routine laboratory tests, such as glycemia and serum electrolytes, were assessed every 1-3 months in occasion of pre-transfusion cross-match testing. Weight, height and Tanner stage were assessed every 6 months in patients <18 years of age. The study protocol was approved by Ethical Committees and Institutional Review Boards of all the participating centers and was conducted in accordance with the Declaration of Helsinki and ICH guidelines for good clinical practice. All patients provided written informed consent.

haematologica | 2022; 107(2)

Endocrine complications in iron overload

Table 1. Cross-tabulation of number of conditions at baseline and number of conditions occurred during the follow-up.

N. of endocrine diseases at baseline 0 0 1 2 3 4 Total

75(63.6%) (23.3%) 87(71.9%) (27.0%) 86(80.4%) (26.7%) 59(90.8%) (18.3%) 15(100.0%) (4.7%) 322(75.6%) (100%)

N. of new endocrine diseases occurred during follow-up 1 2 32(27.1%) (36.4%) 30(24.8%) (34.1%) 21(19.6%) (23.9%) 5(7.7%) (5.7%) 0 (0.0%) 88(20.7%) (100%)

10(8.5%) (66.7%) 4(3.3%) (26.7%) 0(0.0%) (0.0%) 1(1.5%) (6.7%) n/a 15(3.5%) (100%)

Total 3 1(0.9%) (100.0%) 0(0.0%) (0.0%) 0(0.0%) (0.0%) n/a n/a (3.5%) 1(0.2%) (100%)

118 (100%) (27.7%) 121 (100%) (28.4%) 107 (100%) (25.1%) 65 (100%) (15.3%) 15 (100%) 426 (100%) (100%)

Values represent the total number of patients in each category, with row and column percentages in parentheses.

A detailed description of all methods used is available in the Online Supplementary Appendix.

Statistical analysis Data were cleaned before the analysis: we checked all variables for missing, illogical or implausible values, also through crosschecks with related variables (e.g., chronologic orders). Continuous variables were checked for abnormal distributions and outliers. We used Cox-regression to fit survival analyses with follow-up days as the underlying time variable. Survival time was measured as the number of days passed from the beginning of the treatment regimen with the drug DFX to the first of either the diagnosis of the first new endocrine disease, side effect due to DFX leading to therapy suspension, death, or censoring. We centered the covariates before interaction analyses. We adopted two strategies for the development of the multiple models: we either started with the covariates having higher biological plausibility of effect or with those with a lower P-value from at the bivariate stage. The two approaches reached the same final model. The assumption of proportional hazards was checked by using the Schoenfeld residuals test. We computed the proportion of variation explained by the models (adjusted R2) using the Royston method with bootstrap confidence intervals (5,000 replications).19 We derived the predicted probabilities of developing a new endocrine disease within 5 years and 1 year by using the margins command in Stata v.14. Two patients were excluded from the analysis as they already had all possible five endocrine diseases at baseline.

Results Out of 426 patients enrolled, accounting for 3,517 person per years, 104 participants developed at least one new endocrine disease after a mean and median followup time of 8 years (range, 1 month–18.5 years). The mean iron intake at baseline was 0.28+/-0.08 mg/kg/day (range, 0.14–0.49 mg/kg/day) and at the end of study was 0.26+/0.12 mg/kg/day (range, 0.16–0.50 mg/kg/day). The mean hemoglobin level was 9.8+/-0.68 g/dL (range, 9.4–10.6 g/dL) indicating the majority of patients had good control of their chronic anemia. No deaths were recorded. Overall, 18 (4%) patients experienced adverse events (AE) that determined tempohaematologica | 2022; 107(2)

rary or permanent DFX discontinuation. The most frequent AE were related to gastrointestinal intolerance (epigastralgia, heartburn, abdominal pain; n=8) and increased transaminases (n=8). Increased in serum creatinine (n=1) and Lichen planus (n=1) were reported as other AE which caused DFX interruption. In nine (2%) cases DFX was discontinued because of treatment failure, reported as increase in serum ferritin (n=6), cardiac T2* (n=2), LIC (n=1). In one case treatment failure was reported along with gastrointestinal intolerance. Table 1 shows a cross-tabulation between the number of endocrine diseases at baseline and the number of new endocrine diseases that occurred during the follow-up. The 75.6% of the total sample (322 of 426) did not develop any new endocrine disease during the follow-up (95% Confidence Interval [CI]: 71.2–79.6). Out of the 104 (24.4%) with newly diagnosed endocrinopathies, 84.6% developed only one endocrine disease (95% CI: 76.2– 90.9). Out of 118 patients with no endocrine diseases at baseline, 43 (36.4%) developed at least one endocrine disease during the follow-up (95% CI: 27.8–45.8). Out of 121 patients having one endocrine disease at baseline already, 34 (28.1%) developed at least one additional endocrine disease during the follow-up (95% CI: 20.3– 37.0). Among the 118 patients with no endocrine diseases at baseline, BMT disorders occurred the most (17.8% [95% CI:11.4–25.9]), followed by hypogonadism (12.7% [95% CI=7.3–20.1]). Those two conditions were also the most prevalent ones in patients with one disease at a baseline (80.2% [95% CI: 71.9–86.9] and 11.6% [95% CI: 6.5– 18.7] respectively) and were those that most likely occurred as additional diseases during the follow-up. Figure 1 shows the overall crude risks for all 104 first incidents, by incident type and age group. It appears that most of the new incidents occurred after the age of 20 years, with a new spike between 35 and 45 years. As for pediatric patients, the increase seems to start after the age of 12 years. No cases of insulin-dependent diabetes were reported in patients with no endocrine disorders at baseline (Figure 2). Kaplan-Meier survival probability curves with age as the underlying time variable are reported in the Online Supplementary Appendix. Tables 2A and 2B show a description of the sample by 469

M. Casale et al.

prevalent endocrine diseases at baseline and by incident endocrine diseases during the follow-up. Age, TSH, and low BMD were associated with both prevalent and incident disorders. Among the markers of iron overload, ferritin and T2* were associated with prevalent but not with incident disorders, whereas LIC was not associated with any of them. During the follow-up, iron overload test results either decreased by a small extent or remained stable over time, while the standard deviation of those differences was more than three times their means. The number of prevalent endocrine disorders was inversely associated with the incidence of a new disorder. Only 11 patients (2.6%) had a side-effect related to DFX administration (gastrointestinal disorder).

Tables 3A and 3B show the results from the multiple Cox regression models. In both models, the adjusted hazard rate of developing a new endocrine disorder decreased by about 50% for each prevalent endocrine disease at baseline (P<0.001) and increased by about 25% for each mIU/L of TSH at baseline (P<0.001). The two models differ in the way in which the variable age was treated. In model 3A age was treated as a log-linear variable whereas in model 3B age was treated as a linear variable, but in that case, also a binary indicator was included for pediatric/adult patient and an interaction term between age and the indicator as well. The latter model showed a higher adjusted R2 (0.25 vs. 0.22) although that difference was not significant (95% CI: 0.19–0.42 vs. 95%

Figure 1. Overall crude risks for all 104 first incidents (n=426), by incident type and age group.

Figure 2. Crude risks for all 43 first incidents in patients with no endocrinopathies at baseline (n=118), by incident type and age group.

470

haematologica | 2022; 107(2)

Endocrine complications in iron overload

Table 2A. Sample description stratified by number of endocrine diseases at baseline.

Factors and categories* 0 N=118 Age (years) Pediatric patient (<16 years) Age if child (n=109) Age if adult (n=317) Sex male Splenectomised Used drugs other than DFX in the past Heart disease Thyroid disorder Parathyroid disorder Gonadal disorder Glucose metabolism disorder BMD Ferritin (ng/mL) Ferritin >2,000 (ng/mL) LIC (mg Fe/g dry weight) LIC <3 3715+ Heart T2* (msec) EF (%) TSH (mIU/L) On levotyroxine FT4 (pmol/L) TSH index PTH (pg/mL) Glycemia (mg/dL) Calcium (mg/dL) Phosphorus (mg/dL) BMD femur (g/cm2) BMD femur (z score) BMD femur (t score) BMD L1-L4 (g/cm2) BMD L1-L4 (z score) BMD L1-L4 (t score) New endocrine disease occurred Thyroid disorder occurred Parathyroid disorder occurred Gonadal disorder occurred Glucose metabolism disorder occurred BMD Side effect occurred D Ferritin† D LIC† D T2*†

Number of endocrine diseases at baseline 1 2+ N=121 N=187

9.1 (5.4-23.2) 68.6% (81/118) 6.9 (3.7) 31.3 (11.9) 50.8% (60/118) 23.7% (28/118) 27.1% (32/118) 4.4% (5/114) 0.0% (0/118) 0.0% (0/118) 0.0% (0/118) 0.0% (0/118) 0.0% (0/118) 1342.1 (×/2.0) 28.4% (29/102) 4.9 (×/2.4)

28.9 (18.6-36.7) 17.4% (21/121) 10.9 (3.5) 32.0 (10.5) 49.6% (60/121) 47.9% (58/121) 18.2% (22/121) 7.8% (9/116) 6.6% (8/121) 1.7% (2/121) 11.6% (14/121) 0.0% (0/121) 80.2% (97/121) 937.2 (×/2.3) 18.3% (21/115) 4.0 (×/2.4)

34.7 (29.6-39.7) 3.7% (7/187) 10.9 (3.9) 35.6 (7.3) 38.5% (72/187) 75.4% (141/187) 36.4% (68/187) 19.5% (34/174) 35.8% (67/187) 7.0% (13/187) 88.2% (165/187) 26.7% (50/187) 93.0% (174/187) 844.9 (×/2.2) 15.0% (26/173) 3.9 (×/2.2)

28.6% (20/70) 38.6% (27/70) 22.9% (16/70) 10.0% (7/70) 36.0 (11.3) 64.0 (6.6) 2.5 (×/1.5) 0.0% (0/104) 14.8 (3.3) 2.8 (0.7) 25.0 (×/1.6) 84.5 (×/1.1) 9.4 (9.1-9.7) 4.3 (3.5-5.0) 0.9 (0.8-1.0) -0.4 (1.5) -0.5 (1.5) 1.0 (0.9-1.1) -0.8 (-1.4--0.3) -0.8 (-1.4--0.3) 36.4% (43/118) 5.1% (6/118) 0.8% (1/118) 12.7% (15/118) 3.4% (4/118) 24.6% (29/118) 0.8% (1/118) -476.6 (1519.6) -1.4 (5.2) 0.7 (13.8)

43.2% (41/95) 27.4% (26/95) 22.1% (21/95) 7.4% (7/95) 35.7 (9.4) 62.9 (5.8) 2.1 (×/1.6) 5.2% (6/115) 14.2 (2.9) 2.6 (0.7) 27.3 (×/1.8) 84.2 (×/1.1) 9.3 (8.9-9.6) 4.1 (3.5-4.7) 0.7 (0.6-0.9) -1.5 (1.1) -1.6 (0.9) 0.8 (0.7-0.9) -2.1 (-2.9--1.3) -2.2 (-3.0--1.3) 28.1% (34/121) 8.8% (10/113) 1.7% (2/119) 12.1% (13/107) 2.5% (3/121) 41.7% (10/24) 4.1% (5/121) -310.8 (1419.9) -1.6 (6.6) 1.9 (11.5)

37.4% (55/147) 38.1% (56/147) 19.7% (29/147) 4.8% (7/147) 30.5 (12.0) 64.4 (6.8) 2.1 (×/2.1) 27.8% (49/176) 14.5 (4.6) 2.7 (0.9) 21.9 (×/2.3) 95.1 (×/1.3) 9.3 (9.0-9.7) 4.0 (3.4-4.5) 0.7 (0.6-0.8) -2.0 (1.0) -2.1 (1.0) 0.8 (0.7-0.9) -2.5 (-3.2--1.9) -2.8 (-3.4--2.0) 14.4% (27/187) 5.0% (6/120) 4.0% (7/174) 18.2% (4/22) 5.8% (8/137) 23.1% (3/13) 2.7% (5/187) -430.6 (1103.1) -1.8 (6.6) 5.9 (12.8)

P-value

<0.001 <0.001 <0.001 0.001 0.054 <0.001 0.002 <0.001

<0.001 0.023 0.18 0.34

<0.001 0.33 0.061 <0.001 0.54 0.39 0.19 <0.001 0.25 0.17 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.39 0.18 0.74 0.36 0.22 0.28 0.62 0.93 0.008 continued on following page

haematologica | 2022; 107(2)

471

M. Casale et al. continued from previous page

D TSH† D TSH index† D BMD femur (g/cm2)† D BMD L1-L4 (g/cm2)†

-0.1 (1.4) -0.0 (0.8) -0.0 (0.1) -0.2 (0.7)

0.1 (1.5) 0.0 (0.8) 0.0 (0.3) 0.0 (0.1)

0.5 (5.1) 0.2 (0.9) 0.0 (0.2) 0.0 (0.2)

0.41 0.092 0.74 0.007

For normally-distributed variables, data are presented as mean (standard deviation [SD]) with P-value from ANOVA. For log-normal variables, data are presented as geometric mean (×/geometric SD) with P-value from ANOVA on logged values. For continuous variables with other types of distributions, data are presented as median (interquartile range [IQR]) with P-value from Kruskal-Wallis test. For categorical variables, data are presented as % (n/total) with P-value from Pearson's chi-squared test. *Measured at baseline, unless otherwise specified. †Intra-individual difference between measures taken at the end and at the beginning of the follow-up. DFX: deferasirox ; BMD: bone metabolism disorder; LIC: liver iron concentration; EF: ejection fractions; FT4: free thyroxine; TSH: thyrotropin; PTH: parathyroid hormone.

Table 2B. Sample description stratified by categories of outcome measure.

Factors and categories* Age (years) Pediatric patient (<16 years) Age if child (n=109) Age if adult (n=317) Sex male Splenectomised Used drugs other than DFX in the past Heart disease Thyroid disorder Parathyroid disorder Gonadal disorder Glucose metabolism disorder Bone metabolism disorder N. of endocrine diseases at baseline 0 1 2+ Ferritin (ng/mL) Ferritin >2,000 (ng/mL) LIC (mg Fe/g dry weight) LIC <3 3715+ Heart T2* (msec) EF (%) TSH (mIU/L) On levotyroxine FT4 (pmol/L) TSH index PTH (pg/mL) Glycemia (mg/dL) Calcium (mg/dL) Phosphorus (mg/dL) BMD femur (g/cm2) BMD femur (z score) BMD femur (t score) BMD L1-L4 (g/cm2) BMD L1-L4 (z score) BMD L1-L4 (t score)

New incident

P-value

No N=322

Yes N=104

30.5 (18.2-38.3) 23.3% (75/322) 7.5 (4.0) 34.2 (8.6) 43.2% (139/322) 54.7% (176/322) 28.9% (93/322) 10.7% (33/309) 19.9% (64/322) 3.1% (10/322) 46.3% (149/322) 14.0% (45/322) 70.8% (228/322)

27.0 (11.8-35.4) 32.7% (34/104) 8.9 (3.7) 33.2 (11.1) 51.0% (53/104) 49.0% (51/104) 27.9% (29/104) 15.8% (15/95) 10.6% (11/104) 4.8% (5/104) 28.8% (30/104) 4.8% (5/104) 41.3% (43/104)

23.3% (75/322) 27.0% (87/322) 49.7% (160/322) 979.3 (×/2.2) 18.5% (55/297) 4.1 (×/2.4)

41.3% (43/104) 32.7% (34/104) 26.0% (27/104) 995.8 (×/2.4) 22.6% (21/93) 4.2 (×/2.1)

37.3% (90/241) 33.2% (80/241) 22.8% (55/241) 6.6% (16/241) 32.9 (10.7) 64.0 (6.0) 2.1 (×/1.9) 15.8% (47/298) 14.5 (4.0) 2.7 (0.8) 25.1 (×/2.0) 89.3 (×/1.3) 9.3 (9.0-9.6) 4.1 (3.5-4.7) 0.7 (0.6-0.8) -1.7 (1.1) -1.9 (1.1) 0.8 (0.7-0.9) -2.3 (-3.0--1.6) -2.6 (-3.3--1.7)

36.6% (26/71) 40.8% (29/71) 15.5% (11/71) 7.0% (5/71) 34.1 (13.5) 63.2 (7.9) 2.5 (×/1.6) 8.2% (8/97) 14.5 (3.2) 2.8 (0.8) 21.8 (×/2.1) 86.8 (×/1.2) 9.3 (9.0-9.8) 4.0 (3.4-4.7) 0.8 (0.7-0.9) -1.5 (1.3) -1.5 (1.3) 0.9 (0.8-1.0) -1.8 (-2.6--1.0) -1.8 (-2.8--1.0)

0.029 0.056 0.082 0.42 0.16 0.32 0.84 0.18 0.030 0.41 0.002 0.012 <0.001 <0.001

0.86 0.39 0.85 0.51

0.46 0.41 0.019 0.063 0.92 0.24 0.26 0.29 0.41 0.54 0.050 0.32 0.079 0.019 0.020 0.003 continued on following page

472

haematologica | 2022; 107(2)

Endocrine complications in iron overload continued from previous page

Side effect D Ferritin† D LIC† D T2*† D TSH† D TSH index† D BMD femur (g/cm2)† D BMD L1-L4 (g/cm2)†

2.8% (9/322) -405.3 (1263.5) -1.7 (6.6) 3.7 (12.1) 0.2 (4.0) 0.1 (0.8) 0.0 (0.2) 0.0 (0.2)

1.9% (2/104) -413.4 (1481.1) -1.7 (4.9) 3.4 (14.7) 0.3 (1.9) 0.2 (0.9) -0.0 (0.2) -0.1 (0.5)

0.63 0.96 0.97 0.85 0.89 0.29 0.23 0.009

CI: 0.10–0.37). According to model 3A, each 50% increase in age was associated with an increase of about 18% in the hazard of an incident new disease (P=0.005) after having adjusted for TSH and number of previous endocrine conditions. Kaplan-Meier survival probability curves are reported in the Online Supplementary Appendix. Table 4A shows the 5-year risk predictions according to levels of age, TSH, and number of endocrine diseases at baseline, based on estimates from model 3B. On average, the whole cohort of patients had a risk of 9.7% (95% CI: 6.3–13.1) of developing an additional endocrine disease within 5 years from the start of therapy with the drug DFX. However, there was considerable variation according the baseline conditions. For example, an average 14year-old patient with a TSH of 3 mIU/L who already suffered from one endocrine disorder had a risk of developing another disorder within 5 years of about a 10%, whereas a 35-year-old patient with a TSH of 5 mIU/L and no disease at baseline had a risk of 50%. Table 4B shows the 1-year risk predictions according to levels of age, TSH and number of endocrine diseases at baseline, based on estimates from model 3B. The overall 1-year risk was 1.1% (95% CI: 0.6–1.7). Fifty-five patients were on therapy with levothyroxine at the beginning of the follow-up. We carried out a sensitivity analysis by running the same analysis on patients who were and were not on levothyroxine separately to see if levothyroxine modified the estimates from the final models. In patients who were (n=55) and were not (n=371) on levothyroxine, the adjusted hazard ratio [HR] for 1 mIU/L increase in TSH was 1.26 (95% CI: 1.02–1.55, P=0.032) and 1.29 (95% CI: 1.07–1.56, P=0.006) respectively. The estimates from the other predictors did not change either. Therefore, TSH was a predictor of additional endocrine disease incidence regardless of levothyroxine administration. In addition, we conducted stratified analyses after splitting the sample at the median follow-up time, or at the age of 16 years, or at 0/1+ prevalent endocrine diseases at baseline. The results from those subgroup analyses were similar to the main one. Given that chronic iron overload is supposed to be the main driver of endocrine complications due to blood transfusions, we have not only used the baseline markers of iron overload (ferritin, LIC, and T2*) in our predictive models, but we have also tested the latest available measures and the difference between initial and final measures. In no cases had those markers any effect on the incidence of endocrine complications. TSH was not correlathaematologica | 2022; 107(2)

Table 3A. Risk factors for developing a new endocrine disease during the follow-up: results from the simplest multiple Cox regression model.

Variable at the beginning of follow-up

Mutually-adjusted hazard ratio

Diseases at baseline (1 increase) TSH (1 mIU/L increase) Age (50% increase)

0.53 1.25 1.18

(95% CI)

(0.43 0.66) (1.13 1.38) (1.05 1.33)

<0.001 <0.001 0.005

CI: Confidence Interval; TSH: thyrotropin.

Table 3B. Risk factors for developing a new endocrine disease during the follow-up: results from the multiple Cox regression model showing the highest adjusted R2, which was used to draw the risk charts.

Variable at the beginning of follow-up

Mutually-adjusted hazard ratio

Diseases at baseline (1 increase) TSH (1 mIU/L increase) Age (5-year increase) Child vs. Adult Interaction Age*Child

0.54 1.26 1.12 6.70 1.59

(95% CI)

(0.43 0.67) <0.001 (1.15 1.39) <0.001 (1.00 1.26) 0.053 (1.32 34.02) 0.022 (1.02 2.47) 0.041

CI: Confidence Interval; TSH: thyrotropin.

ed with any marker of iron overload (Spearman's rhos <0.07, P>0.18).

Discussion Endocrine complications remain the most common and resource-consuming disorders secondary to iron overload in TDT patients. In historical cohorts, disturbances of sexual axis affected 80% of patients, while BMD and short stature were reported in up to 60% and 50% of the overall study population, respectively. Prevalence of hypothyroidism and diabetes ranged from 6% to 14%, while hypoparathyroidism was reported up to 25%.18 In our long-term cohort study of patient affected by TDT treated with the iron-chelating drug DFX, the risk of developing an endocrine complication is generally lower than the previously reported risk, but there is considerable risk variation, according to several parameters such as patient’s age, number of endocrine complications already present before the start of the therapy, and TSH serum concentration. We developed a simple risk chart enabling clinicians to derive an approximate estimate of their patients’ risk. Ferritin, LIC and cardiac T2* are considered as markers of iron overload, but the correlation between those markers 473

M. Casale et al.

and risk of endocrine complications is controversial,20 since many studies have shown no correlation,21–23 confirming our results. This disconnection with iron-overload parameters has been observed also in chronic metabolic syndromes, although substantial evidence shows that the clinical course of these disorders is affected by iron overload.24 The different mechanisms of iron uptake and accumulation among different organs may be responsible of that phenomenon. Iron accumulates in the liver due to transferrin-mediated mechanisms and LIC has inadequate ability to predict that risk in extrahepatic organs.20 The endocrine glands and the

heart, instead, develop pathologic iron overload exclusively through uptake of non-transferrin bound iron (NTBI). The mechanism by which this uptake occurs is controversial, too, but L-type calcium must play a role as it is present in large quantities in cardiomyocytes, pancreatic b cells, in various cell types of the anterior pituitary gland (including gonadotrophs, thyrotrophs, and corticoptrophs), and in the parathyroid-hormone-producing cells of the parathyroid gland.25 Although NTBI composes a very small fraction of body iron, it produces oxidative stress and organ damage.26 While elevated LIC increases patients’ risk of iron-overload

Table 4A. Predicted risk chart for developing a new endocrine disease within 5 years, in percentages. The overall 5-year risk was 9.7%

Age 1 2 4 6 8 10 12 14 16 18 20 25 30 35 40 45 50

Diseases at baseline = 0 TSH=1 TSH=3 TSH=5 6.3 6.5 7.1 7.7 8.4 9.3 10.3 11.4 12.8 14.3 16.1 21.5 27.8 33.5 37.2 39.0 40.5

9.9 10.2 11.1 12.0 13.1 14.4 15.8 17.5 19.4 21.6 23.9 30.5 36.9 41.4 43.9 45.9 47.9

15.2 15.8 17.1 18.5 20.1 21.9 24.0 26.3 28.9 31.6 34.5 41.5 46.9 50.2 52.7 55.2 57.9

Diseases at baseline = 1 TSH=1 TSH=3 TSH=5 3.4 3.6 3.9 4.2 4.6 5.1 5.7 6.3 7.1 8.0 9.1 12.7 17.7 23.6 29.1 32.5 34.0

5.4 5.6 6.1 6.6 7.3 8.0 8.9 9.9 11.0 12.4 14.0 18.9 25.0 31.0 35.1 37.2 38.6

8.5 8.8 9.6 10.4 11.4 12.5 13.8 15.2 16.9 18.9 21.0 27.3 33.8 38.7 41.5 43.3 45.2

Diseases at baseline = 2 TSH=1 TSH=3 TSH=5 1.9 1.9 2.1 2.3 2.5 2.8 3.1 3.5 3.9 4.4 5.1 7.2 10.5 15.2 20.8 26.2 29.5

3.0 3.1 3.3 3.6 4.0 4.4 4.9 5.4 6.1 6.9 7.9 11.1 15.6 21.3 27.1 31.0 32.8

4.7 4.8 5.3 5.7 6.3 6.9 7.6 8.5 9.6 10.8 12.2 16.6 22.4 28.5 33.2 35.5 36.8

Diseases at baseline = 3 TSH=1 TSH=3 TSH=5 1.0 1.0 1.1 1.2 1.4 1.5 1.7 1.9 2.1 2.4 2.8 4.0 6.0 9.1 13.5 19.0 24.2

1.6 1.7 1.8 2.0 2.2 2.4 2.6 3.0 3.3 3.8 4.3 6.2 9.2 13.4 18.9 24.4 28.3

2.5 2.6 2.9 3.1 3.4 3.8 4.2 4.7 5.3 6.0 6.8 9.6 13.8 19.2 25.0 29.5 31.6

Age in years. TSH: thyrotropin.

Table 4B. Predicted risk chart for developing a new endocrine disease within 1 year, in percentages. The overall 1-year risk was 1.1%

Age 1 2 4 6 8 10 12 14 16 18 20 25 30 35 40 45 50

Diseases at baseline = 0 TSH=1 TSH=3 TSH=5 0.6 0.7 0.7 0.8 0.9 1.0 1.1 1.2 1.4 1.5 1.8 2.6 3.9 6.1 9.3 14.0 19.5

1.0 1.1 1.1 1.3 1.4 1.5 1.7 1.9 2.1 2.4 2.8 4.1 6.1 9.2 13.6 19.1 24.4

1.6 1.7 1.8 2.0 2.2 2.4 2.7 3.0 3.4 3.9 4.4 6.3 9.3 13.6 19.1 24.6 28.4

Diseases at baseline = 1 TSH=1 TSH=3 TSH=5 0.3 0.4 0.4 0.4 0.5 0.5 0.6 0.6 0.7 0.8 1.0 1.4 2.1 3.4 5.4 8.5 13.0

0.5 0.6 0.6 0.7 0.7 0.8 0.9 1.0 1.2 1.3 1.5 2.2 3.4 5.2 8.2 12.4 17.8

0.9 0.9 1.0 1.1 1.2 1.3 1.5 1.6 1.8 2.1 2.4 3.5 5.3 8.0 12.1 17.4 22.8

Diseases at baseline = 2 TSH=1 TSH=3 TSH=5 0.2 0.2 0.2 0.2 0.3 0.3 0.3 0.3 0.4 0.4 0.5 0.8 1.2 1.9 3.0 4.9 7.9

0.3 0.3 0.3 0.4 0.4 0.4 0.5 0.6 0.6 0.7 0.8 1.2 1.8 2.9 4.7 7.5 11.6

0.5 0.5 0.5 0.6 0.6 0.7 0.8 0.9 1.0 1.1 1.3 1.9 2.9 4.5 7.1 11.0 16.1

Diseases at baseline = 3 TSH=1 TSH=3 TSH=5 0.1 0.1 0.1 0.1 0.1 0.1 0.2 0.2 0.2 0.2 0.3 0.4 0.6 1.0 1.7 2.8 4.6

0.2 0.2 0.2 0.2 0.2 0.2 0.3 0.3 0.3 0.4 0.4 0.6 1.0 1.6 2.6 4.3 7.0

Age in years. TSH: thyrotropin.

474

haematologica | 2022; 107(2)

0.3 0.3 0.3 0.3 0.3 0.4 0.4 0.5 0.5 0.6 0.7 1.0 1.6 2.5 4.1 6.5 10.3

Endocrine complications in iron overload

complications, there is not a LIC threshold below which cardiac and endocrine iron accumulation does not occur.27 The explanation of this paradox is that many chronically transfused patients have fully-saturated transferrin, regardless of their LIC,28 and, as heart and endocrine glands exclusively accumulate NTBI, it is possible for them to be in positive iron balance even if the total body iron balance (LIC) is neutral or negative.27 Patients who miss chelator doses expose their extrahepatic organs to unrestricted uptake of labile iron species.29 Previous studies reported a correlation between cardiac T2* and manifest endocrinopathies.21,22 However, those studies concerned patients with severe iron overload, with T2* <20 msec, while our sample had average ferritin <1,000 ng/mL, LIC <5 mg/dry weight (dw), and T2* >30 msec which are considered as the acceptable target levels to reach during iron-chelating therapy.18 It has been shown that the iron overload of endocrine glands preceded that of the heart, although both phenomena are mediated by NTBI.11 However, endocrine organs have superior reserve capacity and the clinical manifestations concerning them may appear years after silent iron accumulation.12 When iron overload continues, due to the lack of patient compliance or due to an inadequate dose of iron-binding therapy, the heart starts to show signs of overload, which can be identified through MRI-T2*.20 Therefore, cardiac T2* is not an early indicator of iron overload. We have not found a correlation between T2* and endocrine complications in our cohort of patients, as the vast majority of our patients had an acceptable iron balance. Abnormal cardiac T2* is an excellent marker of NTBI control, but it is insensitive because exposure must be severe and quite prolonged. As a result, abnormal cardiac T2* has a very high positive predictive value for endocrine iron deposition. However, once the heart has been successfully de-ironed, endocrine glands typically retain moderate iron deposition. Finally, even when the endocrine glands have been successfully de-ironed, their functional reserve has been destroyed.4,11,12,20-22 As there is considerable variation in the risk of endocrinopathies in patients without signs of heart and liver overload,17,21,22,30–32 and because those kinds of endocrinopathies, when manifest, are irreversible,17,21,22,30–32 further clinical and laboratory predictors in addition to MRI imaging of endocrine glands are needed to prevent endocrine complications. We proposed our risk chart on the model 3B, in which the association between risk of complication and age was considered as being linear, while adding a binary marker of adulthood and an interaction parameter between age and adulthood. We preferred this model to the simpler model 3A (log-linear age alone) because it had a slightly better R2 and mostly because adults and children affected by TDT are usually treated in separate health care centers and as a matter of fact they define two separate categories. We have developed the largest analysis on endocrine complications in TDT patients ever developed so far and this is the first study providing clear benchmarks for patients’ management. However, the predictive power of our risk chart must be improved and validated. It is plausible that a diagnosis of an endocrinopathy produces a warning effect that is similar to that observed after an abnormal cardiac T2*, which makes patients have better compliance and clinicians increase iron-chelating dose.33 That may explain why previous endocrinopathies were associated with lower incidence of new ones in our haematologica | 2022; 107(2)

sample. Another reason may be that the therapy for an endocrine disease ameliorates the function of other endocrine axes. This has been previously shown for BMD, metabolic syndrome, and glucose and lipid metabolism disorders after treatment for hypothyroidism and hypogonadism.34–36 Furthermore, endocrine glands are not equally vulnerable to the iron toxicity, and patients with more endocrinopathies have already wiped out the most endangered endocrine glands. All our patients were on regular iron-chelation therapy and had acceptable levels of iron load. Therefore, the markers of iron overload were expected to be stable over time or to have minor fluctuations. However, iron overload increased for same patients. If that phenomenon was due to scarce patient compliance and if compliance was associated with our explanatory variables, our estimates may be biased. However, also iron overload measures taken contemporaneously with disease incidence have shown no effect, as well as their deltas. A lack of compliance could have been assumed if at least the latest tests assessing iron overload were associated with higher risk. Heightened TSH has been associated with endothelial dysfunction,37 defined as a diminished bioavailability of nitric oxide (NO) and/or an increase in vasoconstrictive factors such as endothelin (ET-1). That condition has been well documented in thalassemia patients and is associated with cardiac, hepatic and endocrine clinical complications.32,38 Endothelial dysfunction in TDT is a progressive process, starting from childhood, and recent studies found significantly higher plasma levels of asymmetric dimethylarginine (ADMA), a novel risk marker of cardiovascular disease implicated in the pathogenesis of endothelial dysfunction, in very young TDT children.39 So, increased TSH may be an early expression of systemic endothelial dysfunction in TDT which is considered an independent risk factor of future complications.32 TSH appears the best marker of systemic endocrine gland dysfunction, as its measurement is very accurate and widely used in clinical practice,40 differently from the several limits in the assessment of other pituitary hormones, as growth hormone (GH), adrenocorticotropic hormone (ACTH), LH/FSH.41 Furthermore, production of TSH is the last affected by the progressive damage of pituitary gland in TDT patients which impairs firstly GH secretion, followed by LH/FSH and ACTH.42 For all these reasons, TSH may be the sentinel for endocrine gland dysfunction. Along with TSH, age is also associated with endothelial dysfunction, which could be the main driver of endocrine and cardiovascular risk.32 These observations pave the way for the early identification of clinical complications in other metabolic diseases, which have been reported greatly affected by iron overload.24 The variation in risk of complications that our best model could explain was insufficient (25%). Therefore, there must be factors other than those we considered that have some effect on the incidence on endocrine complications. These may include NTBI, transferrin saturation, smoking, other markers of endothelial dysfunction, pancreas and pituitary R2* which weren’t considered. Furthermore, the different chelation history among the study cohort (older patients treated for longer with subcutaneous DFX compared to younger patients treated for longer with oral DFX) creates an inherent age effect to be taken into consideration. The apparent increase in endocrine complications after 475

M. Casale et al.

the age of 12 years is certainly related to different factors, such as the current inability to recognize hypogonadism prior to puberty; the effect of hypogonadism on bone metabolism due to the impact of steroids on bone mineralization; the delay between the start of iron damage in the gland tissues and the onset of overt clinical complications, e.g., diabetes.4,11-13 Furthermore, adolescence is also marked by less physical activity and more adverse body habitus and nutrition that worsen insulin sensitivity. For all these reasons, children are not protected by iron damage in endocrine glands and conversely, they require more aggressive prophylaxis to avoid pituitary and pancreatic iron accumulation which will be clinically manifested only years later, when the functional reserve has been destroyed. Different chelation goals (such as transferrin desaturation and the use of direct endocrine imaging) and alternative chelation strategies are necessary to better protect endocrine glands. In conclusion, although in DFX-treated TDT patients the risk of developing an endocrine complication is generally lower than the previously reported risk, there is considerable risk variation and the burden of these complications remains high. This is the first study providing a practical tool for physicians to identify patients at higher risk of developing endocrine complications. Future research will look at increasing the amount of variation explained from our model and testing further clinical and laboratory predictors, including the assessment of direct endocrine MRI.

References 1. Modell B, Khan M, Darlison M, Westwood MA, Ingram D, Pennell DJ. Improved survival of thalassaemia major in the UK and relation to T2* cardiovascular magnetic resonance. J Cardiovasc Magn Reson. 2008;10 (1):42. 2. Rund D. Thalassemia 2016: modern medicine battles an ancient disease. Am J Hematol. 2016;91(1):15-21. 3. Pinto VM, Poggi M, Russo R, Giusti A, Forni GL. Management of the aging beta-thalassemia transfusion-dependent population the Italian experience. Blood Rev. 2019;38: 100594. 4. Wood JC. Impact of iron assessment by MRI. Hematololgy Am Soc Hematol Educ Program. 2011;2011:443-450. 5. Maffei L, Sorrentino F, Caprari P, et al. HCV infection in Thalassemia syndromes and hemoglobinopathies: new perspectives. Front Mol Biosci. 2020;7:7. 6. El Kholy M, Elsedfy H, Soliman A, Anastasi S, Raiola G, De Sanctis V. Towards an optimization of the management of endocrine complications of thalassemia. J Pediatr Endocrinol Metab. 2014(9-10);27:801-805. 7. Casale M, Filosa A, Ragozzino A, et al. Long-term improvement in cardiac magnetic resonance in b-thalassemia major patients treated with deferasirox extends to patients with abnormal baseline cardiac function. Am J Hematol. 2019;94(3):312-318. 8. Pennell DJ, Porter JB, Cappellini MD, et al. Deferasirox for up to 3 years leads to continued improvement of myocardial T2* in patients with b-thalassemia major. Haematologica 2012; 97(6):842-848. 9. Wood JC, Kang BP, Thompson A, et al. The effect of deferasirox on cardiac iron in thalassemia major: impact of total body iron

476

Disclosures Università degli Studi della Campania “Luigi Vanvitelli” sponsored the study (VALERE project) and received a partial financial support to trial costs from and Novartis Farma SpA which had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication; MC received speaker honoraria and advisory board fees from Novartis Farma SpA; GLF received advisory board fees from Novartis Farma SpA; MP received consulting fees and advisory board fees from Novartis Farma SpA; SP received grant support paid to his institution, advisory board fees and speaker honoraria from Novartis Farma SpA. No other potential conflict of interest relevant to this article was reported. All other authors declare no conflict of interests of any kind. Contributions MC, SP, and AIL designed the study. Each author collected the data from his/her own center and takes responsibility for the accuracy of the data provided. AIL carried out the statistical analysis; MC and AIL drafted the manuscript. All authors contributed to the interpretation of the data and approved the manuscript. The centers in Naples, Genoa, Milan, and Padua are part of the European Reference Network on Rare Hematological Diseases (ERNEuroBloodNet). All centers involved in the study are part of the Italian Society for Thalassemia and Hemoglobinopathies (SITE) and pediatric centers are part of the Italian Association of Pediatric Hematology and Oncology.

stores. Blood. 2010;116(4):537-543. 10. Deugnier Y, Turlin B, Ropert M, et al. Improvement in liver pathology of patients with b-thalassemia treated with deferasirox for at least 3 years. Gastroenterology. 2011;141(4):1202-1211. 11. Noetzli LJ, Papudesi J, Coates TD, Wood JC. Pancreatic iron loading predicts cardiac iron loading in thalassemia major. Blood. 2009;114(19):4021-4026. 12. Noetzli LJ, Panigrahy A, Mittelman SD, et al. Pituitary iron and volume predict hypogonadism in transfusional iron overload. Am J Hematol. 2012;87(2):167-171. 13. Noetzli LJ, Mittelman SD, Watanabe RM, Coates TD, Wood JC. Pancreatic iron and glucose dysregulation in thalassemia major. Am J Hematol. 2012;87(2):155-160. 14. Belhoul KM, Bakir ML, Saned M-S, Kadhim AM, Musallam KM, Taher AT. Serum ferritin levels and endocrinopathy in medically treated patients with b thalassemia major. Ann Hematol. 2012;91(7):1107-1114. 15. Casale M, Citarella S, Filosa A, et al. Endocrine function and bone disease during long-term chelation therapy with deferasirox in patients with b-thalassemia major. Am J Hematol. 2014 89(12):11021106. 16. Poggi M, Sorrentino F, Pugliese P, et al. Longitudinal changes of endocrine and bone disease in adults with b-thalassemia major receiving different iron chelators over 5 years. Ann Hematol. 2016;95(5):757-763. 17. Thuret I, Pondarré C, Loundou A, et al. Complications and treatment of patients with b-thalassemia in France: results of the National Registry. Haematologica. 2010;95(5):724-729. 18. Cappellini MD, Cohen A, Porter J, Taher A, Viprakasit V, editors. Guidelines for the management of transfusion dependent

Thalassaemia (TDT), 3rd edn. Nicosia (CY): Thalassaemia International Federation, 2014 http://www.ncbi.nlm.nih.gov/books/NBK2 69382/ (accessed August 15, 2020). 19. Royston P. Explained variation for survival models. Stata J. 2006;6(1):83-96. 20. Wood JC. Use of magnetic resonance imaging to monitor iron overload. Hematol Oncol Clin North Am. 2014;28(4):747-764. 21. Au W-Y, Lam WW-M, Chu WWC, et al. A cross-sectional magnetic resonance imaging assessment of organ specific hemosiderosis in 180 thalassemia major patients in Hong Kong. Haematologica. 2008;93(5):784-786. 22. Ang AL, Tzoulis P, Prescott E, Davis BA, Barnard M, Shah FT. History of myocardial iron loading is a strong risk factor for diabetes mellitus and hypogonadism in adults with b thalassemia major. Eur J Haematol. 2014;92(3):229-236. 23. Pinto VM, Bacigalupo L, Gianesin B, et al. Lack of correlation between heart, liver and pancreas MRI-R2*: results from long-term follow-up in a cohort of adult b-thalassemia major patients. Am J Hematol. 2018;93(3): E79-82. 24. Fernández-Real JM, Manco M. Effects of iron overload on chronic metabolic diseases. Lancet Diabetes Endocrinol. 2014;2(6):513526. 25. Oudit GY, Trivieri MG, Khaper N, Liu PP, Backx PH. Role of L-type Ca2+ channels in iron transport and iron-overload cardiomyopathy. J Mol Med. 2006;84(5):349-364. 26. Cabantchik ZI. Labile iron in cells and body fluids: physiology, pathology, and pharmacology. Front Pharmacol. 2014;5:45. 27. Noetzli LJ, Carson SM, Nord AS, Coates TD, Wood JC. Longitudinal analysis of heart and liver iron in thalassemia major. Blood. 2008;112(7):2973-2978. 28. Piga A, Longo F, Duca L, et al. High non-

haematologica | 2022; 107(2)

Endocrine complications in iron overload

transferrin bound iron levels and heart disease in thalassemia major. Am J Hematol. 2009;84(1):29-33. 29. Wood JC, Glynos T, Thompson A, et al. Relationship between labile plasma iron, liver iron concentration and cardiac response in a deferasirox monotherapy trial. Haematologica 2011;96(7):1055-1058. 30. Cunningham MJ, Macklin EA, Neufeld EJ, Cohen AR, Thalassemia Clinical Research Network. complications of beta-thalassemia major in North America. Blood. 2004;104 (1):34-39. 31. Vogiatzi MG, Macklin EA, Trachtenberg FL, et al. Differences in the prevalence of growth, endocrine and vitamin D abnormalities among the various thalassaemia syndromes in North America. Br J Haematol. 2009;146(5):546-556. 32. Taher AT, Cappellini MD, Bou-Fakhredin R, Coriu D, Musallam KM. Hypercoagulability and vascular disease. Hematol Oncol Clin

haematologica | 2022; 107(2)

North Am. 2018;32(2):237-245. 33. Pennell DJ, Udelson JE, Arai AE, et al. Cardiovascular function and treatment in bthalassemia major: a consensus statement from the American Heart Association. Circulation. 2013;128(13):281-308. 34. Salzano A, Marra AM, Arcopinto M, et al. Combined effects of growth hormone and testosterone replacement treatment in heart failure. ESC Heart Fail. 2019;6(6):1216-1221. 35. Rochira V. Late-onset hypogonadism: bone health. Andrology. 2020;8(6):1539-1550. 36. Feingold KR, Anawalt B, Boyce A, et al., editors. Endotext. South Dartmouth (MA): MDText.com, Inc., 2000 http://www.ncbi.nlm.nih.gov/books/NBK2 78943/ (accessed August 15, 2020). 37. Niknam N, Khalili N, Khosravi E, Nourbakhsh M. Endothelial dysfunction in patients with subclinical hypothyroidism and the effects of treatment with levothyroxine. Adv Biomed Res. 2016;5:38.

38. Aggeli C, Antoniades C, Cosma C, et al. Endothelial dysfunction and inflammatory process in transfusion-dependent patients with beta-thalassemia major. Int J Cardiol. 2005;105(1):80-84. 39. Gursel O, Tapan S, Sertoglu E, et al. Elevated plasma asymmetric dimethylarginine levels in children with beta-thalassemia major may be an early marker for endothelial dysfunction. Hematology. 2018;23(5):304-308. 40. Clerico A, Trenti T, Aloe R, et al. A multicenter study for the evaluation of the reference interval for TSH in Italy (ELAS TSH Italian Study). Clin Chem Lab Med. 2018;57(2): 259-267. 41. Bidlingmaier M, Strasburger CJ. Growth hormone assays: current methodologies and their limitations. Pituitary. 2007;10(2):115119. 42. Jameson JL, De Groot LJ. Endocrinology-EBook: Adult and Pediatric. Elsevier Health Sciences, 2010.

477

ARTICLE Ferrata Storti Foundation

Iron Metabolism & its Disorders

UBA6 and NDFIP1 regulate the degradation of ferroportin Lisa Traeger,1 Steffen B. Wiegand,1 Andrew J. Sauer,1 Benjamin H.P. Corman,1 Kathryn M. Peneyra,1 Florian Wunderer,1,2 Anna Fischbach,1 Aranya Bagchi,1 Rajeev Malhotra,3 Warren M. Zapol1 and Donald B. Bloch1,4 1

Haematologica 2022 Volume 107(2):478-488

Anesthesia Center for Critical Care Research of the Department of Anesthesia, Critical Care and Pain Medicine, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA; 2Department of Anesthesiology, Intensive Care Medicine and Pain Therapy, University Hospital Frankfurt, Goethe University, Frankfurt, Germany; 3 Cardiovascular Research Center and the Cardiology Division of the Department of Medicine, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA and 4Division of Rheumatology, Allergy and Immunology of the Department of Medicine, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA

ABSTRACT

Correspondence: DONALD B. BLOCH dbloch@mgh.harvard.edu LISA TRAEGER email@lisatraeger.de Received: February 6, 2021. Accepted: July 22, 2021. Pre-published: July 29, 2021

epcidin regulates iron homeostasis by controlling the level of ferroportin, the only membrane channel that facilitates export of iron from within cells. Binding of hepcidin to ferroportin induces the ubiquitination of ferroportin at multiple lysine residues and subsequently causes the internalization and degradation of the ligand-channel complex within lysosomes. The objective of this study was to identify components of the ubiquitin system that are involved in ferroportin degradation. A HepG2 cell line, which inducibly expresses ferroportingreen fluorescent protein (FPN-GFP), was established to test the ability of small interfering (siRNA) directed against components of the ubiquitin system to prevent BMP6- and exogenous hepcidin-induced ferroportin degradation. Of the 88 siRNA directed against components of the ubiquitin pathway that were tested, siRNA-mediated depletion of the alternative E1 enzyme UBA6 as well as the adaptor protein NDFIP1 prevented BMP6- and hepcidin-induced degradation of ferroportin in vitro. A third component of the ubiquitin pathway, ARIH1, indirectly inhibited ferroportin degradation by impairing BMP6-mediated induction of hepcidin. In mice, the AAV-mediated silencing of Ndfip1 in the murine liver increased the level of hepatic ferroportin and increased circulating iron. The results suggest that the E1 enzyme UBA6 and the adaptor protein NDFIP1 are involved in iron homeostasis by regulating the degradation of ferroportin. These specific components of the ubiquitin system may be promising targets for the treatment of iron-related diseases, including iron overload and anemia of inflammation.

Introduction https://doi.org/10.3324/haematol.2021.278530

478

Iron is an essential element that is required for a spectrum of cellular and biological processes including oxygen transport, DNA synthesis and the production of energy. High levels of iron, in the presence of oxygen, may catalyze the production of reactive oxygen species, which are free radicals that may damage cellular proteins and membranes. The level of iron in the body must be tightly regulated to provide sufficient levels to permit fundamental biological processes while preventing the damaging effects of excess iron.1,2 The hepatic hormone hepcidin is a critical regulator of systemic iron homeostasis.3–5 Hepcidin expression is controlled by at least three stimuli: i) increased serum and liver iron, which induce hepcidin via the bone morphogenetic protein (BMP) signaling pathway; ii) increased mediators of inflammation (IL-1b and IL-6), which increase hepcidin via the Jak/Stat pathway; and iii) the hormone erythroferrone, which inhibits BMP signaling by sequestering BMP6 in response to increased erythropoietic demand.6–9 Hepcidin regulates iron homeostasis by controlling the cell surface level of ferroportin, which is the only known membrane channel that facilitates export of iron from within cells.10 Ferroportin is a member of the superfamily

haematologica | 2022; 107(2)

Regulation of ferroportin degradation

of transporters of small molecules, which includes more than 300 membrane-bound proteins.11,12 Ferroportin is predominantly expressed in tissues associated with iron transport including enterocytes, hepatocytes, macrophages and erythrocytes.10,11,13 The protein has 12 membrane-spanning domains, which create a channel through which iron is transported. Binding of hepcidin to the main extracellular cavity of ferroportin causes ligation of ubiquitin molecules to multiple intracellular lysine residues.14 Polyubiquitination of ferroportin induces the internalization of the hepcidinferroportin complex followed by degradation within lysosomes.10,14,15 Degradation of ferroportin results in decreased serum iron, because enterocytes, hepatocytes and macrophages are no longer able to transfer intracellular iron to the circulation.16 In addition to the ability of hepcidin to induce degradation of ferroportin, the hormone is also able to inhibit iron export by directly occluding the iron channel.17 Occlusion of the channel may be especially important for cells, such as mature red blood cells, which lack the endocytic machinery required to degrade ferroportin.17 Ubiquitin is a 76 amino acid polypeptide that can be attached to lysine residues in proteins. The attachment of ubiquitin to a protein regulates the protein’s localization, stability and/or activity.18 The process of ubiquitination involves the activation and transient conjugation of ubiquitin to a carrier protein, with subsequent final ligation of the ubiquitin molecule to a substrate. In general, ubiquitination requires three different kinds of enzymes: a ubiquitin activating enzyme (E1), a ubiquitin conjugating enzyme (E2), and a ubiquitin ligase (E3). The human ubiquitin system encodes two different E1 enzymes (UBA1 and UBA6), approximately 50 different E2 enzymes, and more than 600 E3 enzymes.19–21 The ubiquitin E3 ligases are important for substrate recognition and are divided into three different classes.22 Depending on the class of ligase, the E3 enzyme either directly transfers ubiquitin to a substrate (“HECT” and “RBR” ligases) or acts as an adaptor to facilitate the transfer of ubiquitin from an E2 enzyme directly to the substrate (“RING” E3 ligases).22,23 Binding of the E3 enzyme to the substrate may also require an adaptor protein that acts as a scaffold between the E3 enzyme and the target protein. In this study, an in vitro small interfering RNA (siRNA) screen was performed to determine which proteins in the ubiquitin pathway are involved in ferroportin degradation. A previous study used a modified HEK293 cell line, in which expression of ferroportin was induced by the addition of ponasterone.24 Exogenous hepcidin and putative inhibitors of ferroportin degradation were added to this cell line and the level of ferroportin was then measured. To permit screening for specific enzymes involved in ferroportin ubiquitination without using exogenous hepcidin, we established a HepG2 cell line that expresses the ferroportingreen fluorescent protein (FPN-GFP) fusion protein in response to doxycycline. In this cell line, BMP6 can be used to gradually induce the expression of endogenous hepcidin. The HepG2-FPN-GFP cell line was used to show that the alternative E1 enzyme UBA6 as well as the NEDD4 family binding protein NDFIP1 are able regulate the degradation of ferroportin in response to BMP6, as well as exogenous hepcidin. Depletion of the E3 ligase ARIH1 indirectly inhibited ferroportin degradation by impairing BMP6-mediated hepcidin induction. In vivo, depletion of Ndfip1 in the murine liver increased the level of hepatic ferroportin and increased circulating iron. haematologica | 2022; 107(2)

Methods HepG2-FPN-GFP cell line A plasmid encoding human ferroportin (NM_016917) fused to GFP was a gift from Tomas Ganz (David Geffen School of Medicine, UCLA).14 DNA encoding the fusion protein was ligated into the NheI and NotI sites of plasmid pTRE2hyg (Clontech cat#631014). The plasmid was transfected into the HepG2 “tet-on advanced” cell line (Clonetech cat#630932) using Effectene transfection reagents (Qiagen, Germantown, MD, USA) according to the manufacturer’s instructions and successfully transfected cells were selected using hygromycin (0.4 mg/mL). Individual cell lines were established and were confirmed to express FPN-GFP in the presence of doxycycline (2 µg/mL). One colony, which expressed a high level of FPN-GFP after treatment with doxycycline, was selected for further experiments. HepG2-FPN-GFP cells were maintained in Eagle Minimum Essential Medium, 10% FBS, L-glutamine (2 mM), G418 (100 ng/mL), hygromycin (0.4 mg/mL), penicillin (100 units/mL), and streptomycin (100 µg) at 37°C in 5% CO2 and 95% humidity.

Adeno-associated virus administration All experiments using mice were approved by the Partners Subcommittee on Research Animal Care at Massachusetts General Hospital, Boston, MA, USA (Protocol # 2007N000052). Wild-type mice on a C57BL/6J background were purchased from Jackson Laboratories (Bar Harbor, ME, USA). Animals were fed a standard diet (380 ppm iron). Adeno-associated virus (AAV) short hairpin RNA (shRNA) AAV2/8-GFP-U6-m-Ndfip1-shRNA (AAV2/8-shNdfip1) and AAV2/8-GFP-U6-scrmb-shRNA (AAV2/8shControl) were obtained from Vector BioLabs (Malvern, PA, USA). Eight-week-old male mice were injected intravenously with 1x1011 particles of AAV2/8-shNdfip1 or AAV2/8-shControl via the tail vein. Six weeks later, mice were anesthetized with 4% isoflurane and whole blood was collected by cardiac puncture. Liver and spleen were harvested for further analysis.

Statistical analysis All statistical analyses were performed using GraphPad Prism 8.3.0 (GraphPad Software, San Diego, CA, USA). Data are expressed as mean ± standard deviation (SD). The Shapiro-Wilk test was performed to test for normality. Correlation analysis were performed using Pearson correlation. Comparison of two groups was performed using the Student’s t-test for parametric data and the Mann-Whitney-U test for non-parametric data. Comparison of more than two groups was performed using one-way ANOVA with Tukey post hoc test (for parametric data) or the Kruskal-Wallis test with Dunn’s post hoc test (for non-parametric data). After adjusting for multiple comparisons, a P value <0.05 was considered statistically significant. A detailed description of other methods can be found in the Online Supplementary Appendix.

Results Preparation and characterization of the HepG2-FPNGFP cell line Binding of hepcidin to ferroportin induces the polyubiquitination, internalization and lysosomal degradation of the ligand-channel complex.10 To identify the specific enzymes that mediate ubiquitination of ferroportin, we established a stable HepG2 cell line that inducibly expresses FPN-GFP (HepG2-FPN-GFP) in the presence of doxycycline. Treatment of HepG2-FPN-GFP cells with 2 µg/mL of doxy479

L. Traeger et al.

cycline for 18 hours (h) induced expression of the fusion protein, which was detected at the cell surface (Figure 1A). The FPN-GFP fusion protein was able to export iron, as indirectly indicated by decreased levels of intracellular FTL and FTH1 after FPN-GFP induction (Figure 1B; Online Supplementary Figure S1A). Treatment with hepcidin (50 ng/mL) for 90 minutes (min) caused FPN-GFP to localize to punctate structures in the cytoplasm (Figure 1A), and treatment with hepcidin for 18 h caused degradation of the fusion protein (Figure 1C). Because BMP6 induces HepG2 cells to express hepcidin25 (Figure 1D), we were able to investigate the effect of gradual induction of endogenous hepcidin on ferroportin degradation. Treatment with BMP6 (10 ng/mL) for 18 h caused degradation of the FPN-GFP fusion protein as detected by indirect fluorescence and immunoblotting (Figure 1A and C). Pretreatment with chloroquine (100 µM for 2 h), an inhibitor of lysosomal degradation, prevented BMP6-mediated FPN-GFP degradation and caused FPN-GFP to localize to lysosomes in the cytoplasm (Online Supplementary Figure S1B). BMP6 induces expression of hepcidin through the BMP receptor-SMAD 1/5/8 pathway. After binding to the BMP receptor complex, activated BMP type I receptors phosphorylate SMAD 1/5/8 proteins, which translocate together with SMAD4 to the nucleus, and induce hepcidin expression.26 The siRNA-mediated inhibition of SMAD4 in HepG2-FPN-GFP cells prevented BMP6-mediated degradation of FPN-GFP (Figure 1E). Taken together, these results show that the HepG2-FPN-GFP cell line expresses inducible, functional FPN-GFP. Both BMP6-induced endogenous hepcidin and exogenous hepcidin cause internalization and degradation of the FPN-GFP fusion protein.

E1 enzyme UBA6 is required for ubiquitination of ferroportin The human ubiquitin system encodes two E1 enzymes: UBA1 (also known as in UBE1) and UBA6. To identify the E1 enzyme that is involved in ferroportin degradation, HepG2-FPN-GFP cells were transfected with siRNA that targeted each of the E1 enzymes or with a control siRNA (siControl). Twenty-four hours after transfection with siRNA, cells were treated overnight with doxycycline to induce the expression of FPN-GFP and were then incubated with BMP6 for 18 h. Cells that were treated with siControl and BMP6 had decreased cell surface expression of FPNGFP (Figure 2A). Depletion of UBA1 did not prevent the BMP6-induced localization of FPN-GFP to lysosomes and subsequent degradation. However, depletion of UBA6 prevented BMP6-mediated degradation of ferroportin, as indicated by the persistence of the FPN-GFP fusion protein at the cell surface (Figure 2A). Immunoblotting confirmed that depletion of UBA6, but not UBA1, impaired degradation of FPN-GFP (Figure 2B; Online Supplementary Figure 2A). The successful depletion of each of the E1 enzymes by the appropriate siRNA was confirmed by quantitative polymerase chain reaction (qPCR) (Figure 2C). Successful depletion of UBA6 by siUBA6 was not affected by the addition of BMP6 (Online Supplementary Figure S2B). Depletion of UBA6 might block degradation of FPN-GFP or prevent BMP6-induced expression of hepcidin. To consider this latter possibility, the ability of UBA6 depletion to inhibit the BMP signal transduction pathway was investigated. Depletion of UBA6 had no effect on BMP6-mediated phosphorylation of SMAD1/5/8 (Figure 2B). In addition, siRNA directed against UBA6 did not prevent expression of 480

endogenous hepcidin in HepG2-FPN-GFP cells (Figure 2D). In contrast, siRNA directed against SMAD4 blunted BMP6induced expression of hepcidin. To further demonstrate that depletion of UBA6 blocks degradation of ferroportin, independent of an effect on BMP-induced expression of hepcidin, the effect of exogenous hepcidin on the cellular localization of FPN-GFP in HepG2 cells was investigated. To determine the amount of hepcidin produced by HepG2 cells after treatment with BMP6, HepG2 cells were incubated with BMP6 (10 ng/mL) for 18 h and the amount of hepcidin in the tissue culture medium was measured by enzyme-linked immunosorbant assay (ELISA). Under these conditions, BMP6 induced 3.9 ng/mL (± 0.4 ng/mL) of hepcidin, and this concentration of hepcidin (rather than the much higher dose of 30-50 ng/mL used in other studies10,24) was used to treat cells in subsequent experiments. Cells were transfected with siControl, siUBA1, or siUBA6 and treated overnight with hepcidin (4 ng/mL). In the presence of this low concentration of hepcidin, FPN-GFP expression at the cell surface persisted in UBA6-depleted cells but not in siControl-treated- or siUBA1-treated cells (Figure 2E). The inability of hepcidin to degrade the FPN-GFP fusion protein in siUBA6 treated cells was confirmed by immunoblot (Online Supplementary Figure S2C). Taken together, these results show that UBA6 is required for hepcidin induced internalization and degradation of FPN-GFP.

The adaptor protein NDFIP1 regulates ferroportin degradation To identify additional components of the ubiquitin pathway that might be involved in ferroportin degradation, siRNA directed against different E2 and E3 enzymes, as well as other known components of the ubiquitin pathway, were tested for the ability to inhibit BMP6-mediated degradation of FPN-GFP (Online Supplementary Table S1, n=77). A commercially available library (Dharmacon, Lafayette, CO, USA), which contains siRNA that were previously verified to silence the corresponding targets and to minimize off-target effects, was used in these studies. HepG2 cells transfected with siRNA directed against SMAD4 were used as positive controls for inhibition of BMP6-mediated degradation of FPN-GFP; siControl was used as a negative control. Eighteen hours after treatment with BMP6, the localization of FPN-GFP was determined by immunofluorescence. In the first screen, we identified 23 siRNA directed against different E2 and E3 enzymes that appeared to block FPN-GFP relocalization to the lysosome based on FPN-GFP persistence at the cell surface after BMP6 treatment. In second and tertiary screens, all positive candidates were re-evaluated to exclude false positives. Depletion of each of three E2 enzymes, UBE2R2, UBE2E2 and UBE2J2 partially blocked the internalization of FPNGFP (Online Supplementary Figure S3A and B), while depletion of other, individual E2 enzymes did not impair BMP6mediated FPN-GFP degradation (data not shown). Treatment with pairwise combinations of UBE2R2, UBE2E2 and UBE2J2 or all three of the E2 enzymes did not further prevent the degradation of ferroportin (data not shown), suggesting that additional E2 enzymes participate in FPN ubiquitination. In an initial screen, depletion the NEDD family interacting protein NDFIP1 and the E3 enzyme ARIH1 impaired BMP6-induced FPN-GFP localization to lysosomes and subsequent degradation of the fusion protein (Figure 3A). haematologica | 2022; 107(2)

Regulation of ferroportin degradation

Figure 1. Characterization of the HepG2-FPN-GFP cell line. (A) Images of untreated HepG2 cells, and cells treated with doxycycline (Dox; 2 µg/mL) alone, with Dox followed by hepcidin (50 ng/mL) for 90 minutes (min), with Dox followed by BMP6 (10 ng/mL) for 18 hours (h) are shown. (B) Treatment with Dox induced the expression of the ferroportin-green fluorescent protein (FPN-GFP) fusion protein. Dox-treated cells had reduced levels of intracellular ferritin light-chain (FTL), consistent with increased iron export in cells expressing FPN-GFP. GAPDH was used as a loading control. (C) In the absence of Dox, the fusion FPN-GFP protein was not detected by immunoblot (lanes 1 to 3). In the presence of Dox, FPN-GFP was expressed (lane 4). Treatment with hepcidin (50 ng/mL; lane 5), or BMP6 (10 ng/mL; lane 6), for 18 h caused degradation of the FPN-GFP fusion protein. (D) BMP6 stimulation (10 ng/mL) for 18 h induced hepcidin mRNA expression in HepG2-FPN-GFP cells, as determined by quantitative polymerase chain reaction (qPCR) (mRNA expression relative to control; **P<0.01; Mann-Whitney-U test). (E) Images of siSMAD4 transfected cells treated with Dox (left panel), siSMAD4 transfected cells treated with Dox followed by BMP6 (10 ng/mL; middle panel) and siControl transfected cells treated with Dox followed by BMP6 (right panel) are shown. Small interfering RNA (siRNA)-mediated inhibition of SMAD4 prevented BMP6-mediated degradation of the FPN-GFP fusion protein. The location of nuclei in (A) and (E) are indicated by staining with DAPI (blue). White bar indicates 100 µm.

haematologica | 2022; 107(2)

481

L. Traeger et al.

siNDFIP1 successfully depleted NDFIP1 mRNA in both the absence (Figure 3B) and the presence (Online Supplementary Figure S3C) of exogenous BMP6. siRNA directed against NDFIP1 had no effect on the ability of BMP6 to induce hepcidin expression, demonstrating that the BMP signaling pathway was intact (Figure 3C). To confirm that depletion of NDFIP1 blocks degradation of FPN-GFP, NDFIP1 deplet-

ed cells were treated with exogenous hepcidin (4 ng/mL). Compared to cells that were transfected with siControl, depletion of NDFIP1 inhibited hepcidin-mediated degradation of the FPN-GFP fusion protein (Figure 3D; Online Supplementary Figure S2C). To investigate the possibility that NDFIP1 interacts with ferroportin, HepG2 cells were incubated in the presence or

Figure 2. UBA6 is required for hepcidin-mediated degradation of the ferroportin-green fluorescent protein fusion protein. (A) Cells were transfected with siControl, siUBA1 or siUBA6, and treated with doxycycline (Dox) or Dox followed by BMP6 (10 ng/mL) for 18 hours (h), as indicated. Dox-induced expression of ferroportin-green fluorescent protein (FPN-GFP); depletion of UBA6, but not UBA1, prevented BMP6-mediated FPN degradation. (B) The levels of FPN-GFP and phosphorylated SMAD 1/5/8 in siControl-, siUBA1- and siUBA6-transfected HepG2-FPN-GFP cells in the presence (+) or absence (-) of BMP6 (10 ng/mL for 18 h) are shown. In the absence of Dox, control cells did not express the FPN-GFP fusion protein. Treatment with BMP6 increased the level of pSMAD 1/5/8. UBA6 depletion prevented BMP6-mediated degradation of FPN-GFP. GAPDH was used as a loading control. The immunoblot is representative of 3 separate experiments. (C) UBA1 and UBA6 were successfully depleted using the appropriate small interfering RNA (siRNA), as determined by quantitative polymerase chain reaction (qPCR) (mRNA expression relative to control; *P<0.05; **P<0.01; Mann-Whitney-U test). (D) Treatment with BMP6 (10 ng/mL for 18 h) induced the expression of hepcidin in siControl-transfected cells. Depletion of SMAD4, but not depletion of UBA6, blunted the BMP6-mediated induction of hepcidin in HepG2-FPN-GFP cells (mRNA expression relative to control; **P<0.01; Kruskal-Wallis test). (E) Images of cells transfected with siControl, siUBA1 and siUBA6 are shown. Cells were treated with Dox or Dox followed by hepcidin (4 ng/mL for 18 h), as indicated. Treatment with Dox induced expression of FPN-GFP in siControl treated cells, while incubation with hepcidin caused the degradation of the fusion protein. Depletion of UBA6, but not UBA1, prevented hepcidin-mediated degradation of the FPN-GFP fusion protein. The location of nuclei in (A) and (E) are indicated by staining with DAPI (blue). White bar indicates 100 µm.

482

haematologica | 2022; 107(2)

Regulation of ferroportin degradation

absence of exogenous hepcidin for 20 min, and protein lysates were immunoprecipitated with an antibody directed against NDFIP1. In the absence of hepcidin, a small amount of FPN-GFP was detected in the immunoprecipitated protein lysate. Treatment with hepcidin caused an increase in the amount of FPN-GFP that co-immunoprecipitated with NDFIP1 (Figure 3E). Taken together, the results suggest that NDFIP1 interacts with FPN and is involved in hepcidin-induced FPN internalization and degradation. A second adaptor protein (NDFIP2), which like NDFIP1 facilitates ubiquitination by HECT E3 enzymes, shares 79% similarity with NDFIP127. To investigate the potential role of NDFIP2 in the regulation of FPN, the effect of NDFIP2 depletion on BMP6-induced FPN-GFP degradation was assessed. While siNDFIP1 treatment prevented degradation of FPN-GFP, depletion of NDFIP2 had no effect on BMP6-mediated degradation of the FPN-GFP (Figure 3F; Online Supplementary Figure S3D). NDFIP1 recruits members of the NEDD4 family of E3 ligases to target proteins.28 To investigate whether NEDD4 family members (NEDD4, NEDD4L, ITCH, WWP1, WWP2, SMURF1, SMURF2, HECW1, HECW229) regulate FPN levels, the localization of FPN-GFP in cells treated with siRNA directed against each of these enzymes was examined. None of the siRNA directed against members of the NEDD4 family, either alone or in pair-wise combinations, prevented BMP6 mediated FPN-GFP degradation (Online Supplementary Figure S4A and B). These results indicate that either more than two of these enzymes are involved in BMP6-induced FPN degradation or additional, as yet unidentified, enzymes are able to interact with NDFIP1 and mediate FPN degradation.

ARIH1 indirectly regulates ferroportin by inhibiting BMP6-mediated induction of hepcidin ARIH1 is a member of the Ariadne family of RBR E3 ligases. Treatment of HepG2-FPN-GFP cells with siRNA directed against ARIH1 inhibited BMP6-mediated degradation of FPN-GFP (Figure 3A). ARIH1 was successfully depleted by transfection of siARIH1 in both the absence (Figure 4A) and the presence of BMP6 (Online Supplementary Figure S4C), as determined by qPCR. The addition of low dose exogenous hepcidin to HepG2-FPN-GFP cells, however, reduced the level of FPN-GFP on the surface of ARIH1depleted cells (Figure 4B). The ability of exogenous hepcidin to degrade FPN-GFP in siARIH1 treated cells was confirmed by immunoblot (Online Supplementary Figure S2C). We considered the possibility that depletion of ARIH1 inhibits FPN degradation by interfering with the ability of BMP signaling to induce hepcidin gene expression. In the absence of BMP6, the depletion of ARIH1 reduced basal hepcidin mRNA levels (Figure 4C). Depletion of siARIH1 impaired BMP6-stimulated induction of hepcidin mRNA by 80% (Figure 4D). ARIH1 depletion also inhibited BMP6-mediated induction of ID1, another target of the BMP signaling pathway (Figure 4E). Interestingly, BMP6-induced phosphorylation of SMAD1/5/8 proteins was not affected by ARIH1 depletion (Figure 4F). These results suggest that ARIH1 has an indirect effect on the stability of FPN by altering BMP6- mediated hepcidin induction through a noncanonical pathway. The Ariadne RBR E3 ligase ARIH2 (also known as TRIAD1) is the closest relative to ARIH1 with 54% similarity.30 To consider the possibility that this second member of the Ariadne family is involved in the indirect regulation of haematologica | 2022; 107(2)

FPN, the effect of ARIH2 depletion on BMP6-induced FPNGFP degradation was assessed. In contrast to ARIH1, depletion of ARIH2 had no effect on BMP6-mediated degradation of the FPN-GFP protein expression (Figure 4F; Online Supplementary Figure S4D).

Silencing of Ndfip1 stabilizes hepatic ferroportin in vivo The adaptor protein NDFIP1 was identified as a protein that is involved in FPN degradation in vitro. To address whether NDFIP1 is important for FPN degradation in vivo, mice were injected with an AAV2/8 encoding a shRNA directed against Ndfip1, under the control of a U6 promoter. The AAV serotype 8 was used in these studies because it has a high efficiency of transduction in hepatocytes.31 In both AAV2/8-shNdfip1 and AAV2/8-shControl injected animals, GFP expression was detected in the liver, indicating successful systemic administration of the virus (Figure 5A). In animals injected with AAV2/8-shNdfip1, hepatic Ndfip1 mRNA levels were significantly reduced compared to control animals (Figure 5B). Mice injected with AAV2/8shNdfip1 had a 3-fold increase in FPN protein level in the liver compared to control mice (Figure 5C and D). Hamp mRNA and serum hepcidin levels were similar in both groups, suggesting that higher FPN levels were not caused by induction of hepcidin (Figure 5E and F). Increased hepatic FPN was associated with a 28% increase in serum iron levels in AAV2/8-shNdfip1, compared to AAV2/8-shControl, mice (Figure 5G) and there was a correlation between serum iron and FPN levels (Online Supplementary Figure S5A). Hepatic FTL levels were increased and TfR1 mRNA was decreased in AAV2/8-shNdfip1-treated mice (Online Supplementary Figure S5B and D). As expected because of the targeting of AAV8 to the liver,32 splenic Ndfip1 mRNA levels were not decreased in AAV2/8-shNdfip1 mice (Online Supplementary Figure S5E). The results show that the AAV2/8-mediated depletion of Ndfip1 increases the level of hepatic FPN and that Ndfip1 is required for FPN degradation in the liver.

Discussion This study identified components of the ubiquitin system that are important for FPN degradation. A HepG2 cell line that inducibly expresses functional FPN-GFP fusion protein was established. BMP6-induced expression of hepcidin, which caused the internalization and degradation of the fusion protein and permitted analysis of FPN degradation under conditions in which the level of hepcidin increases gradually. In vitro, the alternative E1 enzyme UBA6, as well as the adaptor protein NDFIP1, were critical for hepcidininduced FPN degradation. Depletion of either UBA6 or NDFIP1 inhibited hepcidin-induced internalization and degradation of FPN-GFP. The E3 ligase ARIH1 indirectly regulated FPN stability by altering BMP6-mediated hepcidin induction through a non-canonical pathway. In vivo, the depletion of Ndfip1 in the murine liver increased the level of hepatic FPN and increased circulating iron. In 2007, UBA6 was identified as a second ubiquitin activating E1 enzyme. UBA1 and UBA6 have non-redundant functions and each enzyme is essential for biological processes.33,34 UBA6 is widely expressed in different tissues but contributes to only approximately 1% of overall cellular ubiquitination.33,35 In addition to activating ubiquitin for 483

L. Traeger et al.

subsequent transfer to the E2 enzyme, UBA6 also activates the ubiquitin-like protein FAT10, which plays a role in the immune response, obesity and aging. However, Fat10-deficient mice do not develop iron overload,36,37 suggesting that FAT10 does not play a direct role in iron homeostasis. In the

present study UBA6 was found to be the E1 enzyme involved in FPN regulation in vitro; depletion of UBA6, but not UBA1, prevented hepcidin-induced FPN degradation in HepG2 cells. In contrast to UBA1, which is known to charge multiple E2 enzymes with ubiquitin, UBA6 transfers

Figure 3. NDFIP1 interacts with ferroportin and regulates ferroportin degradation. (A) Cells were transfected with siControl, siNDFIP1 or siARIH1 and treated with doxycycline (Dox) (left panel) or Dox followed by BMP6 (10 ng/mL for 18 hours [h]), as indicated. Small interfering RNA (siRNA) directed against NDFIP1 or ARIH1 prevented BMP6- mediated ferroportin-green fluorescent protein (FPN-GFP) degradation. White bar indicates 100 µm. (B) Transfection with siNDFIP1 successfully depleted Ndfip1, as determined by quantitative polymerase chain reaction (qPCR) (mRNA expression relative to control; **P<0.01; Student’s t-test). (C) BMP6 (10 ng/mL for 18 h) induced hepcidin expression in siControl transfected cells. Depletion of NDFIP1 did not impair the ability of BMP6 to induce the expression of hepcidin mRNA, as determined by qPCR (mRNA expression relative to control; **P<0.01; One-way ANOVA and Student’s t-test). (D) Cells were transfected with siControl or siNDFIP1 and treated with Dox or Dox followed by hepcidin (4 ng/mL for 18 h) as indicated. In the presence of Dox, the expression of the FPN-GFP fusion protein was induced. Treatment with hepcidin caused FPN-GFP internalization and its subsequent degradation in siControl-treated cells, but not in siNDFIP1-treated cells. White bar indicates 100 µm. (E) A low level of FPN-GFP co-immunoprecipitated with NDFIP1 in Dox-treated HepG2-FPN-GFP cells that were not treated with hepcidin. The level of FPN co-immunoprecipitating with NDFIP1 increased after treatment with hepcidin (50 ng/mL) for 20 minutes (min). Immunoprecipitation was performed using rabbit anti-NDFIP1 antibody. FPN-GFP was detected using a mouse anti-GFP antibody. The immunoblot is representative of 3 separate experiments. (F) HepG2FPN-GFP cells were transfected with siControl-, siNDFIP1- or siNDFIP2 and incubated in the presence (+) or absence (-) of BMP6 (10 ng/mL for 18h). In the absence of Dox, control cells did not express the FPN-GFP fusion protein. (siRNA) directed against NDFIP1, but not NDFIP2, prevented BMP6-induced degradation of the FPNGFP fusion protein, as determined by immunoblot. GAPDH was used as a loading control. The immunoblot is representative of 4 separate experiments.

484

haematologica | 2022; 107(2)

Regulation of ferroportin degradation

Figure 4. ARIH1 regulates BMP6-mediated induction of hepcidin. (A) Transfection with siARIH1 successfully depleted ARIH1 in HepG2-FPN-GFP cells, as determined by quantitative polymerase chain reaction (qPCR) (mRNA expression relative to control; ***P<0.001; Student’s t-test). (B) HepG2-FPN-GFP cells were transfected with siControl or siARIH1 and were treated with doxycycline (Dox) or Dox followed by hepcidin (4 ng/mL for 18 hours [h]) as indicated. In the presence of Dox, the expression of the ferroportin-green fluorescent protein (FPN-GFP) fusion protein was induced. Treatment with hepcidin caused FPN-GFP localization to lysosomes and its subsequent degradation in siControl-treated cells as well as in siARIH1-treated cells. White bar indicates 100 µm. (C) Treatment of HepG2 cells with siARIH1 reduced the basal expression of hepcidin mRNA, as determined by qPCR (mRNA expression relative to control; ***P≤0.001; Student’s t-test). (D) Pretreatment of HepG2 cells with small interfering RNA (siRNA) directed against ARIH1 reduced BMP6-mediated hepcidin mRNA expression (relative to control), as determined by qPCR (***=P≤0.001; One-way ANOVA). (E) BMP6 (10 ng/mL for 18 h) induced the expression of ID1 in siControl transfected cells. Pretreatment of HepG2 cells with siRNA directed against ARIH1 blunted BMP6 induced expression of ID1, as determined by qPCR (mRNA expression relative to control; **P<0.01; ***P<0.001; Oneway ANOVA). (F) Immunoblot showing levels of FPN-GFP and phosphorylated SMAD1/5/8 in siARIH1- or siARIH2-transfected cells in the presence (+) or absence (-) of BMP6 (10 ng/mL for 18 h). The level of pSMAD 1/5/8 was increased in all BMP6-treated cells. siRNA directed against ARIH1, but not ARIH2, prevented BMP6induced degradation of the FPN-GFP fusion protein. GAPDH was used as a loading control. The immunoblot is representative of 4 separate experiments.

haematologica | 2022; 107(2)

485

L. Traeger et al. B

Figure 5. Ndfip1 regulates ferroportin in vivo. (A) Green fluorescent protein (GFP) was detected in the liver of AAV2/8-shControl- and AAV2/8-shNdfip1-injected mice. GAPDH was used as a loading control. (B) AAV2/8-shNdfip1 reduced the level of Ndfip1 mRNA in the liver, as determined by quantitative polymerase chain reaction (qPCR) (mRNA expression relative to control; **P<0.01; Student’s t-test). (C) The level of ferroportin (FPN) in the liver of AAV2/8-shNdfip1-treated mice was increased compared to AAV2/8-shControl-treated mice, as determined by immunoblot. GAPDH was used as a loading control. (D) Densitometric analysis of immunoblot in (C) (***P<0.001; Student’s t-test). (E) HAMP mRNA expression was similar in the liver of AAV2/8-shNdfip1-treated mice compared to AAV2/8-shControl-treated mice as determined by qPCR (mRNA expression relative to control). (F) Serum hepcidin levels were similar in AAV2/8-shNdfip1- and AAV2/8-shControl-treated mice as determined by enzyme-linked immunosorbant assay. (G) Serum iron levels were increased in mice treated with AAV2/8-shNdfip1 (*P<0.05; Student’s t-test).

486

haematologica | 2022; 107(2)

Regulation of ferroportin degradation

ubiquitin to a small number of E2 enzymes.35 Although some E2 enzymes interact with both UBA1 and UBA6, one E2 enzyme (USE1, also known as UBE2Z) is exclusively charged by UBA635,38. In this study, we found that inhibition of USE1 did not interfere with hepcidin-induced FPN degradation (data not shown), indicating that an E2 enzyme other than (or in addition to) USE1 is involved in FPN regulation. In UBA6-depleted cells, UBA1 failed to induce hepcidin-mediated FPN degradation, indicating non-redundant functions of UBA1 and UBA6 in FPN regulation. The results suggest that an as yet unidentified E2 enzyme, exclusively charged by UBA6, plays a role in FPN degradation. Members of the NEDD4 family of HECT-type E3 ligases contain a “WW” domain that interacts with a proline rich PPXY (PY) motif in the target protein. However, some target proteins lack a PY domain and ubiquitination of these proteins requires the presence of adaptor proteins NDFIP1 or NDFIP2 to act as a scaffold between the two proteins. NDFIP proteins contain three transmembrane domains as well as two PY motifs, which interact with the WW domain of several members of the NEDD4 family of E3 ligases.28 In this study, NDFIP1 was shown to interact with FPN in HepG2 cells in vitro and regulates the level of FPN in the liver in vivo. None of the WW domain-containing NEDD4 family members that were tested individually or in pair-wise combination prevented BMP6-induced FPN degradation. The results suggest that several NEDD4 family members may have a redundant role in FPN degradation. Another possibility is that an as yet unknown E3 ligase interacts with the adaptor protein NDFIP1 to ubiquitinate FPN. ARIH1 is a member of the Ariadne family of E3 RBR ligase. ARIH1 is highly expressed in the nucleus, where it interacts with Cajal and PML nuclear bodies.39 ARIH1 associates with neddylated Cullin-RING E3 ligases (CRL) and monoubiquitinates CRL targets.40 In this study, ARIH1 was shown to indirectly regulate FPN stability by altering BMP6-mediated hepcidin induction through a non-canonical pathway. Depletion of ARIH1 blunted basal, as well as BMP6-mediated, hepcidin and ID1 mRNA expression without altering the phosphorylation of SMAD 1/5/8 proteins in response to BMP6. Further studies are needed to elucidate the mechanism as to how ARIH1 regulates hepcidin expression in response to BMP6. NDFIP1 was previously shown to have a role in iron

References 1. Salahudeen AA, Bruick RK. Maintaining mammalian iron and oxygen homeostasis: sensors, regulation, and cross-talk. Ann N Y Acad Sci. 2009;1177:30-38. 2. Pantopoulos K, Porwal SK, Tartakoff A, Devireddy L. Mechanisms of mammalian iron homeostasis. Biochemistry. 2012;51(29):5705-5724. 3. Pigeon C, Ilyin G, Courselaud B, et al. A new mouse liver-specific gene, encoding a protein homologous to human antimicrobial peptide hepcidin, is overexpressed during iron overload. J Biol Chem. 2001;276(11):7811-7819. 4. Nicolas G, Viatte L, Lou D-Q, et al. Constitutive hepcidin expression prevents iron overload in a mouse model of hemochromatosis. Nat Genet. 2003;34(1):97101. 5. Park CH, Valore EV, Waring AJ, Ganz T. Hepcidin, a urinary antimicrobial peptide

haematologica | 2022; 107(2)

homeostasis.41–43 NDFIP1 binds to divalent metal transporter 1 (DMT1), the major iron transporter for non-heme iron import.44 NDFIP1 recruits the NEDD4 family member WWP2 to ubiquitinate DMT143. In vivo, Ndfip1 is involved in the regulation of DMT1 in enterocytes.41 The expression of Dmt1 in enterocytes of Ndfip1 deficient mice is increased under normal iron conditions as well as during iron deficiency. The increased level of Dmt1 leads to increased iron absorption, and under normal dietary iron conditions Ndfip1-deficient mice develop a phenotype resembling classic hereditary hemochromatosis, with increased hepatic, duodenal and serum iron levels.43,45 In this study we show that depletion of Ndfip1 in the liver increased the level of FPN. Ndfip1 appears to regulate two steps in iron metabolism: iron import by DMT1 in enterocytes and iron export by FPN in the liver. Ndfip1-deficient mice were not used in this study, because Ndfip1 deficiency results in a severe inflammatory phenotype caused by hyperactivation of T cells.45,46 In summary, this study demonstrated that the E1 enzyme UBA6 and the adaptor protein NDFIP1 are important for iron homeostasis, regulating the degradation of hepatic FPN. In the future, it may be possible to target specific components of the ubiquitin pathway with small molecules;47 the results of this study may offer novel approaches to treating disorders of iron metabolism. Disclosures No conflicts of interest to disclose. Contributions LT, FW and DBB designed and conceived the study; LT and DBB wrote and edited the manuscript; LT, SBW, AJS, BHPC, KP, AF and DBB performed experiments; LT, RM, AB and DBB analyzed and interpreted the data; DBB and WMZ supervised the study. All authors approved the final version for submission. Funding This study was supported by Luisa Hunnewell and Larry Newman (DBB), the German Research Foundation (TR 1642/11 to LT, WI 5162/2-1 to SBW, Wu 841/1-1 to FW, FI 2429/1-1 to AF), NIH (R01HL142809 to RM, R01DK125786 to AB) and the American Heart Association (18TPA34230025 to RM) 20IOA35360009 to AB).

synthesized in the liver. J Biol Chem. 2001;276(11):7806-7810. 6. Dev S, Babitt JL. Overview of iron metabolism in health and disease. Hemodial Int. 2017;21(Suppl 1):S6-S20. 7. Sebastiani G, Wilkinson N, Pantopoulos K. Pharmacological targeting of the hepcidin/ferroportin axis. Front Pharmacol. 2016;7:160. 8. Wang C-Y, Xu Y, Traeger L, et al. Erythroferrone lowers hepcidin by sequestering BMP2/6 heterodimer from binding to the BMP type I receptor ALK3. Blood. 2020;135(6):453-456. 9. Wunderer F, Traeger L, Sigurslid HH, Meybohm P, Bloch DB, Malhotra R. The role of hepcidin and iron homeostasis in atherosclerosis. Pharmacol Res. 2020;153:104664. 10. Nemeth E, Tuttle MS, Powelson J, et al. Hepcidin regulates cellular iron efflux by binding to ferroportin and inducing its internalization. Science. 2004;306(5704):2090-

2093. 11. Drakesmith H, Nemeth E, Ganz T. Ironing out ferroportin. Cell Metab. 2015;22(5):777787. 12. Lin L, Yee SW, Kim RB, Giacomini KM. SLC transporters as therapeutic targets: emerging opportunities. Nat Rev Drug Discov. 2015;14(8):543-560. 13. Zhang D-L, Ghosh MC, Ollivierre H, Li Y, Rouault TA. Ferroportin deficiency in erythroid cells causes serum iron deficiency and promotes hemolysis due to oxidative stress. Blood. 2018;132(19):2078-2087. 14. Qiao B, Sugianto P, Fung E, et al. Hepcidininduced endocytosis of ferroportin is dependent on ferroportin ubiquitination. Cell Metab. 2012;15(6):918-924. 15. Ross SL, Tran L, Winters A, et al. Molecular mechanism of hepcidin-mediated ferroportin internalization requires ferroportin lysines, not tyrosines or JAK-STAT. Cell Metab. 2012;15(6):905-917.

487

L. Traeger et al. 16. Camaschella C, Nai A, Silvestri L. Iron metabolism and iron disorders revisited in the hepcidin era. Haematologica. 2020;105 (2):260-272. 17. Aschemeyer S, Qiao B, Stefanova D, et al. Structure-function analysis of ferroportin defines the binding site and an alternative mechanism of action of hepcidin. Blood. 2018;131(8):899-910. 18. Pickart CM, Eddins MJ. Ubiquitin: structures, functions, mechanisms. Biochim Biophys Acta. 2004;1695(1-3):55-72. 19. Morreale FE, Walden H. Types of ubiquitin ligases. Cell. 2016;165(1):248-248. 20. hen M, Schmitt S, Buac D, Dou QP. Targeting the ubiquitin-proteasome system for cancer therapy. Expert Opin Ther Targets. 2013;17(9):1091-1108. 21. Stewart MD, Ritterhoff T, Klevit RE, Brzovic PS. E2 enzymes: more than just middle men. Cell Res. 2016;26(4):423-440. 22. Ardley HC, Robinson PA. E3 ubiquitin ligases. Essays Biochem. 2005;41:15-30. 23. Weber J, Polo S, Maspero E. HECT E3 ligases: a tale with multiple facets. Front Physiol. 2019;10:370. 24. Fung E, Sugianto P, Hsu J, Damoiseaux R, Ganz T, Nemeth E. High-throughput screening of small molecules identifies hepcidin antagonists. Mol Pharmacol. 2013;83(3):681690. 25. Meynard D, Kautz L, Darnaud V, CanonneHergaux F, Coppin H, Roth M-P. Lack of the bone morphogenetic protein BMP6 induces massive iron overload. Nat Genet. 2009;41(4):478-481. 26. Xiao X, Alfaro-Magallanes VM, Babitt JL. Bone morphogenic proteins in iron homeostasis. Bone. 2020;138:115495. 27. Shearwin-Whyatt LM, Brown DL, Wylie FG, Stow JL, Kumar S. N4WBP5A (Ndfip2), a Nedd4-interacting protein, localizes to multivesicular bodies and the Golgi, and has a

488

potential role in protein trafficking. J Cell Sci. 2004;117(Pt 16):3679-3689. 28. Harvey KF, Shearwin-Whyatt LM, Fotia A, Parton RG, Kumar S. N4WBP5, a potential target for ubiquitination by the Nedd4 family of proteins, is a novel Golgi-associated protein. J Biol Chem. 2002;277(11):93079317. 29. Ingham RJ, Gish G, Pawson T. The Nedd4 family of E3 ubiquitin ligases: functional diversity within a common modular architecture. Oncogene. 2004;23(11):1972-1984. 30. Kelsall IR, Duda DM, Olszewski JL, et al. TRIAD1 and HHARI bind to and are activated by distinct neddylated Cullin-RING ligase complexes. EMBO J. 2013;32(21):2848-2860. 31. Nam H-J, Lane MD, Padron E, et al. Structure of adeno-associated virus serotype 8, a gene therapy vector. J Virol. 2007;81(22):1226012271. 32. Snyder RO, Miao CH, Patijn GA, et al. Persistent and therapeutic concentrations of human factor IX in mice after hepatic gene transfer of recombinant AAV vectors. Nat Genet 1997;16(3):270-276. 33. Barghout SH, Schimmer AD. E1 Enzymes as Therapeutic Targets in Cancer. Pharmacol Rev. 2021;73(1):1-56. 34. Groettrup M, Pelzer C, Schmidtke G, Hofmann K. Activating the ubiquitin family: UBA6 challenges the field. Trends Biochem Sci. 2008;33(5):230-237. 35. Jin J, Li X, Gygi SP, Harper JW. Dual E1 activation systems for ubiquitin differentially regulate E2 enzyme charging. Nature. 2007;447(7148):1135-1138. 36. Canaan A, Yu X, Booth CJ, et al. FAT10/diubiquitin-like protein-deficient mice exhibit minimal phenotypic differences. Mol Cell Biol. 2006;26(13):5180-5189. 37. Canaan A, DeFuria J, Perelman E, et al. Extended lifespan and reduced adiposity in mice lacking the FAT10 gene. Proc Natl Acad

Sci U S A. 2014;111(14):5313-5318. 38. Wang F, Zhao B. UBA6 and its bispecific pathways for ubiquitin and FAT10. Int J Mol Sci. 2019;20(9):2250. 39. Elmehdawi F, Wheway G, Szymanska K, et al. Human homolog of drosophila Ariadne (HHARI) is a marker of cellular proliferation associated with nuclear bodies. Exp Cell Res. 2013;319(3):161-172. 40. Scott DC, Rhee DY, Duda DM, et al. Two distinct types of E3 ligases work in unison to regulate substrate ubiquitylation. Cell. 2016;166(5):1198-1214. 41. Foot NJ, Leong YA, Dorstyn LE, et al. Ndfip1deficient mice have impaired DMT1 regulation and iron homeostasis. Blood. 2011;117(2):638-646. 42. Foot NJ, Gembus KM, Mackenzie K, Kumar S. Ndfip2 is a potential regulator of the iron transporter DMT1 in the liver. Sci Rep. 2016;6:24045. 43. Foot NJ, Dalton HE, Shearwin-Whyatt LM, et al. Regulation of the divalent metal ion transporter DMT1 and iron homeostasis by a ubiquitin-dependent mechanism involving Ndfips and WWP2. Blood. 2008;112(10): 4268-4275. 44. Yanatori I, Kishi F. DMT1 and iron transport. Free Radic Biol Med. 2019;133:55-63. 45. Oliver PM, Cao X, Worthen GS, et al. Ndfip1 protein promotes the function of itch ubiquitin ligase to prevent T cell activation and T helper 2 cell-mediated inflammation. Immunity. 2006;25(6):929-940. 46. Nemeth E, Valore EV, Territo M, Schiller G, Lichtenstein A, Ganz T. Hepcidin, a putative mediator of anemia of inflammation, is a type II acute-phase protein. Blood. 2003;101(7):2461-2463. 47. Deng L, Meng T, Chen L, Wei W, Wang P. The role of ubiquitination in tumorigenesis and targeted drug discovery. Signal Transduct Target Ther. 2020;5(1):11.

haematologica | 2022; 107(2)

ARTICLE

Non-Hodgkin Lymphoma

Early detection of T-cell lymphoma with T follicular helper phenotype by RHOA mutation analysis Rachel Dobson,1 Peter Y. Du,1 Lívia Rásó-Barnett,2 Wen-Qing Yao,1 Zi Chen,1 Calogero Casa,2 Hesham EI-Daly,2 Lorant Farkas,2,3 Elizabeth Soilleux,1,4 Penny Wright,4 John W. Grant,4 Manuel Rodriguez-Justo,5 George A. Follows,6 Hala Rashed,7 Margarete Fabre,6,8 E. Joanna Baxter,6 George Vassiliou,6,8 Andrew Wotherspoon,9 Ayoma D. Attygalle,9 Hongxiang Liu2 and Ming-Qing Du1,4 1

Division of Cellular and Molecular Pathology, Department of Pathology, University of Cambridge, Cambridge, UK; 2The Haematopathology and Oncology Diagnostic Service, Addenbrooke's Hospital, Cambridge University Hospitals NHS Foundation Trust, Cambridge, UK; 3Department of Pathology, Akershus University Hospital, Lorenskog, Norway; 4Department of Histopathology, Addenbrooke's Hospital, Cambridge University Hospitals NHS Foundation Trust, Cambridge, UK; 5Department of Pathology, Royal Free and University College Medical School, London, UK; 6Department of Haematology, Addenbrooke's Hospital, Cambridge University Hospitals NHS Foundation Trust, Cambridge, UK; 7Department of Cellular Pathology, University Hospitals of Leicester, East Midlands Pathology Services, Leicester, UK; 8Wellcome-MRC Cambridge Stem Cell Institute, Jeffrey Cheah Biomedical Centre, University of Cambridge, Cambridge, UK and 9Histopathology Department, The Royal Marsden Hospital, London, UK

Ferrata Storti Foundation

Haematologica 2022 Volume 107(2):489-499

ABSTRACT

ngioimmunoblastic T-cell lymphoma (AITL) and peripheral T-cell lymphoma with T follicular helper phenotype (PTCL-TFH) are a group of complex clinicopathological entities that originate from T follicular helper cells and share a similar mutation profile. Their diagnosis is often a challenge, particularly at an early stage, because of a lack of specific histological and immunophenotypic features, paucity of neoplastic T cells and prominent polymorphous infiltrate. We investigated whether the lymphoma-associated RHOA Gly17Val (c.50G>T) mutation, occurring in 60% of cases, is present in the early “reactive” lesions, and whether mutation analysis could help to advance the early diagnosis of lymphoma. The RHOA mutation was detected by quantitative polymerase chain reaction with a locked nucleic acid probe specific to the mutation, and a further peptide nucleic acid clamp oligonucleotide to suppress the amplification of the wild-type allele. The quantitative polymerase chain reaction assay was highly sensitive and specific, detecting RHOA Gly17Val at an allele frequency of 0.03%, but not other changes in Gly17, nor in 61 controls. Among the 37 cases of AITL and PTCL-TFH investigated, RHOA Gly17Val was detected in 62.2% (23/37) of which 19 had multiple biopsies including preceding biopsies in ten and follow-up biopsies in 11 cases. RHOA Gly17Val was present in each of these preceding or follow-up biopsies including 18 specimens that showed no evidence of lymphoma by combined histological, immunophenotypic and clonality analyses. The mutation was seen in biopsies 0-26.5 months (mean 7.87 months) prior to the lymphoma diagnosis. Our results show that RHOA Gly17Val mutation analysis is valuable in the early detection of AITL and PTCL-TFH.

Introduction Angioimmunoblastic T-cell lymphoma (AITL), peripheral T-cell lymphoma with T follicular helper phenotype (PTCL-TFH) and follicular T-cell lymphoma are a group of complex clinicopathological entities that originate from T follicular helper (TFH) cells. These lymphomas are the most common among various T-cell malignancies in Western countries. Patients with these lymphomas commonly show an aggressive clinical course, with a median 3-year survival rate of only 30%

haematologica | 2022; 107(2)

Correspondence: MING-QING DU mqd20@cam.ac.uk Received: July 9, 2020. Accepted: January 22, 2021. Pre-published: February 11, 2021. https://doi.org/10.3324/haematol.2020.265991

489

R. Dobson et al.

(https://www.hmrn.org/ statistics/disorders/34). The poor clinical outcome is largely a consequence of the lack of understanding of their molecular mechanism and targeted therapy. The diagnosis of these lymphoma entities, particularly at an early stage, is a challenge because of the lack of specific clinical and histological features, low tumor cell content and the presence of prominent polymorphous inflammatory infiltrates that often mask the neoplastic cells. To help the diagnosis, T-cell clonality analysis is commonly used, but this is frequently not helpful because of the paucity of malignant T cells. The diagnostic difficulty is further exacerbated by the increasing use of needle core biopsies to avoid more invasive surgical excision. When a lymphoproliferative lesion is suspected but uncertain on histological diagnosis, patients are commonly subjected to a “watch and wait” approach, and further biopsied when showing signs of disease progression. There is a wide range of genetic changes in T-cell lymphomas with TFH phenotype, including early TET2 and DNMT3A mutations, which occur in hematopoietic stem cells, and late genetic changes, which are specifically seen in lymphoma cells.1-4 Among the latter, RHOA mutation is the most frequent,2,5-9 occurring in 60-70% of cases, with Gly17Val (c.50G>T) accounting for 95% of the changes. Interestingly, studies of mouse models have suggested that RHOA Gly17Val mutation induces TFH cell differentiation and, together with loss-of-function TET2 mutations, can promote the development of AITL-like lymphomas.10-12 Importantly, RHOA mutation is preferentially associated with the above lymphoma entities and has, so far, been shown only in the neoplastic T-cell population, but not in reactive B and T cells in these malignant conditions.2,5,13,14 Studies to date suggest that the RHOA Gly17Val mutation could be used as a marker for the diagnosis of AITL and PTCL-TFH.6,15 However, it is unclear whether the mutation is present in tissue biopsies prior to the diagnosis of AITL and whether mutation analysis could help in the early detection of these lymphomas. To investigate these issues, we established a highly sensitive quantitative polymerase chain reaction (qPCR) assay to detect the RHOA mutation, and screened a large cohort (n=37 cases) of AITL and PTCL-TFH with multiple sequential biopsies together with 61 controls.

nine cases of AITL or PTCL-TFH with known RHOA mutations (Gly17Val, c.50G>T, n=6 with variant allele frequency [VAF] ranging from 2-32%; Gly17Leu, c.49-50GG>TT, n=2; Gly17Glu, c.50G>A, n=1; Ser26Arg, c.76A>C, n=1) were available from our previous study,2 while crude DNA samples were available from routine clonality analysis or similarly prepared. The histology of these specimens was reviewed by specialist hematopathologists.

Quantitative polymerase chain reaction with peptide nucleic acid clamp and locked nucleic acid probe The qPCR with peptide nucleic acid (PNA) clamp and locked nucleic acid (LNA) probe was performed as previously described.16 The qPCR assay contained two probes and a PNA clamp (Figure 1A, Online Supplementary Table S1). The total probe served as an internal control to monitor the PCR performance. The PNA clamp specifically and strongly binds to the wild-type DNA sequence, resisting the 5’ nuclease activity of Taq DNA polymerase, and thus blocks the wild-type allele from PCR amplification. This results in preferential amplification of the mutant allele which is detected specifically by the LNA mutant probe. PCR primers with various probes were first tested using purified DNA samples, with the optimized conditions outlined below (Online Supplementary Table S1). Briefly, the PNA-LNA PCR was carried out in a 20 µL reaction containing 10 mL Premix Ex Taq (Probe qPCR) Master Mix (Takara, Shiga, Japan), 0.2 mM of each forward and reverse primer, 0.1 µM of each total and mutant probe, 0.05 mM of the PNA clamp probe, and 2 mL of crude DNA or 25 ng of purified DNA. Real-time PCR was carried out in triplicate using Quantstudio 6 (Thermo Fisher Scientific, Waltham, MA, USA) with denaturation at 95°C for 30 s followed by 45 cycles at 95°C for 3 s, and at 62°C for 30 s.

Targeted sequencing using the Fluidigm Access Array and Illumina MiSeq Targeted sequencing was performed on a selected case with consecutive tissue biopsies as described in our recent study.2 Each of the DNA samples was investigated in duplicate for mutations in TET2, DNMT3A, IDH2, RHOA, PLCG1, CCND3, CD28 and TNFRSF21 by Fluidigm PCR and Illumina MiSeq sequencing. Sequence reads alignment, variant calling, filtering to eliminate false positive and benign changes were carried out according to our previously established protocols.2 Only the reproducible variants that appeared in both replicates were regarded as true changes.

BaseScope in situ hybridization Methods Tissue materials and DNA extraction The use of archival tissues for research was approved by the ethics committees of the institutions involved. In total, 78 tissue specimens from 37 patients (multiple consecutive specimens in 29 cases) with AITL (n=35) or PTCL-TFH (n=2), together with 61 controls (13 with reactive lymph nodes with paracortical expansion of T cells, 10 with classic Hodgkin lymphoma and 1 with marginal zone lymphoma with a prominent background of T cells) and peripheral blood DNA samples from 16 individuals (TET2 and DNMT3A mutations in 2, TET2 mutation in 4, DNMT3A mutation in 5) with clonal hematopoiesis of indeterminate potential (CHIP) (1 individual classed as having high-risk CHIP) were successfully investigated. DNA samples were prepared from a whole tissue section of each specimen. Purified DNA was obtained using a QIAamp DNA-Micro kit (Qiagen), while crude DNA was obtained by digesting samples at 37°C overnight with proteinase K and NP-40 buffer. Purified DNA samples from

490

This was performed on a selected case. The sequence of the clonal TRB rearrangement in a case of AITL was available from a recent study.2 Based on the unique VDJ junctional sequence, unique BaseScope probes were designed and used to identify the lymphoma T cells by in situ hybridization. The BaseScope in situ hybridization was carried out according to the manufacturer's instructions (Advanced Cell Diagnostics, Newark, CA, USA),2 with the conditions optimized using the AITL specimen from which the clonal TRB rearrangement was sequenced. Tonsils were used as a negative control.

Results Peptide nucleic acid – locked nucleic acid quantitative polymerase chain reaction is highly sensitive and specific for the detection of RHOA Gly17Val (c.50G>T) Upon the optimization of PCR conditions for RHOA mutation detection, we first determined the sensitivity of haematologica | 2022; 107(2)

RHOA mutation analysis in early lymphoma diagnosis

Figure 1. Detection of RHOA p.Gly17Val (c.50G>T) by real time polymerase chain reaction with peptide nucleic acid/locked nucleic acid probes. (A) Schematic illustration of the real time polymerase chain reaction (PCR) design. The total probe is used as an internal control to monitor the PCR performance. The peptide nucleic acid (PNA) clamp binds to the wild-type sequence, resists the 5’ nuclease activity of Taq DNA polymerase and thus blocks the wild-type allele from PCR amplification. This results in preferential amplification of the mutant allele which is detected specifically by the locked nucleic acid (LNA) mutant probe. (B) The PCR assay is highly sensitive, capable of detecting the RHOA mutation at a variant allele frequency (VAF) of 0.03% based on serial dilutions of an angioimmunoblastic T-cell lymphoma sample with known mutation allele frequency by next-generation sequencing (left panel). For simplicity, only the LNA mutant but not total probe signals are shown. The PCR assay shows a linear correlation among the serial dilutions (right panel).

the qPCR assay using serial dilutions of three purified DNA samples with known VAF of Gly17Val (c.50G>T) into tonsil DNA. The PNA-LNA qPCR was highly sensitive, capable of detecting the mutation at a VAF of 0.032% (Figure 1B). As expected, the qPCR was highly specific to Gly17Val (c.50G>T) and showed no detectable signal of the LNA mutant probe with the DNA samples harboring other RHOA mutations including Gly17Leu (c.49-50GG>TT) and Gly17Glu (c.50G>A), or tonsillar DNA (Online Supplementary Figure S1A). As crude DNA preparations were routinely used for clonality analysis in our clinical diagnostic laboratories, we tested whether such crude DNA samples were amenable to the above PNA-LNA qPCR (Online Supplementary Figure S1C). Of the ten crude DNA samples initially tested, nine yielded excellent amplification comparable with the results from purified DNA samples. We then used crude DNA preparations for the qPCR in the remaining investigations, with high quality results from 139 of the 144 samples investigated. The reason that the other five samples failed to support qPCR was haematologica | 2022; 107(2)

essentially the poor quality of the DNA, as evidenced by quality control PCR.

RHOA Gly17Val (c.50G>T) is detected in initial biopsies not diagnostic for lymphoma by conventional approaches To confirm that the RHOA mutation was lymphomaspecific, not associated with reactive conditions proliferations or other lymphoproliferative disorders with enriched T cells, we investigated 45 tissue biopsies including 27 from lymph nodes, for which T-cell clonality analysis was requested during routine histological diagnosis, but showed no evidence of a T-cell lymphoma. These included 13 specimens showing paracortical T-cell expansion or enriched CD4+ T cells, and ten classic Hodgkin lymphomas with a prominent background of T cells. The qPCR was successful for all these specimens, as indicated by the total probe control, but they were all negative for the RHOA mutation. Similarly, we also showed absence of the RHOA mutation in peripheral blood lymphocytes (Online Supplementary Figure S1D) 491

R. Dobson et al.

Figure 2. Analysis of RHOA (c.50G>T; p.Gly17Val) mutation by quantitative polymerase chain reaction in the RHOA-positive cases with longitudinal biopsies in patients with angioimmunoblastic T-cell lymphoma (only including cases with at least one biopsy classed as not diagnostic [“group B” or “group C”]). The diagnoses of these specimens were reviewed, and categorized into three groups: group A (filled circles): lymphoma diagnosed by histology and immunophenotype, further supported by clonal TCR gene rearrangement; group B (half-filled circles): lymphoma not diagnostic by histology and immunophenotype alone, but ascertained by clonal TCR gene rearrangement; group C (open circles): lymphoma not diagnosed by combined analyses. Red (regardless of the symbol: including outline only, half-filled, and completely filled symbols) indicates RHOA p.Gly17Val mutant positive biopsy. Gray indicates RHOA p.Gly17Val status unknown. E denotes extranodal biopsies. *denotes lymph node excision biopsies and all others are core biopsy specimens. An open square denotes a diagnosis of classic Hodgkin lymphoma with no apparent evidence of angioimmunoblastic T-cell lymphoma (AITL). Case 38 lacks a final diagnosis as the patient died, indicated by a dashed line. Cases with only group A multiple biopsies are not included in this figure.

from 16 individuals with CHIP (mutational information provided in Online Supplementary Table S2), in keeping with the findings from whole exome or panel sequencing.17,18 We then investigated 37 cases of AITL (n=35) or nodal PTCL-TFH (n=2) with unknown RHOA mutation status, of which 29 had multiple longitudinal biopsies (2-5) available for RHOA analysis. RHOA mutation was seen in 23 cases, but was negative in the remaining 14 cases. Among the 23 cases with RHOA mutation, 19 cases had multiple biopsies. We reviewed the original histological diagnosis in each of these specimens and categorized them into three groups. Group A specimens (n=26) showed clear evidence of lymphoma by histology and immunophenotype, further supported by clonal TCR rearrangement. Group B (n=6) had suspicious histological and immunophenotypic findings, although not entirely diagnostic for lymphoma, which was ascertained by detection of strong clonal TCR gene rearrangements. Finally, group C mutations (n=18) were not diagnostic for lymphoma by combined histological, immunophenotypic 492

and clonality analyses (Online Supplementary Table S3). Interestingly, RHOA mutation was detected in each of these specimens, in the positive cases, including those not diagnostic for lymphoma by conventional integrated diagnostic investigations (Figure 2).

Characteristics of biopsies not diagnostic for lymphoma but positive for RHOA mutation Of the 23 cases of AITL or PTCL-TFH with RHOA mutation, the initial biopsy was not diagnostic in ten cases including two with an excisional lymph node specimen. The time from the initial non-diagnostic biopsy to that establishing AITL or PTCL-TFH ranged from 0 to 26.5 months, with an average of 7.87 months. All these initial non-diagnostic biopsies showed polyclonal (n=4), weak clonal (n=5) or oligoclonal (n=2) TCR gene rearrangements or failed (n=1) by BIOMED-2 TRB and TRG clonality analysis. Seven of the initial non-diagnostic biopsies were also subjected to BIOMED-2 B-cell clonality analysis and six showed a clonal IG gene rearrangement. Epstein-Barr virus (EBV)-encoded small RNA haematologica | 2022; 107(2)

RHOA mutation analysis in early lymphoma diagnosis

Figure 3. Histological and immunophenotypic findings in case 30. The first lymph node excision biopsy shows partial effacement of the lymph node architecture by polymorphous infiltrates, particularly B cells with plasmacytoid differentiation. A large proportion of B cells were EBER-positive and showed IG κ light chain restriction (not shown). The lymphoid follicles appear to be reactive and shows no apparent expansion of T follicular helper (TFH) cells, with only a few CD10-positive cells spilling out of the germinal center, consistent with pattern-1 histology of angioimmunoblastic T-cell lymphoma (AITL). The second lymph node biopsy shows effacement of the lymph node architecture by medium-sized atypical lymphoid cells with regressed follicles. There is a prominent proliferation of follicular dendritic cell meshworks and high endothelial venules with atypical lymphoid cells clustered in their vicinity. The atypical lymphoid cells are T cells expressing TFH markers, spilling out of the germinal center to the interfollicular region. EBER in situ hybridization shows only scattered positive cells.

haematologica | 2022; 107(2)

493

R. Dobson et al.

CD3

CD4

PD1

BCL6

CD20

CD30

EBER

CD30

CD20

OCT2

CD4

Figure 4. Histological and immunophenotypic findings in case 7. The first biopsy shows a subcutaneous perivascular infiltrate of CD3+ T cells with vasculitic features. The third biopsy shows cardinal features of angioimmunoblastic T-cell lymphoma (AITL) and also a prominent pleomorphic infiltrate including an EBER-positive B-cell population with Hodgkin and Reed/Sternberg (HRS)-like morphology and immunophenotype. The fifth biopsy shows no apparent evidence of AITL, but a polymorphous infiltrate with a more prominent EBER-positive B-cell population that has HRS cell morphology and immunophenotype, rosetting by CD4+ T cells.

494

haematologica | 2022; 107(2)

RHOA mutation analysis in early lymphoma diagnosis

Table 1. Summary of histological, immunohistochemical, clonality analysis and genetic findings in case 7.

First biopsy Time after the 1st biopsy Clinical presentation

Skin rash, fevers; All self-resolved with no therapy

Biopsy site

Skin punch biopsy Histology Mild perivascular and infiltrate, predominantly Immuno-phenotype CD3+ T cells, admixed histiocytes and B cells

Diagnosis T-cell clonality TRG-A TRG-B TRB-A TRB-B TRB-C B-cell clonality BaseScope-ISH for TRB-V5Targeted sequencing

RHOA c.50G>T

Second biopsy

Third biopsy

Fourth biopsy

Fifth biopsy

11 months later

11.5 months later

33 months later

35 months later

Episode of non-specific symptoms; lymphadenopathy revealed by CT, but relatively low PET signal; opted for close surveillance. Lymph node core biopsy Predominant infiltrate of small to medium sized T cells with vasculocentric pattern, CD4+, some CD10+ and BCL6+ T cells possibly spilled outside B-cell follicle, expanded FDC meshworks; scattered large EBER+ B cells: CD30+, CD15 weak+

Left axillary lymph node core biopsy Very similar to the 2nd biopsy. Predominantly small to medium sized T cells associated with HEV proliferation, CD4+, PD1+, BCL6 variable+; Expanded FDC and pleomorphic infiltrate, scattered large B cells with HRS morphology CD30+, CD15 weak+ AITL lymphoma

Slowing enlarging lymph node with increasing symptoms. Treated with 6 cycles of R-GCVP* and achieved CR; Ongoing remission in May 2020 Right axillary lymph Right axillary lymph node core biopsy node core biopsy Lymph node Normal structure structure effaced by largely effaced polymorphous infiltrate by polymorphous with scattered infiltrate with large atypical cells prominent large expressing CD30, atypical cells CD20 (weak), expressing CD30, CD79, PAX5, BCL6, CD20, CD79a, MUM1, OCT2, BOB1, PAX5, MUM1, and EBER+, BCL6(weak), but not CD15. and EBER+, but CD3 T cells: not CD15, CD10, + largely CD4 , CYCLIN D1, ALK some ICOS+, and CD25. occasional PD1+, but No detectable CD10– and CXCL13– T-cell abnormalities Classic Hodgkin Classic Hodgkin lymphoma

Panniculitis

AITL

n/a n/a n/a n/a n/a n/a Few positive cells

217 bp Poly 245 bp Poly 302 bp n/a Diffuse positive

217 bp Poly 245 bp Poly Poly n/a Diffuse positive

Poly Poly Poly Poly Poly Poly n/a

Poly Poly Poly Poly 187bp Poly Negative

n/a

DNMT3A (VAF: 34%) (c.920C>T; p.P307L) TET2 (VAF: 8%) (c.3646C>T; p.R1216X) TET2 (VAF: 36%) (c.3781C>A; p.R1261S) TET2 (VAF: 3%) (c.3866G>T; p.C1289F) TET2 (VAF: 18%) (c.4947T>A; p.Y1649X) RHOA (VAF: 20%) (c. 50G>T; p.G17V) Strong positive

n/a

DNMT3A (VAF:20%) (c.920C>T; p.P307L) TET2 (VAF: 4%) (c.3646C>T; p.R1216X) TET2 (VAF: 20%) (c.3781C>A, pR1261S) TET2 (VAF: 8%) (c.3866G>T; p.C1289F)

Weak positive

Strong positive

Weak positive

RHOA (VAF: 1%) (c.50G>T; p.G17V) Weak positive

n/a: not available; R-GCVD: rituximab, gemcitabine, cyclophosphamide, vincristine, prednisolone; CR: complete remission; VAF: variant allele frequency.

haematologica | 2022; 107(2)

495

R. Dobson et al.

(EBER) in situ hybridization was carried out in eight cases during the routine diagnostic workup, and four showed variable positivity from a few scattered to numerous EBER-positive cells, of which three displayed clonal IG gene rearrangement. In four cases, EBV-driven proliferation or classic Hodgkin lymphoma were considered in the initial diagnosis. Six follow-up biopsies were not diagnostic of involvement by AITL or PTCL-TFH by routine diagnostic investigations; these included two bone marrow and two skin specimens. In each specimen, a CD4+ T-cell infiltrate was noted, but a definite aberrant immunophenotype and expression of TFH markers could not be ascertained. Tcell clonality analysis showed weak polyclonality in two and a weak clonal or oligoclonal pattern in two.

Representative cases Case 30. A 79-year-old man presented with bilateral tender neck lymph nodes, and had mild sweats, but no weight loss or fever. Clinical examination revealed multiple bilaterally enlarged neck and groin lymph nodes (up to 1.5 cm diameter), and palpable liver and spleen. Right level II neck lymph node excision biopsy showed partial effacement of the lymph node architecture and expansion of the interfollicular area by a polymorphous population of lymphoid cells, including B cells with plasmacytoid differentiation, and scattered large cells (Figure 3). These B cells expressed pan B-cell markers (CD20, CD79a, CD19), and MUM1, but were negative for CD10 and BCL6 (data not shown). A high proportion of the B cells were EBERpositive and showed IG k light chain restriction. The lymphoid follicles appeared to be reactive and showed no apparent expansion of TFH cells, with only a few CD10positive cells spilling out of the germinal centers (Figure 3). BIOMED-2 clonality analyses showed clonal IGH and IGK gene rearrangements, but a weak oligoclonal pattern with TRG and TRB. The histological diagnosis was uncertain: a clonal EBV-positive polymorphous lymphoproliferation was considered. Close follow-up with a low threshold for re-biopsy was recommended. Two months later, the patient presented with increasing fatigue, fever, maculopapular chest rash and increased size of peripheral lymphadenopathy. Positron-emission tomography (PET) scan revealed extensive bilateral cervical, mediastinal, bilateral iliac and groin lymphadenopathy. Left level V neck lymph node excision biopsy showed partial effacement of the lymph node architecture by an infiltrate of medium-sized atypical lymphoid cells with regressed follicles (Figure 3). There was hyperplasia of follicular dendritic cell meshworks and high endothelial venules with the atypical lymphoid cells clustered in their vicinity. The atypical lymphoid cells were positive for CD3, CD5 and TFH markers (PD1, CD10, ICOS, BCL6) (Figure 3). EBER in situ hybridization revealed only scattered positive cells. BIOMED-2 clonality analyses showed clonal TRB and TRG gene rearrangements, and also weak clonal IGH and IGK gene rearrangements which were different from those of the previous biopsy in the size of their amplified IG products. A diagnosis of AITL was made. The patient was initially treated with six cycles of CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisone), then avelumab (cycle 12 at the most recent follow-up) under the AVAIL-T trial, and was well, showing no constitutional symptoms or palpable cervical lymph nodes, 20 months following the AITL diagnosis. 496

Retrospective analysis showed the presence of RHOA Gly17Val mutation in the above two specimens by qPCR and targeted sequencing with 3% and 23% VAF in the first and second biopsy, respectively. The targeted sequencing also revealed a pathogenic nonsense substitution in DNMT3A (c.2311C>T, p.R771*) in the first and second biopsy with 17% and 26% VAF, respectively. Case 7. An 82-year-old man, with remission of a previous mantle cell lymphoma, presented with a skin rash on the right calf. A punch biopsy showed a mild perivascular infiltrate by CD3+ T cells, with a histological diagnosis of panniculitis (Figure 4, Table 1). To investigate potential lymphoma relapse, lymph node core biopsies were taken at 11 and 12 months of follow-up: neither specimen showed evidence of mantle cell lymphoma, but both revealed cardinal features of AITL and also a prominent pleomorphic infiltrate including an EBER-positive B-cell population with Hodgkin and Reed/Sternberg (HRS)-like morphology and immunophenotype (Figure 4, Table 1). T-cell clonality analysis demonstrated clonal rearrangements by TRG-A and TRB-A with identical sized amplified products between the two biopsies, and an additional clonal rearrangement by TRB-C in the second biopsy. Further follow-up biopsies were taken at 25 and 26 months, and neither showed apparent evidence of AITL, but both had polymorphous infiltrates with a more prominent EBER-positive B-cell population that had HRScell morphology and immunophenotype (Figure 4, Table 1). Neither specimen showed any evidence of the clonal TRG/TRB rearrangements seen in the early lymph node biopsies, although a fifth biopsy displayed an isolated clonal rearrangement by TRB-C. B-cell clonality analyses demonstrated polyclonal IG gene rearrangements in both specimens (Figure 4, Table 1). A classic Hodgkin lymphoma arising from the EBV-positive B-cell component of the AITL was considered. In a retrospective study, the TRB-A and B PCR products from the second biopsy were sequenced using an Illumina MiSeq platform and a dominant TRBV5-J2 rearrangement (86%) was identified.2 Based on the unique VDJ junctional sequence, we designed unique BaseScope probes to identify the lymphoma T cells by in situ hybridization (Online Supplementary Figure S2). As expected, both the second and third biopsies with an AITL diagnosis showed diffuse positivity with the lymphoma clone-specific probe. Interestingly, the initial skin biopsy also displayed isolated positive cells, while the fifth biopsy with a diagnosis of EBV lymphoproliferative disease (LPD) gave a negative result. Both the second (AITL) and fifth (EBV-LPD) biopsies were investigated by panel sequencing for recurrent somatic mutations, and this identified five shared mutations – one DNMT3A, three TET2 and one RHOA changes – between the two specimens, and one further TET2 mutation only in the second specimen (Table 1). In general, the mutation load in each of the above shared changes was much higher in the second biopsy (AITL) than in the fifth biopsy (EBV-LPD). This was particularly striking for the RHOA Gly17Val mutation, with a VAF of 20% in the second biopsy but of only 1% in the fifth biopsy. As expected, qPCR for the RHOA mutation demonstrated its strong positivity in both the second and third biopsies showing AITL, but a weak positive signal in the initial skin biopsy, and the fourth and fifth biopsies displaying classic Hodgkin lymphoma-like EBV-LPD (Figure 4, Online Supplementary Figure S2). haematologica | 2022; 107(2)

RHOA mutation analysis in early lymphoma diagnosis

Discussion By using a highly sensitive qPCR assay, we confirmed that RHOA Gly17Val (c.50G>T) mutation is specifically associated with AITL or PTCL-TFH, but is not found in other lymphoproliferative conditions including those with florid infiltration of T-helper cells. More importantly, detection of RHOA Gly17Val (c.50G>T) mutation could help early detection of AITL and PTCL-TFH, and also diagnosis of their extranodal involvement. The difficulty in making a histological diagnosis of AITL or PTCL-TFH is well recognized, particularly when the neoplastic cell content is low, inconspicuous by histological and immunophenotypic assessment and undetectable by analysis of TCR gene rearrangements. Apart from the paucity of neoplastic T cells, the polymorphous infiltrate, particularly the presence of numerous EBV-positive B cells, including HRS-like cells (EBV-positive or -negative), may lead to diagnostic consideration of an EBV-driven B-cell proliferation.19-21 Such misleading diagnostic features may also be reinforced by the frequent demonstration of clonal IG gene rearrangement. As shown in this study, EBV-driven B-cell proliferation was considered as a diagnosis in four of 11 of the initial biopsies. Indeed, it has been reported by several independent studies that EBV-associated lymphoproliferation is a frequent pitfall in the diagnosis of AITL.19-22 In the present study, the majority of the initial specimens that were not diagnostic were core biopsies. It is possible that these core biopsies were not totally representative, with characteristic lymphoma components missed because of sampling errors. Nevertheless, two initial nondiagnostic biopsies were excisional lymph node specimens, indicating that an absence of diagnostic features of AITL/PTCL-TFH in the initial biopsies was also a real issue. Interestingly, the time interval between the initial non-diagnostic biopsies and the follow-up biopsies that established the diagnosis varied considerably, ranging from 0 to 26.5 months. In the cases with a long interval, it is likely that the initial biopsy represented an early premalignant lesion, while the follow-up biopsy reflected more progressed disease, thus having more cardinal features for making the AITL/PTCL-TFH diagnosis. While in the cases with a short interval, it is possible that the enlarged lymph nodes were variably involved by AITL/PTCL-TFH and the initial non-diagnostic biopsies represented early involvement by the lymphoma. Both the above possibilities may exist, without excluding the other. It is impossible to distinguish the two scenarios based on the analysis of a single biopsy, including RHOA mutation analysis. Nonetheless, detection of RHOA Gly17Val (c.50G>T) mutation can certainly raise the alarm to perform more in-depth histological and immunophenotypic investigations, for example a more careful search for evidence of TFH cell expansion, as shown in the first biopsy in case 30. It is important to emphasize that detection of a RHOA mutation is not equivalent to a diagnosis of lymphoma because mutational analysis by qPCR or targeted sequencing is highly sensitive, and the mutation is seen in biopsies without histological evidence of AITL. As discussed above, the RHOA mutation-positive non-diagnostic biopsy may represent a premalignant lesion or early involvement by the lymphoma. Therefore, RHOA mutation analysis should be used as an auxiliary tool with the results interpreted in the haematologica | 2022; 107(2)

context of histological and immunophenotypic findings. If a histological diagnosis of AITL/PTCL-TFH cannot be made, an excision biopsy or a low threshold for an early follow up biopsy, when appropriate, is indicated. In a few case studies, RHOA mutation was detected in circulating free DNA or peripheral blood lymphocytes from patients with AITL/PTCL-TFH,15 and the mutation burden appeared to correlate with the treatment outcome.23,24 It remains to be investigated whether the mutation is detectable in circulating free DNA or peripheral blood lymphocytes at the time of initial non-diagnostic biopsies. Given that the RHOA mutation burden in circulating free DNA and peripheral blood lymphocytes reflects, at least theoretically, the overall lymphoma load, simultaneously analyzing blood samples, in addition to tissue biopsy, could add further value in lymphoma diagnosis, particularly when a specimen is not representative. Apart from the initial biopsies, the diagnosis of extranodal involvement by AITL/PTCL-TFH such as skin and bone marrow is also difficult for reasons similar to those discussed above, including low tumor cell content and polymorphous infiltrate.25 In particular, extranodal infiltrates may not harbor the cardinal features of AITL such as TFH marker expression in the neoplastic T cells and follicular dendritic cell meshworks. The DNA sample from these extranodal sites is often not informative for clonality analysis because of insufficient lymphoid cells and/or poor DNA quality. As shown in our study, RHOA mutation analysis is highly valuable for confirming AITL/PTCL-TFH involvement in follow-up biopsies. Similarly, RHOA mutation analysis is valuable in the diagnosis of AITL/PTCL-TFH from cytological specimens.9 Aside from the RHOA mutation, there are several other genetic changes including IDH2, CD28, PLCG1, VAV1 and TNFRSF21 mutations, and VAV1-STAP2, CTLA4-CD28 and ITK-SYK fusions, which occur at variable frequencies in AITL and PTCL-TFH.2,26-28 Of note, mutations in VAV1, a signaling molecule downstream of TCR, occurs in 8.2% of AITL and appears to be mutually exclusive of the RHOA mutation.28 Detection of these additional lymphoma-associated genetic changes could also help the early detection of AITL/PTCL-TFH, together with RHOA mutation detection, potentially being valuable for diagnosis in up to 90% of these T-cell lymphomas. These diverse genetic changes could be readily investigated by targeted sequencing either alone or as part of a comprehensive lymphoma panel. The detailed analyses of multiple biopsies in case 7 also revealed the mutation burden in non-malignant T cells. The fifth lymph node biopsy showed little involvement by AITL as the lymphoma clone was undetectable by BaseScope in situ hybridization and the VAF of the RHOA mutation was only 1%, but the specimen had a high TET2 mutation burden (20% VAF). This implies that the TET2 mutation must have been present in a large proportion of reactive B and T cells,1,5,29-31 probably up to 40% of the total cell population, which is well above the EBER-positive cell fraction (~5%). Hence, this enlarged lymph node was essentially caused by lymphoid proliferation driven by the TET2 mutation and/or EBV infection. These findings further highlight the markedly variable histological presentation of enlarged lymph nodes in patients with AITL and, hence, the danger of potential sampling errors in the diagnosis of AITL. In this context, it is pertinent to 497

R. Dobson et al.

investigate RHOA and other lymphoma-associated mutations in patients with advanced age and biopsies showing EBV-positive polymorphous infiltrates, to explore how such mutation analyses can help to circumvent this diagnostic pitfall. At the genetic level, the high number of TET2 mutations and their frequent presence in non-neoplastic T cells suggest that the patient most likely had an underlying CHIP with TET2 and DNMT3A mutations that occurred in hematopoietic stem cells, consequently extending to the progenies of these cells. In general, patients with AITL typically have an aggressive clinical course and respond poorly to currently available therapies. Case 7 appeared to be an exception to this rule, showing slow disease progression in the absence of any treatment, although it is not possible to rule out a response to the R-GCVP (rituximab, gemcitabine, cyclophosphamide, vincristine, prednisolone) that was aimed at treating the EBV-associated proliferations (Table 1). An indolent clinical course has been previously reported for some patients with AITL showing pattern-1 histology, including those treated only with steroids.22,32 It remains to be investigated how to identify such indolent cases at the time of diagnosis and stratify their clinical management accordingly, particularly in view of the advance in early detection of AITL. In summary, we have shown that RHOA mutation is specifically associated with AITL, PTCL-TFH and their related lesions, and investigation of the mutation is highly valuable in early detection of these T-cell lymphomas and their extranodal involvement.

References 1. Nguyen TB, Sakata-Yanagimoto M, Asabe Y, et al. Identification of cell-type-specific mutations in nodal T-cell lymphomas. Blood Cancer J. 2017;7(1):e516. 2. Yao WQ, Wu F, Zhang W, et al. Angioimmunoblastic T-cell lymphoma contains multiple clonal T-cell populations derived from a common TET2 mutant progenitor cell. J Pathol. 2020;250(3):346-357. 3. Iqbal J, Amador C, McKeithan TW, Chan WC. Molecular and genomic landscape of peripheral T-cell lymphoma. Cancer Treat Res. 2019;176:31-68. 4. Xie M, Lu C, Wang J, et al. Age-related mutations associated with clonal hematopoietic expansion and malignancies. Nat Med. 2014;20(12):1472-1478. 5. Sakata-Yanagimoto M, Enami T, Yoshida K, et al. Somatic RHOA mutation in angioimmunoblastic T cell lymphoma. Nat Genet. 2014;46(2):171-175. 6. Nakamoto-Matsubara R, SakataYanagimoto M, Enami T, et al. Detection of the G17V RHOA mutation in angioimmunoblastic T-cell lymphoma and related lymphomas using quantitative allele-specific PCR. PLoS One. 2014;9(10):e109714. 7. Palomero T, Couronne L, Khiabanian H, et al. Recurrent mutations in epigenetic regulators, RHOA and FYN kinase in peripheral T cell lymphomas. Nat Genet. 2014; 46(2):166-170. 8. Ondrejka SL, Grzywacz B, Bodo J, et al. Angioimmunoblastic T-cell lymphomas with the RHOA p.Gly17Val mutation have classic clinical and pathologic features. Am J Surg Pathol. 2016;40(3):335-341. 9. Lee PH, Weng SW, Liu TT, et al. RHOA

498

Disclosures No conflicts of interest to disclose. Contributions RD, PD, WQ, and CC designed the experiments and collected and analyzed data; WQ and ZC performed the Illumina sequencing analysis and variant calling; LRB, HED, LF, ES, PW, JWG, MRJ, GAF, HR, AW, ADA, HL, MF, EJB, and GV contributed cases and pathology review; MQD and RD wrote and prepared the manuscript; MQD was responsible for research funding, study design and coordination. All authors commented on the manuscript and approved its submission for publication. Acknowledgments The authors thank Shubha Anand and Yuanxue Huang for their assistance with using TapeStation, Graeme Clark and Ezequiel Martin for their assistance with Illumina sequencing and Fangtian Wu for carrying out DNA extraction on a control case. Funding The research in MQD's laboratory was supported by grants from Blood Cancer UK (13 006, 15 019), CRUK (C8333/A29707), and the Kay Kendall Leukaemia Fund (KKL582) UK. WY was supported by a research fellowship from the China Scholarship Council, and an International Collaborative Award from the Pathological Society of Great Britain and Ireland, UK. The Human Research Tissue Bank is supported by the NIHR Cambridge Biomedical Research Centre.

G17V mutation in angioimmunoblastic Tcell lymphoma: a potential biomarker for cytological assessment. Exp Mol Pathol. 2019;110:104294. 10. Zang S, Li J, Yang H, et al. Mutations in 5methylcytosine oxidase TET2 and RhoA cooperatively disrupt T cell homeostasis. J Clin Invest. 2017;127(8):2998-3012. 11. Cortes JR, Ambesi-Impiombato A, Couronné L, et al. RHOA G17V induces T follicular helper cell specification and promotes lymphomagenesis. Cancer Cell. 2018;33(2):259-273. 12. Ng SY, Brown L, Stevenson K, et al. RhoA G17V is sufficient to induce autoimmunity and promotes T-cell lymphomagenesis in mice. Blood. 2018;132(9):935-947. 13. Dobay MP, Lemonnier F, Missiaglia E, et al. Integrative clinicopathological and molecular analyses of angioimmunoblastic T-cell lymphoma and other nodal lymphomas of follicular helper T-cell origin. Haematologica. 2017;102(4):e148-e151. 14. Yoo HY, Sung MK, Lee SH, et al. A recurrent inactivating mutation in RHOA GTPase in angioimmunoblastic T cell lymphoma. Nat Genet. 2014;46(4):371-375. 15. Hayashida M, Maekawa F, Chagi Y, et al. Combination of multicolor flow cytometry for circulating lymphoma cells and tests for the RHOA(G17V) and IDH2(R172) hotspot mutations in plasma cell-free DNA as liquid biopsy for the diagnosis of angioimmunoblastic T-cell lymphoma. Leuk Lymphoma. 2020;61(10):2389-2398. 16. Tanzima Nuhat S, Sakata-Yanagimoto M, Komori D, et al. Droplet digital polymerase chain reaction assay and peptide nucleic acid-locked nucleic acid clamp method for RHOA mutation detection in angioim-

munoblastic T-cell lymphoma. Cancer Sci. 2018;109(5):1682-1689. 17. 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. 18. Lewis NE, Petrova-Drus K, Huet S, et al. Clonal hematopoiesis in angioimmunoblastic T-cell lymphoma with divergent evolution to myeloid neoplasms. Blood Adv. 2020;4(10):2261-2271. 19. Steciuk MR, Massengill S, Banks PM. In immunocompromised patients, EpsteinBarr virus lymphadenitis can mimic angioimmunoblastic T-cell lymphoma morphologically, immunophenotypically, and genetically: a case report and review of the literature. Hum Pathol. 2012;43(1):127-133. 20. Nicolae A, Pittaluga S, Venkataraman G, et al. Peripheral T-cell lymphomas of follicular T-helper cell derivation with Hodgkin/ Reed-Sternberg cells of B-cell lineage: both EBV-positive and EBV-negative variants exist. Am J Surg Pathol. 2013;37(6):816-826. 21. Laforga JB, Gasent JM, Vaquero M. Potential misdiagnosis of angioimmunoblastic T-cell lymphoma with Hodgkin's lymphoma: a case report. Acta Cytol. 2010;54(5 Suppl):840-844. 22. Tan LH, Tan SY, Tang T, et al. Angioimmunoblastic T-cell lymphoma with hyperplastic germinal centres (pattern 1) shows superior survival to patterns 2 and 3: a meta-analysis of 56 cases. Histopathology. 2012;60(4):570-585. 23. Sakata-Yanagimoto M, NakamotoMatsubara R, Komori D, et al. Detection of the circulating tumor DNAs in angioimmunoblastic T- cell lymphoma. Ann Hematol. 2017;96(9):1471-1475.

haematologica | 2022; 107(2)

RHOA mutation analysis in early lymphoma diagnosis

24. Nguyen TB, Sakata-Yanagimoto M, Fujisawa M, et al. Dasatinib Is an effective treatment for angioimmunoblastic T-cell lymphoma. Cancer Res. 2020;80(9):18751884. 25. Attygalle AD, Diss TC, Munson P, Isaacson PG, Du MQ, Dogan A. CD10 expression in extranodal dissemination of angioimmunoblastic T-cell lymphoma. Am J Surg Pathol. 2004;28(1):54-61. 26. Vallois D, Dobay MP, Morin RD, et al. Activating mutations in genes related to TCR signaling in angioimmunoblastic and other follicular helper T-cell-derived lym-

haematologica | 2022; 107(2)

phomas. Blood. 2016;128(11):1490-1502. 27. Rohr J, Guo S, Huo J, et al. Recurrent activating mutations of CD28 in peripheral Tcell lymphomas. Leukemia. 2016; 30(5):1062-1070. 28. Fujisawa M, Sakata-Yanagimoto M, Nishizawa S, et al. Activation of RHOAVAV1 signaling in angioimmunoblastic Tcell lymphoma. Leukemia. 2018;32(3):694702. 29. Couronné L, Bastard C, Bernard OA. TET2 and DNMT3A mutations in human T-cell lymphoma. N Engl J Med. 2012;366(1):95-96. 30. Quivoron C, Couronné L, Della Valle V, et

al. TET2 inactivation results in pleiotropic hematopoietic abnormalities in mouse and is a recurrent event during human lymphomagenesis. Cancer Cell. 2011;20(1):2538. 31. Schwartz FH, Cai Q, Fellmann E, et al. TET2 mutations in B cells of patients affected by angioimmunoblastic T-cell lymphoma. J Pathol. 2017;242(2):129-133. 32. Ch'ang HJ, Su IJ, Chen CL, et al. Angioimmunoblastic lymphadenopathy with dysproteinemia--lack of a prognostic value of clear cell morphology. Oncology. 1997;54(3):193-198.

499

ARTICLE Ferrata Storti Foundation

Haematologica 2022 Volume 107(2):500-509

Non-Hodgkin Lymphoma

Efficacy and safety assessment of prolonged maintenance with subcutaneous rituximab in patients with relapsed or refractory indolent non-Hodgkin lymphoma: results of the phase III MabCute study Simon Rule,1 Wolney Gois Barreto,2 Javier Briones,3 Angelo M. Carella,4 Olivier Casasnovas,5 Chris Pocock,6 Clemens-Martin Wendtner,7 Francesco Zaja,8 Susan Robson,9 Lachlan MacGregor,9 Roger R. Tschopp,9 Sonja Nick9 and Martin Dreyling10 Derriford Hospital and Plymouth University Medical School, Plymouth, UK; 2Hemocentro Ribeirão Preto, University of São Paulo, São Paulo, Brazil; 3Hospital de la Santa Creu i Sant Pau, Barcelona, Spain; 4IRCCS AOU San Martino-IST, Genova, Italy; 5Centre Hospitalier Universitaire François Mitterand, Dijon, France; 6Kent & Canterbury Hospital, Canterbury, UK; 7Munich Clinic Schwabing, Academic Teaching Hospital, LudwigMaximilians University, Munich, Germany; 8DSM, University of Trieste, Trieste, Italy; 9 F. Hoffmann-La Roche Ltd, Basel, Switzerland and 10Klinikum der Universität München, Munich, Germany 1

ABSTRACT

Correspondence: MARTIN DREYLING martin.dreyling@med.uni-muenchen.de Received: November 10, 2020. Accepted: May 3, 2021. Pre-published: June 17, 2021. https://doi.org/10.3324/haematol.2020.274803

ituximab plus chemotherapy induction followed by rituximab maintenance for up to 2 years confers a long-term benefit in terms of progression-free survival in patients with indolent non-Hodgkin lymphoma. It is not known whether further prolonged maintenance with rituximab provides additional benefit. The phase III MabCute study enrolled 692 patients with relapsed or refractory indolent non-Hodgkin lymphoma. Patients who responded to induction with rituximab plus chemotherapy and were still responding after up to 2 years’ initial maintenance with subcutaneous rituximab were randomized to extended maintenance with subcutaneous rituximab (n=138) or observation only (n=138). The primary endpoint of investigator-assessed progression-free survival in the randomized population was un-addressed by the end of study because of an insufficient number of events (129 events were needed for 80% power at 5% significance if approximately 330 patients were randomized). In total, there were 46 progression-free survival events, 19 and 27 in the rituximab and observation arms, respectively (P=0.410 by stratified log-rank test; hazard ratio 0.76 [95% confidence interval: 0.37– 1.53]). The median progression-free survival was not reached in either randomized arm. There were no new safety signals; however, adverse events were seen slightly more frequently with rituximab than with observation during extended maintenance. Maintenance for up to 2 years with rituximab after response to initial induction therefore remains the standard of care in patients with relapsed or refractory indolent nonHodgkin lymphoma. (Clinicaltrials.gov identifier: NCT01461928)

500

Introduction Non-Hodgkin lymphoma (NHL) accounts for approximately 85% of lymphomas.1 Indolent forms include follicular lymphoma (FL), Waldenström macroglobulinemia/ lymphoplasmacytic lymphoma and marginal zone lymphoma. Of these, FL is the most common,1,2 accounting for 5/100,000 cases in Western Europe.3 Indolent NHL usually develops slowly (and may not need immediate treatment), follows a relapsing-remitting course, and is often incurable.1 Chemoimmunotherapy based on the human/murine chimeric anti-CD20 monoclonal antibody rituximab is standard treatment for a range of B-cell malignancies, including indolent and aggressive forms of NHL.3-7 Intravenously administered ritux-

haematologica | 2022; 107(2)

Prolonged maintenance with rituximab in indolent NHL

imab prolongs time to disease progression and increases overall survival (OS),8 but is associated with infusion reactions, which can be severe.9,10 Thus, a slow infusion is required during the first antibody administration, which generally takes at least 3.5–4 h.9-11 Faster infusion rates are used for subsequent infusions;10,11 nevertheless, infusion duration remains a challenge for patients and healthcare providers, particularly when multi-agent chemotherapy is being used.11 A subcutaneous (SC) formulation of rituximab and recombinant human hyaluronidase has been developed to address this concern.12 At fixed doses, rituximab SC has shown comparable efficacy and safety to intravenous rituximab in patients with NHL or chronic lymphocytic leukemia, with non-inferior serum trough rituximab concentrations.12-16 Additionally, patients’ preference/satisfaction and time and motion data (active healthcare practitioner time and chair time for patients) favor the use of the SC formulation,17,18 which is currently approved in Europe, the USA and numerous other countries for multiple indications (chronic lymphocytic leukemia, diffuse large B-cell lymphoma and FL).9,19 Dosing advantages over intravenous treatment include administration over 5–7 minutes, with a requirement for only 15 minutes of monitoring.9,19 Rituximab plus chemotherapy induction followed by rituximab maintenance is an approved treatment in FL,9,19 and has shown long-term progression-free survival (PFS) benefit in patients with indolent NHL.20-27 Tumor response and survival data show improvements in outcomes that persist over the longer term when rituximab maintenance therapy is given for up to 2 years.27,28 Whether further and prolonged maintenance therapy (beyond 2 years) would benefit patients with relapsed/refractory (R/R) indolent NHL who have maintained their response to treatment remains unknown. MabCute (NCT01469128) is a phase III trial in which patients with R/R indolent NHL were randomized to prolonged rituximab SC maintenance or observation after completing rituximab SC-based induction and 2 years maintenance therapy, provided that they were in response and willing to continue treatment.

Methods Study design This was a phase III, open-label, multicenter, international, randomized interventional study enrolling patients from 141 centers worldwide (mostly in Europe). MabCute was divided into Induction (6–8 months), Maintenance I (24 months) and Maintenance II (minimum 15 months) phases (Figure 1). The study was carried out in accordance with the Declaration of Helsinki and Good Clinical Practice, local legislation and the approval of institutional review boards. Written informed consent was obtained from participants.

Study population Adults aged ≥18 years with R/R CD20+ grade 1, 2 or 3a FL or other CD20+ indolent NHL (Waldenström macroglobulinemia/lymphoplasmacytic lymphoma or marginal zone lymphoma), and Eastern Cooperative Oncology Group (ECOG) performance status ≤2 were recruited. Details of the baseline assessments are provided in the Online Supplementary Appendix.

Study treatments Eligible patients received eight rituximab cycles, one intra-

haematologica | 2022; 107(2)

venous (375 mg/m2) and seven SC (1,400 mg fixed-dose) with six to eight chemotherapy cycles as induction (Figure 1; further information is available in the Online Supplementary Appendix). Patients with complete or partial response received 2 years’ maintenance with rituximab SC (Maintenance I). Patients with continuing response at the end of Maintenance I were randomized to prolonged maintenance with rituximab SC or to observation (Maintenance II).

Study endpoints and procedures The primary endpoint was PFS from the time of randomization to extended maintenance with rituximab SC or observation in Maintenance II (PFS in the randomized intent-to-treat [ITT ] population). Secondary endpoints included OS from the time of randomization in Maintenance II (OS ), overall response rate (Cheson criteria29) at end of Induction, and partial response to complete response conversion rate at the end of Maintenance I. An exploratory analysis of PFS and OS from enrollment to end of Maintenance I (i.e., the non-randomized part of the study; PFS , OS ) according to induction chemotherapy was also performed. Safety was assessed in all patients who received at least one dose of study medication and included adverse events (using National Cancer Institute Common Toxicity Criteria Version 4.0 and coded with Medical Dictionary for Regulatory Activities version 2.0), laboratory tests and vital signs. rand

rand

reg

Analytical plan Sample size was based on a phase III randomized study of 465 R/R FL patients. Overall, 129 PFS events were required to achieve 80% power for the log-rank-test at a two-sided significance level of 5%; therefore, approximately 700 patients needed to be enrolled to randomize 330 patients (allowing for a 10% dropout) after the 2.5-year Induction plus Maintenance I. Randomization to Maintenance II was 1:1, stratified by indolent NHL subtype and Follicular Lymphoma International Prognostic Index (FLIPI) category.30 The end of study was defined as the time when all patients randomized into Maintenance II had been followed up for ≥15 months, or earlier if at least 129 PFS events had been observed. PFS , OS , PFS and OS were reported with medians, 95% confidence intervals (95% CI), and Kaplan-Meier estimates and their 95% CI. The randomized treatment arms (prolonged rituximab maintenance vs. observation in Maintenance II) were compared using log-rank testing stratified according in indolent NHL subtype and FLIPI category. Cox regression was used to estimate hazard ratios (HR). rand

rand

reg

Results Patients In total, 692 patients received rituximab plus chemotherapy as induction (ITT population for Induction); 60.5% of patients received bendamustine, 12.4% received cyclophosphamide, doxorubicin, vincristine and prednisone (CHOP) and 11.8% received cyclophosphamide, vincristine and prednisone (CVP) (Online Supplementary Table S1); very small numbers received fludarabine, cyclophosphamide and mitoxantrone (FCM) or mitoxantrone, chlorambucil and prednisone (MCP). The distribution of patients who received each induction regimen was maintained out to Maintenance II (Online Supplementary Table S1). Of the patients who received induction therapy, 148 discontinued treatment because of adverse events (70 patients; 10.1%), disease progression (29; 4.2%), patients’ request (16; 2.3%), investigators’ request (7; 1.0%), loss to 501

S. Rule et al.

Figure 1. Study design. aWaldenström macroglobulinemia/lymphoplasmacytic lymphoma or marginal zone lymphoma. bChemotherapy options included bendamustine, CHOP, CVP, FCM, MCP, CHVP-IFN, chlorambucil, fludarabine-containing regimen or GIFOX. cMaintenance started within 8–12 weeks of completion of induction. R/R: relapsed or refractory; FL: follicular lymphoma; Gr: grade; NHL: non-Hodgkin lymphoma; R: rituximab; Cs: cycles; PD: disease progression; CR: complete response; PR: partial response; SD: stable disease; SC: subcutaneous; IV: intravenous; CHOP: cyclophosphamide, doxorubicin, vincristine and prednisone; CVP: cyclophosphamide, vincristine and prednisone; FCM: fludarabine, cyclophosphamide and mitoxantrone; MCP: mitoxantrone, chlorambucil and prednisone; CHVP-IFN: cyclophosphamide, doxorubicin, etoposide and prednisone + interferon-α; GIFOX: gemcitabine, ifosfamide and oxaliplatin.

follow-up (4; 0.6%), death (2; 0.3%) or other reasons (20; 2.9%) (Online Supplementary Figure S1A, B, Online Supplementary Table S2). A further 39 patients withdrew after Induction and before Maintenance I because of disease progression (16; 2.3%), adverse events (6; 0.9%), patients’ request (2; 0.3%), investigators’ request (2; 0.3%), death (1; 0.1%), loss to follow-up (1; 0.1%) or other reasons (11; 1.6%, all with stable disease). Of the 505 patients who continued to Maintenance I, 494 were treated (ITT population for Maintenance I; treatment was not given because of adverse events in 5 patients, disease progression in 2, investigator’s request or death in 1 each, and other reasons in 2). During Maintenance I, 188 patients (38.1%) discontinued study treatment because of disease progression (82; 16.6%), adverse events (66; 13.4%), patients’ request (20; 4.0%), investigators’ request (8; 1.6%), death (3; 0.6%), loss to follow-up (2; 0.4%) or other reasons (7; 1.4%: 2 with stable disease) (Online Supplementary Figure S1C). A further 28 patients (5.7%) completed Maintenance I but discontinued before Maintenance II. Reasons were patients’ request (12; 2.4%), disease progression (9; 1.8%), adverse events (4; 0.8%), investigator’s request (1; 0.2%) and other reasons (2; 0.4%). Two other patients (0.4%) failed to meet randomization criteria; the remaining 276 patients were randomized to Maintenance II (Figure 2). The median durations of the Induction and Maintenance I periods were 8.2 (range 0–18) months and 22.1 (range 0– 31) months, respectively. The primary ITT population included 276 patients who were randomized into Maintenance II (138 each in the rituximab and observation arms). Two further patients were initially planned for randomization but were subsequently found to be ineligible: one had disease progression and one had stable disease. Just over half of all patients were male and approximately two-thirds had Ann Arbor stage IV disease (Table 1). Patients were evenly distributed across FLIPI score categories in both arms. Approximately 40% of patients had bone marrow involvement, and just over half of all patients had FL. Nearly 60% overall had received rituximab plus bendamustine at Induction. Of the ITT popurand

502

lation for Induction (n=692), patients receiving bendamustine were older than those receiving CHOP or CVP, and a greater proportion had a high FLIPI score and Ann Arbor stage III/IV disease at screening (Online Supplementary Table S3). More patients receiving bendamustine and CHOP had FL compared with those receiving CVP. Six of 138 patients in the Maintenance II rituximab arm discontinued before the start of treatment (Figure 2). Maintenance II was completed thereafter by 109 patients randomized to rituximab and 111 randomized to observation; 23 patients (16.7%) randomized to rituximab and 27 (19.6%) randomized to observation discontinued during Maintenance II (Figure 2). The median follow-up time was 28.1 (range, 0–46) months. A single patient had progressive disease at the end of Maintenance I but was randomized to rituximab in error. This was subsequently recorded as a protocol violation.

Rituximab exposure The median duration of exposure to rituximab during Induction was 6.4 (range, 0–11) months. The median number of rituximab cycles was 8.0 (range, 1–9); 522 patients (75.4%) received the planned eight cycles. In Maintenance I, the median duration of exposure to rituximab was 20.3 (range, 0–28) months, with a median of 12.0 (range, 1–12) cycles being given. Of the 494 patients, 295 (59.7%) received the planned maximum 12 injections every 8 weeks for 24 months. The median duration of exposure during Maintenance II treatment was 24.8 (range, 0–43) months, and the median number of rituximab cycles was 14.0 (range, 1–24). Three patients received the highest number of rituximab treatments (44 cycles across the entire study), while two received the lowest (21 cycles).

Safety and tolerability Induction Treatment-emergent adverse events and serious adverse events were reported during Induction in 89.0% (616/692) and 30.1% (208/692) of patients, respectively. Half of all patients experienced at least one treatment-emergent haematologica | 2022; 107(2)

Prolonged maintenance with rituximab in indolent NHL

Figure 2. Patients’ disposition during Maintenance II. aTwo additional patients originally intended for randomization failed to meet continuation criteria and were consequently not treated in Maintenance II. bDerived by subtracting patients who discontinued from treated patients.

adverse event of grade ≥3 intensity (n=344; 49.7%), most commonly neutropenia (n=160; 23.1%). Febrile neutropenia (n=31; 4.5%), pneumonia (n=28; 4.0%) and neutropenia (n=16; 2.3%) were the most commonly reported serious adverse events (occurring in >2% of patients). Infusion/administration-related reactions were reported in 330 patients (47.7%) during Induction; 54 patients (7.8%) had a grade ≥3 event. The most common infusion/administration-related reaction of any grade during Induction was nausea (n=57; 8.2%); neutropenia was the most common grade ≥3 infusion/administration-related reaction (n=15; 2.2%). At least one treatment-emergent adverse event leading to rituximab discontinuation was reported in 66 patients (9.5%) during Induction, most frequently neutropenia (7 patients; 1.0%). At least one treatment-emergent adverse event leading to death was reported in 12 patients (1.7%). Similar incidences of treatment-emergent adverse events were seen across induction chemotherapy regimens (Online Supplementary Table S4). Patients receiving bendamustine experienced more general disorders and administration site conditions overall than those in other groups. Frequencies haematologica | 2022; 107(2)

of neutropenia reported as an adverse event were similar across induction chemotherapy regimens.

Maintenance I Treatment-emergent adverse events and serious adverse events were reported in 380/494 (76.9%) and 134/494 (27.1%) patients, respectively, during Maintenance I. At least one treatment-emergent adverse event of grade ≥3 intensity was reported in 163 patients (33.0%), most commonly neutropenia (n=59; 11.9%). Pneumonia was the most commonly reported serious adverse event affecting >2% of patients during Maintenance I (n=17; 3.4%). Infusion/administration-related reactions were reported in 75 patients (15.2%), with 20 (4.0%) experiencing at least one grade ≥3 event. The most common infusion/administration-related reaction of any grade was decreased neutrophil count (n=14; 2.8%). Neutropenia was the most commonly reported grade ≥3 infusion/administration-related reaction during Maintenance I (9 patients; 1.8%). Rituximab discontinuation due to a treatment-emergent adverse event was reported in 28 patients (5.7%) during Maintenance I. Of these, only neutropenia and pneumonia 503

S. Rule et al.

Table 1. Patient and disease characteristics at the start of Maintenance II.

Characteristic

Number of patients (%) R-SC Observation n=138 n=138

Median age, years (range) 64 (26-89) Male, n (%) 74 (53.6) Ann Arbor stage at diagnosis, n/N (%) I 13/134 (9.7) II 12/134 (9.0) III 21/134 (15.7) IV 88/134 (65.7) FLIPI score, n (%) Low 25 (34.2) Intermediate 22 (30.1) High 26 (35.6) Bone marrow involvement, n (%) 60 (43.5) Median lactate dehydrogenase, 3.26 (1.30-11.77) ukat/L (range) Type of NHL at screening, n (%) FL 73 (52.9) WM/LPL 28 (20.3) MZL 36 (26.1) Induction chemotherapy regimen Bendamustine 80 (58.0) CHOP 20 (14.5) CVP 26 (18.8) Other 12 (8.6)

65 (34-86) 68 (49.3) 8/135 (5.9) 19/135 (14.1) 30/135 (22.2) 78/135 (57.8) 28 (36.4) 27 (35.1) 22 (28.6) 59 (42.8) 3.32 (1.40-9.15)

77 (55.8) 25 (18.1) 35 (25.4) 79 (57.2) 19 (13.8) 22 (15.9) 18 (13.0)

R-SC: subcutaneous rituximab; FLIPI: Follicular Lymphoma International Prognostic Index; NHL: non-Hodgkin lymphoma; FL: follicular lymphoma; WM/LPL: Waldenström macroglobulinemia/lymphoplasmacytic lymphoma; MZL: marginal zone lymphoma; CHOP: cyclophosphamide, doxorubicin, vincristine and prednisone; CVP: cyclophosphamide, vincristine and prednisone.

Table 2. Summary of adverse events occurring during extended maintenance.

Patients with ≥1 event, n (%) ≥1 AE Grade ≥3 AE affecting ≥1% patients in either arm Neutropenia Pneumonia Hypertension Neutrophil count decreased Acute kidney injury Febrile neutropenia Leukopenia Myelodysplastic syndrome Upper respiratory tract infection Sepsis Thrombocytopenia Vomiting Serious AE affecting ≥1% patients in either arm Pneumonia Acute kidney injury Appendicitis Bronchitis Fall Febrile neutropenia Myelodysplastic syndrome Neutropenia Sepsis Squamous cell carcinoma of skin Grade 5 (fatal) AE AE leading to treatment discontinuation

R-SC n=138

Observation n=138

111 (80.4) 48 (34.8)

80 (58.0) 40 (29.0)

12 (8.7) 7 (5.1) 3 (2.2) 3 (2.2) 0 2 (1.4) 0 1 (0.7) 0 2 (1.4) 1 (0.7) 2 (1.4) 31 (22.5)

8 (5.8) 4 (2.9) 0 0 2 (1.4) 0 2 (1.4) 2 (1.4) 2 (1.4) 2 (1.4) 2 (1.4) 0 32 (23.2)

8 (5.8) 0 2 (1.4) 0 0 2 (1.4) 1 (0.7) 0 2 (1.4) 1 (0.7) 5 (3.6) 10 (7.2)

4 (2.9) 2 (1.4) 0 2 (1.4) 2 (1.4) 0 2 (1.4) 2 (1.4) 2 (1.4) 2 (1.4) 5 (3.6) 0

R-SC: subcutaneous rituximab; AE: adverse event.

were seen in more than one patient (2 patients each). At least one treatment-emergent adverse event leading to death was reported in eight patients (1.6%). Adverse events were the most common reason for death during Induction and Maintenance I (40/692; 5.8% and 32/494; 6.5%, respectively). Sepsis was the most frequent event leading to death during these phases (7 patients [1.0%] and 2 patients [0.4%], respectively). Rituximabrelated sepsis was associated with death in four patients (0.6%) during Induction and one patient (0.2%) during Maintenance I.

Maintenance II The original wording of the study protocol led to differences in adverse event reporting between the rituximab and observation arms in Maintenance II (see the Online Supplementary Appendix for details). After a protocol amendment to permit retrospective collection of adverse events of grade ≥3 during this phase (allowing adverse event reporting to be consistent between the rituximab and observation arms), neutropenia and pneumonia were the most frequently reported grade ≥3 adverse events in both the rituximab arm (8.7% and 5.1%, respectively) and the observation arm (5.8% and 2.9%, respectively) (Table 2). However, when looking at median neutrophil counts (based on laboratory data), similar values were observed in both treatment arms in Maintenance II. There were three grade ≥3 infusion/administration-related reactions (1 each of lymphopenia, urinary tract infection and hypertensive crisis). There were no reports of grade ≥3 rash, erythema or skin reaction during Maintenance II. The incidence of seri504

ous adverse events was similar for both arms (22.5% with rituximab and 23.2% for observation) (Table 2), with pneumonia (5.8% and 2.9%, respectively) and sepsis (1.4% for both arms) being most commonly reported. All fatal adverse events (5 in each arm) were considered unrelated to study treatment by the investigators. These events were pneumonia, septic shock, acute myocardial infarction, Crohn disease, abdominal infection and diverticulitis (same patient) in the rituximab arm, and acute myeloid leukemia, cardiopulmonary failure, ventricular tachycardia, pneumonia and lung disorder in the observation arm. Five further deaths in the rituximab arm and two in the observation arm were due to disease progression; a single additional death with unknown cause was recorded in the observation arm. There were no safety concerns or new signals related to hematology, biochemistry or immunological parameters in any phase of the study, and no meaningful changes from baseline in vital signs. There were also no unexpected changes from baseline in worst-on-treatment ECOG scores, and no noteworthy differences in score shifts between the rituximab and observation arms in Maintenance II.

Efficacy The overall response rate at the end of Induction was 84.7% (95% CI: 81.1–87.3), and was similar across different chemotherapies: 86.4% (95% CI: 82.7–89.5) for bendamustine; 87.2% (95% CI: 78.3–93.4) for CHOP; 84.1% (95% CI: 74.4–91.3) for CVP; and 76.9% (95% CI: 67.6– haematologica | 2022; 107(2)

Prolonged maintenance with rituximab in indolent NHL

Figure 3. Survival outcomes during the randomized Maintenance II period. Kaplan-Meier analysis of progression-free (A) and overall survival (B) during the randomized Maintenance II period. R-SC: subcutaneous rituximab.

84.6) for other regimens (including FCM and MCP). All but one patient per arm among the 276 who were randomized in Maintenance II were responders after Induction (Online Supplementary Table S5). Proportions of patients in complete response or partial response at the end of Maintenance I were also comparable between arms among the 276 patients who were randomized (Online Supplementary Table S5). Of the 357 patients who achieved a partial response at the end of Induction, 77 achieved a complete response by the end of Maintenance I, providing a conversion rate of 21.6% (95% CI: 17.4–26.2). The MabCute study was unable to address its primary endpoint (investigator-assessed PFS ) because the number of events reported was insufficient: 129 PFS events were needed for 80% power at 5% significance, with approximately 700 patients needed initially to yield the 330 required for randomization. There were 46 PFS events at the end of study: 19 and 27 in the rituximab and observation arms, respectively: P=0.410 by log-rank test stratified by FLIPI risk category and NHL subtype; HR 0.76 (95% CI: 0.37–1.53), estimated using a Cox regression model with rand

rand

haematologica | 2022; 107(2)

FLIPI risk category and NHL subtype as stratification factors. PFS rates at 6, 9, 12, 15 and 18 months (Kaplan-Meier estimates) were similar for both arms (between 0.97 at 6 months and 0.88 at 18 months for rituximab, and between 0.96 at 6 months and 0.87 at 18 months for observation). The median PFS was not reached in either arm (Figure 3A). One patient, randomized to observation, discontinued from the study and subsequently died 2 months later. This event was not taken into consideration in the primary analysis due to a recording issue. It had no effect on the overall results or conclusions of the study. The median PFS (from enrollment to end of Maintenance I) (Figure 4A) was 46.32 months (95% CI: 42.87–60.02) in patients receiving bendamustine, 39.62 months (95% CI: 27.86–not reached) in patients receiving CHOP, and 37.03 months (95% CI: 33.87–74.12) in patients receiving CVP. Three-year PFS estimates for patients receiving bendamustine, CHOP, and CVP were 0.63 (95% CI: 0.57–0.69), 0.58 (0.46–0.68), and 0.59 (0.28– 0.80), respectively. The median OS (from enrollment to end of Maintenance I) (Figure 4B) was not reached in patients receiving bendamustine (95% CI: 66.86–not rand

rand

reg

505

S. Rule et al.

Figure 4. Survival outcomes from enrollment to end of Maintenance I, according to induction chemotherapy. Kaplan-Meier analysis of progression-free (A) and overall survival (B) from enrollment to end of Maintenance I, according to induction chemotherapy received (bendamustine vs. CHOP and CVP). aIntent to treat population for Induction. Time to event calculated from first induction therapy up to the earliest date of event until randomization; data censored after randomization. CHOP: cyclophosphamide, doxorubicin, vincristine and prednisone; CVP: cyclophosphamide, vincristine and prednisone.

reached) or CVP (95% CI: 46.82–not reached). The median OS was 58.84 months (95% CI: 42.22-not reached) in patients receiving CHOP induction. Three-year OS estimates for patients receiving bendamustine, CHOP, and CVP were 0.83 (95% CI: 0.78–0.87), 0.70 (95% CI: 0.53– 0.82), and 0.82 (95% CI: 0.59–0.93), respectively. PFS and OS by NHL subtype are available in Online Supplementary Table S6. Unfortunately, due to the low numbers no conclusion can be drawn from these data. Response rates at the end of Induction by both chemotherapy regimen and patients remaining at the start of each subsequent study phase showed that 56.9% of patients (157/276) who responded and who ultimately entered Maintenance II had received bendamustine as reg

induction therapy (Online Supplementary Table S7). There were 18 deaths (OS events) in total, ten in the rituximab arm and eight in the observation arm (not including the patient with a retrospective record of death). The median OS was not reached (Figure 3B). rand

rand

reg

506

Discussion The benefit of 2 years of maintenance therapy with rituximab after response to frontline induction in patients with indolent NHL is well established in terms of significantly improved PFS. Early trials of rituximab maintenance in patients with R/R indolent NHL indicated efficahaematologica | 2022; 107(2)

Prolonged maintenance with rituximab in indolent NHL

cy in this setting too.23,24,31,32 The findings prompt the question of whether further and prolonged maintenance therapy (beyond 2 years) would benefit patients with R/R indolent NHL who have maintained their response to treatment. The overall tumor response rate after induction (~85%) in MabCute was consistent with rates observed in previous studies in R/R indolent NHL (75-95%).22-24,31,32 These trials showed significant improvements in response duration and median PFS when rituximab maintenance therapy was given for up to 2 years compared with observation alone, and are supported by a meta-analysis of 2,586 patients participating in nine randomized trials which showed a significant improvement of median OS with rituximab maintenance therapy versus observation only in patients with R/R FL (HR 0.72, 95% CI: 0.57–0.91).28 Maintenance with rituximab for 2 years following the end of Induction in the current study was associated with a rate of partial response to complete response conversion similar to that observed in previous studies.21 Although an OS benefit has been observed following rituximab maintenance in the R/R setting, it has not been demonstrated in the frontline setting. A 10-year followup of the PRIMA study in 1,018 patients with high tumor burden, previously untreated FL showed a significant long-term PFS benefit of rituximab maintenance over observation for 2 years after response to induction with rituximab and chemotherapy.27 Although there was no significant OS benefit, the authors noted that over half of patients in the rituximab arm had not had disease progression over the 10 years, and had not required new antilymphoma treatment. Similar findings (significant PFS improvement but no significant effect on OS) were reported by the ECOG-ACRIN group after a median 11.5 years of follow-up of 387 patients who attained at least stable disease after CVP induction.33 In addition, a prior study by the German STIL group confirmed the benefit of rituximab maintenance in R/R indolent NHL after a bendamustine or fludarabine salvage therapy.34 The key benefits of SC rituximab, with its short administration time, are linked to reductions in healthcare resource utilization18,35 and patients’ preference15,36 relative to the intravenous formulation, particularly for long-term therapy. Unfortunately, MabCute was unable to address its primary endpoint of investigator-assessed PFS in the randomized population. This was due to a much lower than anticipated number of PFS events, representing only a third of the required events to have a power of 80% with a hazard ratio of 0.605. The reason for the low rate of PFS events was not clear, but may have been related to the effectiveness of supportive care and treatment delivery under the study protocol. There were 18 deaths in total, ten in the rituximab arm and eight in the observation arm (not including the patient with a retrospective record of death). This study was not powered to evaluate survival, and follow-up was relatively short at the time of the analysis. The exploratory analysis of PFS and OS from enrollment to end of Maintenance I (i.e. the non-randomized part of the study) showed 3-year PFS and OS rates of 63% and 83%, respectively in patients treated with bendamustine, 58% and 70%, respectively, in those treated with CHOP, and 59% and 82%, respectively, in those treated with CVP. It should be noted that there was a bias rand

rand

reg

haematologica | 2022; 107(2)

reg

in patients’ selection; the investigator could decide what regimens to give to which patients – most patients were treated with bendamustine in the Induction period, and the size of the subgroups is very different. Therefore, a direct comparison between treatment regimens is not appropriate. In MabCute, approximately 60% of patients received bendamustine at Induction, and this proportion of patients was maintained out to the Maintenance II phase. There are few data available on the use of bendamustine in R/R NHL. A study by Sakai et al. recently reported 3-year PFS and OS rates of 71% and 89%, respectively, in a population of patients with R/R FL,37 while the STIL group reported a 1-year PFS of 76% and median OS of 109.7 months in patients with R/R indolent NHL or mantle cell lymphoma.34 However, comparison between these trials is difficult; the PFS survival data from the current study were censored after Maintenance I, and are therefore not comparable with general PFS data. No unexpected toxicities were reported during Maintenance II, and good tolerability and safety were maintained throughout follow-up. The proportion of patients who experienced adverse events during longterm maintenance was slightly greater in the rituximab arm than in the observation arm. These observations were as expected, given the known profile of rituximab SC.12 Rituximab is always given by intravenous infusion for the first cycle, when the risk of infusion-related reactions is greatest, to allow slowing or stopping of the infusion (as a preventative measure). The incidence of infusion-related reactions decreases with subsequent infusions. The overall safety profile of rituximab SC is similar to that of the intravenous formulation, but with a greater incidence of mostly mild-to-moderate infusion/administration-related reactions, primarily injection-site reactions, which decrease in frequency over time.14,15,17,38 This pattern was observed in the current study (i.e. from 47.7% of patients during Induction to 15.2% during Maintenance I and 10.1% of rituximab patients in Maintenance II). Interestingly, in line with prior publications on frontline therapy of FL, a bendamustine-based induction resulted in more frequent pyrexia and neutropenia (Online Supplementary Table S5). In conclusion, the MabCute study was unable to address the question of whether prolonged (beyond 2 years) maintenance therapy with rituximab adds any clear benefit compared with observation only in patients with R/R indolent NHL (who have responded to induction therapy with rituximab plus chemotherapy), due to a low number of PFS events. Extension of treatment was not associated with any important additional toxicity (in particular no additional neutropenia or infection), and no new safety signals were observed. Two years of maintenance with rituximab after response to initial induction therapy therefore remains the standard of care in these patients. reg

Disclosures SR declares a consultation or advisory role for Janssen, AstraZeneca, F. Hoffmann-La Roche Ltd, Sunesis, Pharmacyclics, Celgene, Celltrion, Kite; speakers bureau for Janssen; and research funding from Janssen. WGB declares no conflict of interest. JB declares honoraria from F. Hoffmann-La Roche Ltd, Takeda, Celgene, Novartis, and Gilead; consultation or advisory role for Takeda, Janssen, Celgene, and Gilead; research funding from F. Hoffmann-La Roche Ltd; and travel and/or accommodation expenses from F. Hoffmann La-Roche 507

S. Rule et al.

Ltd, Takeda, Celgene, Janssen, and Gilead. AMC does not declare any conflict of interest. OC declares honoraria from F. Hoffmann-La Roche Ltd, Takeda, BMS, Merck, Gilead, and Janssen; consultation or advisory role for F. Hoffmann-La Roche Ltd, Takeda, BMS, Merck, Gilead, and Janssen; research funding from F. Hoffmann-La Roche Ltd, Takeda, Gilead, and AbbVie; and travel and/or accommodation expenses from F. Hoffmann-La Roche Ltd, Takeda, and Janssen. CP declares honoraria from Janssen and Gilead; consultancy or advisory role for Takeda and Celgene and travel and/or accommodation expenses from Gilead. C-MW declares honoraria from F. Hoffmann-La Roche Ltd, Janssen-Cilag, Gilead, and AbbVie; consultation or advisory role for F. Hoffmann-La Roche Ltd, Janssen-Cilag, Gilead, and AbbVie; research funding from F. Hoffmann-La Roche Ltd, Janssen-Cilag, Gilead, and AbbVie; and travel and/or accommodation expenses from F. Hoffmann-La Roche Ltd, Janssen-Cilag, Gilead, and AbbVie. FZ declares honoraria from F. Hoffmann-La Roche Ltd, Janssen-Cilag, Gilead, Celgene, AbbVie, Takeda, and Novartis; consultation or advisory role for Sandoz, F. Hoffmann-La Roche Ltd, JanssenCilag, Gilead, Celgene, AbbVie, Takeda, and Novartis; research funding from Celgene and Novartis; and travel and/or accommodation expenses from F. Hoffmann-La Roche Ltd, Celgene, AbbVie, Takeda, and Novartis. SR is employed by F. HoffmannLa Roche Ltd. LMacG, RRT, and SN are employed by F. Hoffmann-La Roche Ltd. MD declares honoraria from Bayer, Celgene, Gilead, Janssen, and F. Hoffmann-La Roche Ltd; consultation or advisory role for Acerta, Bayer, Celgene, Gilead, Janssen, Novartis, F. Hoffmann La Roche Ltd, and Sandoz; and

References 1. Ninkovic S, Lambert J. Non-Hodgkin lymphoma. Medicine. 2017;45(5):297-304. 2. Shankland KR, Armitage JO, Hancock BW. Non-Hodgkin lymphoma. Lancet. 2012;380 (9844):848-857. 3. Dreyling M, Ghielmini M, Rule S, et al. Newly diagnosed and relapsed follicular lymphoma: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2021;32(3):298-308. 4. Kastritis E, Leblond V, Dimopoulos MA, et al. Waldenström’s macroglobulinaemia: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up†. Ann Oncol. 2018;29(Suppl 4):iv41-iv50. 5. 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):v116-125. 6. Zelenetz AD, Gordon LI, Abramson JS, et al. NCCN guidelines insights: B-cell lymphomas, version 3.2019. J Natl Compr Canc Netw. 2019;17(6):650-661. 7. Dreyling M, Campo E, Hermine O, et al. Newly diagnosed and relapsed mantle cell lymphoma: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2017;28(Suppl 4):iv62iv71. 8. Salles G, Barrett M, Foa R, et al. Rituximab in B-cell hematologic malignancies: a review of 20 years of clinical experience. Adv Ther. 2017;34(10):2232-2273. 9. MabThera Summary of Product Characteristics. 2020. (Accessed 27 April, 2021, at https://www.ema.europa.eu/en/docu-

508

research funding from Celgene, Janssen, Mundipharma, and F. Hoffmann-La Roche Ltd. Contributions SR, WGB, JB, AMC, OC, CP, C-MW, FZ and MD were involved in accrual and treatment of patients. SR analyzed data. All authors were involved in interpreting the data, critically reviewing the manuscript, approved the manuscript for submission and agree to be accountable for the accuracy and integrity of the study. Funding MabCute was sponsored by F. Hoffmann-La Roche Ltd. Third-party medical writing assistance, under the direction of Simon Rule and Martin Dreyling, was provided by Christopher Dunn and Scott Malkin of Ashfield MedComms, an Ashfield Health company, and was funded by F. Hoffmann-La Roche Ltd. Data-sharing statement Qualified researchers may request access to individual patient level data through the clinical study data request platform (https://vivli.org/). Further details on Roche's criteria for eligible studies are available here (https://vivli.org/members/ourmembers/). For further details on Roche's Global Policy on the Sharing of Clinical Information and how to request access to related clinical study documents, see here (https://www.roche.com/research_and_development/who_we_a re_how_we_work/clinical_trials/our_commitment_to_data_shar ing.htm

ments/product-information/mabthera-eparproduct-information_en.pdf). 10. Highlights of prescribing information. Rituxan® (rituximab) injection, for intravenous use. South San Francisco, CA: Biogen Idec, Inc. and Genentech USA, Inc.; 2019 September. 11. Yelvington BJ. Subcutaneous rituximab in follicular lymphoma, chronic lymphocytic leukemia, and diffuse large B-cell lymphoma. J Adv Pract Oncol. 2018;9(5):530534. 12. Davies A, Berge C, Boehnke A, et al. Subcutaneous rituximab for the treatment of B-cell hematologic malignancies: a review of the scientific rationale and clinical development. Adv Ther. 2017;34(10):2210-2231. 13. Davies A, Merli F, Mihaljevic B, et al. Pharmacokinetics and safety of subcutaneous rituximab in follicular lymphoma (SABRINA): stage 1 analysis of a randomised phase 3 study. Lancet Oncol. 2014;15(3):343352. 14. Davies A, Merli F, Mihaljevic B, et al. Efficacy and safety of subcutaneous rituximab versus intravenous rituximab for first-line treatment of follicular lymphoma (SABRINA): a randomised, open-label, phase 3 trial. Lancet Haematol. 2017;4(6): e272-e282. 15. Lugtenburg P, Avivi I, Berenschot H, et al. Efficacy and safety of subcutaneous and intravenous rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone in first-line diffuse large B-cell lymphoma: the randomized MabEase study. Haematologica. 2017;102(11):1913-1922. 16. Assouline S, Buccheri V, Delmer A, et al. Pharmacokinetics, safety, and efficacy of subcutaneous versus intravenous rituximab plus chemotherapy as treatment for chronic

lymphocytic leukaemia (SAWYER): a phase 1b, open-label, randomised controlled noninferiority trial. Lancet Haematol. 2016;3(3): e128-138. 17. Rummel M, Kim TM, Aversa F, et al. Preference for subcutaneous or intravenous administration of rituximab among patients with untreated CD20+ diffuse large B-cell lymphoma or follicular lymphoma: results from a prospective, randomized, open-label, crossover study (PrefMab). Ann Oncol. 2017;28(4):836-842. 18. De Cock E, Kritikou P, Sandoval M, et al. Time savings with rituximab subcutaneous injection versus rituximab intravenous infusion: a time and motion study in eight countries. PLoS One. 2016;11(6):e0157957. 19. Highlights of prescribing information: Rituxan Hycela™ (rituximab and hyaluronidase human) injection, for subcutaneous use. San Francisco, CA: Genentech, Inc. 2017. 20. Schneider T, Rosta A, Losonczy H, et al. Efficacy and tolerability of a 2-year rituximab maintenance therapy in patients with advanced follicular lymphoma after induction of response with rituximab-containing first line-regimens (HUSOM Study). Pathol Oncol Res. 2018;24(2):199-205. 21. Salles G, Seymour JF, Offner F, et al. Rituximab maintenance for 2 years in patients with high tumour burden follicular lymphoma responding to rituximab plus chemotherapy (PRIMA): a phase 3, randomised controlled trial. Lancet. 2011;377 (9759):42-51. 22. Forstpointner R, Dreyling M, Repp R, et al. The addition of rituximab to a combination of fludarabine, cyclophosphamide, mitoxantrone (FCM) significantly increases the response rate and prolongs survival as com-

haematologica | 2022; 107(2)

Prolonged maintenance with rituximab in indolent NHL

pared with FCM alone in patients with relapsed and refractory follicular and mantle cell lymphomas: results of a prospective randomized study of the German Low-Grade Lymphoma Study Group. Blood. 2004;104 (10):3064-3071. 23. van Oers MH, Klasa R, Marcus RE, et al. Rituximab maintenance improves clinical outcome of relapsed/resistant follicular nonHodgkin lymphoma in patients both with and without rituximab during induction: results of a prospective randomized phase 3 intergroup trial. Blood. 2006;108(10):32953301. 24. van Oers MH, Van Glabbeke M, Giurgea L, et al. Rituximab maintenance treatment of relapsed/resistant follicular non-Hodgkin's lymphoma: long-term outcome of the EORTC 20981 phase III randomized intergroup study. J Clin Oncol. 2010;28(17):28532858. 25. Oh SY, Kim WS, Kim JS, et al. Phase II study of R-CVP followed by rituximab maintenance therapy for patients with advanced marginal zone lymphoma: consortium for improving survival of lymphoma (CISL) study. Cancer Commun (Lond) 2019;39 (1):58. 26. Ferrario A, Pulsoni A, Olivero B, et al. Fludarabine, cyclophosphamide, and rituximab in patients with advanced, untreated, indolent B-cell nonfollicular lymphomas: phase 2 study of the Italian Lymphoma Foundation. Cancer. 2012;118(16):39543961. 27. Bachy E, Seymour JF, Feugier P, et al.

haematologica | 2022; 107(2)

Sustained progression-free survival benefit of rituximab maintenance in patients with follicular lymphoma: long-term results of the PRIMA study. J Clin Oncol. 2019;37(31): 2815-2824. 28. Vidal L, Gafter-Gvili A, Salles G, et al. Rituximab maintenance for the treatment of patients with follicular lymphoma: an updated systematic review and meta-analysis of randomized trials. J Natl Cancer Inst. 2011;103(23):1799-1806. 29. Cheson BD, Pfistner B, Juweid ME, et al. Revised response criteria for malignant lymphoma. J Clin Oncol. 2007;25(5):579-586. 30. Solal-Celigny P, Roy P, Colombat P, et al. Follicular lymphoma international prognostic index. Blood. 2004;104(5):1258-1265. 31. Forstpointner R, Unterhalt M, Dreyling M, et al. Maintenance therapy with rituximab leads to a significant prolongation of response duration after salvage therapy with a combination of rituximab, fludarabine, cyclophosphamide, and mitoxantrone (RFCM) in patients with recurring and refractory follicular and mantle cell lymphomas: Results of a prospective randomized study of the German Low Grade Lymphoma Study Group (GLSG). Blood. 2006;108(13): 4003-4008. 32. Martinelli G, Schmitz SF, Utiger U, et al. Long-term follow-up of patients with follicular lymphoma receiving single-agent rituximab at two different schedules in trial SAKK 35/98. J Clin Oncol. 2010;28(29): 4480-4484. 33. Barta SK, Li H, Hochster HS, et al.

Randomized phase 3 study in low-grade lymphoma comparing maintenance antiCD20 antibody with observation after induction therapy: a trial of the ECOGACRIN Cancer Research Group (E1496). Cancer. 2016;122(19):2996-3004. 34. Rummel M, Kaiser U, Balser C, et al. Bendamustine plus rituximab versus fludarabine plus rituximab for patients with relapsed indolent and mantle-cell lymphomas: a multicentre, randomised, openlabel, non-inferiority phase 3 trial. Lancet Oncol. 2016;17(1):57-66. 35. Rule S, Collins GP, Samanta K. Subcutaneous vs intravenous rituximab in patients with non-Hodgkin lymphoma: a time and motion study in the United Kingdom. J Med Econ. 2014;17(7):459-468. 36. Rule S, Briones J, Smith R, et al. Preference for rituximab subcutaneous (SC) and intravenous (IV) among patients with CD20+ non-Hodgkin's lymphoma (NHL) completing the RASQ measure In randomized phase III studies Prefmab and Mabcute. Value Health. 2014;17(7):A537. 37. Sakai R, Ohmachi K, Sano F, et al. Bendamustine-120 plus rituximab therapy for relapsed or refractory follicular lymphoma: a multicenter phase II study. Ann Hematol. 2019;98(9):2131-2138. 38. Salar A, Avivi I, Bittner B, et al. Comparison of subcutaneous versus intravenous administration of rituximab as maintenance treatment for follicular lymphoma: results from a two-stage, phase IB study. J Clin Oncol. 2014;32(17):1782-1791.

509

ARTICLE Ferrata Storti Foundation

Myeloid Biology

Unique ethnic features of DDX41 mutations in patients with idiopathic cytopenia of undetermined significance, myelodysplastic syndrome, or acute myeloid leukemia Eun-Ji Choi,1* Young-Uk Cho,2* Eun-Hye Hur,1 Seongsoo Jang,2 Nayoung Kim,3 Han-Seung Park,1 Jung-Hee Lee,1 Kyoo-Hyung Lee,1 Si-Hwan Kim,2 Sang-Hyun Hwang,2 Eul-Ju Seo,2 Chan-Jeoung Park2 and Je-Hwan Lee1

Haematologica 2022 Volume 107(2):510-518

1 Department of Hematology, Asan Medical Center, University of Ulsan College of Medicine; 2Department of Laboratory Medicine, Asan Medical Center, University of Ulsan College of Medicine and 3Asan Institution for Life Sciences and Department of Convergence Medicine, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea

*E-JG and Y-UC contributed equally as co-first authors.

ABSTRACT

Correspondence: JE-HWAN LEE jhlee3@amc.seoul.kr Received: August 27, 2020. Accepted: February 2, 2021. Pre-published: February 25, 2021. https://doi.org/10.3324/haematol.2020.270553

510

DX41 mutations are associated with hematologic malignancies including myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML), but the incidence in idiopathic cytopenia of undetermined significance (ICUS) is unknown. We investigated the incidence, genetic characteristics, and clinical features of DDX41 mutations in Korean patients with ICUS, MDS, or AML. We performed targeted deep sequencing of 61 genes including DDX41 in 457 patients with ICUS (n=75), MDS (n=210), or AML (n=172). Germline DDX41 mutations with causality were identified in 28 (6.1%) patients, of whom 27 (96.4%) had somatic mutations in the other position of DDX41. Germline origins of the DDX41 mutations were confirmed in all of the 11 patients in whom germline-based testing was performed. Of the germline DDX41 mutations, p.V152G (n=10) was most common, followed by p.Y259C (n=8), p.A500fs (n=6), and p.E7* (n=3). Compared with non-mutated patients, patients with a DDX41 mutation were more frequently male, older, had a normal karyotype, low leukocyte count, and hypocellular marrow at diagnosis. Three of the four ICUS patients with germline DDX41 mutations progressed to MDS. The incidence of DDX41 mutations in Korean patients was high and there was a distinct mutation pattern, in that p.V152G was a unique germline variant. ICUS harboring germline DDX41 mutations may be regarded as a hereditary myeloid neoplasm. Germline DDX41 mutations are not uncommon and should be explored when treating patients with myeloid malignancies.

Introduction Inherited hematologic malignancies have been established in well-defined hereditary syndromes, which exhibit a particular phenotype, often present in childhood, or have a strong family history.1 There is also an increasing awareness of additional autosomal dominant genetic aberrations with predisposition to myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML), which are primarily sporadic diseases and typically present in older adults. The recently updated World Health Organization (WHO) classification defined myeloid neoplasms with germline predisposition2 and categorized familial myeloid neoplasms into three groups: those with an absence of pre-existing disorder or organ dysfunction (e.g., CEBPA or DDX41 mutations), those with a pre-existing platelet disorder (e.g., RUNX1 mutations), and those with dysfunction of other organs (e.g., GATA2 mutations). DDX41 mutations have recently joined the growing list of genetic alterations in familial myeloid malignancies.3,4 MDS or AML with germline DDX41 mutations usually occurs in the sixth decade of life or beyond, whereas

haematologica | 2022; 107(2)

Ethnic features of DDX41 mutations in ICUS/MDS/AML

most other cases of cancers with germline predisposing mutations typically develop in adolescence or early adulthood.5-11 The DDX41 gene is located at 5q35.3 and encodes a DEAD-box RNA helicase, which is involved in pre-mRNA splicing, RNA processing, and ribosome biogenesis.12 Several mechanisms have been proposed to explain the contributions of DDX41 mutations to the development of hematologic malignancies. DDX41 mutations can: (i) cause aberrant mRNA splicing leading to exon retention or exon skipping, (ii) disrupt the STING-interferon pathway; and (iii) induce aberrant pre-rRNA trimming and ribosome biogenesis.3 DDX41 mutations include both germline and somatic mutations, with the latter being found in over half of the patients with germline mutations in the other allele of DDX41.5 In recent studies, germline DDX41 variants were found in 2.4% of 1,385 patients with MDS or AML,11 and germline or somatic DDX41 variants were found in 3.4% of 1,002 patients with myeloid neoplasms.10 Following advances in genetic testing, clinical next-generation sequencing (NGS)-based leukemia panels are being increasingly used to identify somatic mutations to facilitate the diagnosis, improve prognostication, and select optimal treatment strategies in patients with hematologic malignancies. Some variants found in the panels can also be germline mutations in genes associated with hereditary hematopoietic malignancies.13-15 The myeloid leukemia panel used at our institute includes the DDX41 gene, and the frequencies of DDX41 mutations in Korean patients with MDS or AML appeared to be higher than the previously reported incidences. Importantly, DDX41 mutations have not been evaluated in patients with idiopathic cytopenia of undetermined significance (ICUS), which is a known precursor lesion of MDS. In this study, we investigated the incidence, genetic characteristics, and clinical features of DDX41 mutations in Korean patients with ICUS, MDS, or AML.

Methods Patients We included patients with ICUS, MDS, or AML whose bone marrow samples were collected between 2009 and 2019 at Asan Medical Center (Seoul, Korea). Patients with ICUS or lower-risk MDS were either prospectively enrolled (since January 2018) or retrospectively analyzed, while those with higher-risk MDS or AML were retrospectively analyzed. All patients in the study cohort were unrelated individuals, not including an index case and his or her family members. Diagnoses of MDS and AML were based on the WHO 2016 Classification.2 ICUS was defined by the proposed criteria of the 2007 Consensus Group:16 cytopenia in one or more of cell lineages for ≥6 months (hemoglobin <11 g/dL, neutrophils <1.5x109/L, and platelets <100x109/L) while excluding other causes of cytopenia such as a history of pelvic irradiation or cytotoxic chemotherapy, splenomegaly, heart failure or liver cirrhosis with portal hypertension, active viral infections, and a history of blood or bone marrow diseases. Clonal cytopenia of undetermined significance was defined as ICUS with myeloid neoplasm-related somatic mutations with a variant allele frequency ≥ 2%, or clonal karyotypic abnormalities. Myeloid neoplasm-related somatic mutations were based on those specified in the updated National Comprehensive Cancer Network guideline for MDS.17

haematologica | 2022; 107(2)

The Institutional Review Board of Asan Medical Center approved the protocols of this study (2018-0042 and 2018-0048 [for prospective and retrospective analysis of patients with lowerrisk MDS or ICUS], 2019-0794 [for sequencing the DDX41 gene in DDX41-mutated patients and their family members], and 20200131 [for retrospective analysis of data from patients with higherrisk MDS or AML]), which was carried out in accordance with the 2008 Declaration of Helsinki.

Mutational and cytogenetic analysis For rhe NGS assay, we prepared the sequencing libraries from genomic DNA using customized probes (Integrated DNA Technologies, Inc., Coralville, IA, USA) to capture and enrich the entire coding regions of 61 target genes (HEMEaccuTest DNA Target Enrichment kit; NGeneBio, Seoul, Korea) (Online Supplementary Table S1). We carried out sequencing on the MiSeqDx (Illumina, San Diego, CA, USA) with 2×150 bp, pairedend reads according to the manufacturer’s instructions. Initial read mapping was carried out against the human reference genome (hg19/GRCh37). We subsequently analyzed the sequencing data for variant calling using commercial software (CLC Genomics Workbench; QIAGEN Bioinformatics, Redwood City, CA, USA). We retained the potentially pathogenic variants by filtering out common polymorphisms (minor allele frequency in the population ≥1%) and sequencing/mapping errors, and by filtering in the known oncogenic variants based on the available population or cancer mutation-specific databases. We set the minimum cutoff of variant allele frequency at 2.0% for reporting. We performed the cytogenetic analysis using conventional G-banding techniques based on the analysis of 20 or more metaphase cells.

Germline variant confirmation and determination of causality Variants with allele frequencies between 40% and 60% were considered to be probable germline mutations. We performed germline-based testing in 11 of 34 patients with probable germline DDX41 mutations using sorted blood T cells. This strategy of using sorted T cells was based on recent work confirming that T cells yield sufficient DNA and high rates of somatic variant calls in MDS. It was suggested that, given the challenge of obtaining skin biopsies, T cells would be preferential germline tissues for MDS genomic studies.18 Peripheral blood mononuclear cells were harvested by standard Ficoll (GE Healthcare, Sweden) density gradient centrifugation, and T cells were isolated using the Pan T Cell Isolation Kit, human (MACS Miltenyi Biotec, Auburn, CA, USA) according to the manufacturer’s instructions. The isolated T cells were analyzed with CD3-FITC using a FACSCalibur (Becton Dickinson, Franklin Lakes, NJ, USA), and genomic DNA was purified by the QIAamp DNA mini kit (Qiagen, QIAGEN GmbH, Germany). The pathogenicity of probable germline DDX41 mutation was determined according to the guideline from the American College of Medical Genetics and Genomics (ACMG).19 The concurrence of a somatic DDX41 mutation was considered as strong evidence for causality. Thus, we classified germline DDX41 variants as “causal” if they were either pathogenic (or likely pathogenic) by the ACMG guideline or accompanied by a somatic DDX41 mutation regardless of the ACMG interpretation.

Statistical analysis Categorical variables were compared using the χ2 test or Fisher exact test, and continuous variables were compared using the Mann-Whitney U-test or the Student t-test, as appropriate. Survival was calculated by the Kaplan-Meier method and the resulting survival curves were compared using the log-rank test

511

E.-J. Choi et al.

Table 1. Patients’ characteristics at diagnosis.

Characteristic Sex, n(%) Male Female Median age (range), years ICUS, n(%) CCUS* Non-CCUS WHO classification MDS, n(%) MDS with SLD/RS-SLD MDS with MLD/RS-MLD MDS with EB-1 MDS with EB-2 MDS, unclassifiable MDS with isolated del(5q) Unknown AML, n(%) AML with RGA AML with MRC Therapy-related AML, NOS MPAL Risk stratification, n(%) MDS IPSS-R score ≤ 3.5 IPSS-R score > 3.5 AML# Favorable Intermediate Adverse Unknown Karyotype, n(%) Normal Abnormal

Total (n = 457)

ICUS (n = 75)

MDS (n = 210)

AML (n = 172)

272 (59.5) 185 (40.5) 59 (16-89)

37 (49.3) 38 (50.7) 54 (19-89)

134 (63.8) 76 (36.2) 61 (18-87)

101 (58.7) 71 (41.3) 57 (16-81)

42 (56.0) 33 (44.0)

41 (19.5) 73 (34.8) 45 (21.4) 18 (8.6) 26 (12.4) 2 (1.0) 2 (1.0) 84 (48.8) 40 (23.3) 7 (4.1) 40 (23.3) 1 (0.6)

75 (35.7) 135 (64.3) 63 (36.6) 35 (20.3) 73 (42.4) 1 (0.6) 230 (50.3) 227 (49.7)

69 (92.0) 6 (8.0)

92 (43.8) 118 (56.2)

69 (40.1) 103 (59.9)

ICUS: idiopathic cytopenia of undetermined significance; CCUS: clonal cytopenia of undetermined significance; WHO: World Health Organization; MDS: myelodysplastic syndrome; SLD: single lineage dysplasia; RS: ring sideroblasts; MLD: multilineage dysplasia; EB: excess blasts; AML: acute myeloid leukemia; RGA: recurrent genetic abnormalities; MRC: myelodysplasia-related changes; NOS, not otherwise specified; MPAL, mixed phenotype acute leukemia; IPSS-R, International Prognostic Scoring System-Revised. *CCUS was defined as ICUS with myeloid neoplasm-related somatic mutations of variant allele frequency ≥2%, or clonal karyotypic abnormalities. #Risk stratification of AML according to the 2017 European LeukemiaNet risk stratification.

(univariate analysis). The Kaplan-Meier survival curves were rendered as a graph using Prism version 5.0 (GraphPad Software, Inc., La Jolla, CA, USA). In all analyses, the P-values were two-tailed and those <0.05 were considered statistically significant.

Results Patients’ characteristics The clinical characteristics of the 457 included patients at diagnosis are shown in Table 1. There were 75 patients (16%) with ICUS, 210 (46%) with MDS and 172 (38% with AML. The median age at diagnosis was 59 years (range, 16-89), and 60% were men. Forty-two (56.0%) of the ICUS patients had clonal cytopenia of undetermined significance. Disease risk of the MDS patients was lower512

risk in 75 (35.7%) and higher-risk in 135 (64.3%) according to the Revised International Prognostic Scoring System (IPSS-R).20 Of the AML patients, 63 (36.6%), 35 (20.3%), and 73 (42.4%) were classified into favorable, intermediate, and adverse genetic risk categories, respectively, according to the 2017 European LeukemiaNet risk stratification.21

Frequency and genetic characteristics of DDX41 mutations We detected genetic DDX41 mutations in 39 (8.5%) patients. Thirty-four (7.4%) patients had germline mutations, of whom 27 (79.4%) also had somatic mutations at the other position of DDX41. Five (1.1%) patients had somatic DDX41 mutations only. In 28 patients, the germline DDX41 mutations were considered causal and haematologica | 2022; 107(2)

Ethnic features of DDX41 mutations in ICUS/MDS/AML

Figure 1. Frequency of DDX41 mutations according to the type of hematologic malignancy. ICUS: idiopathic cytopenia of undetermined significance; MDS: myelodysplastic syndrome; AML: acute myeloid leukemia.

only these patients were included in further analyses. The frequency of the causal germline DDX41 mutations was 6.1% (28 of 457); 6.7% (5 of 75) in ICUS, 9.0% (19 of 210) in MDS, and 2.3% (4 of 172) in AML (Figure 1). Detailed information on the DDX41 variants, concurrent mutations of other genes, and karyotypes in the 28 patients are provided in Online Supplementary Table S2. Germline origins of the DDX41 mutations were confirmed in all of the 11 patients who underwent germline-based testing (p.V152G in 5, p.Y259C in 3, p.A500fs in 2, and p.L328R in 1). Of the 55 DDX41 mutations detected in this study, 28 were germline and the other 27 appeared to be somatic. All of the somatic mutations were missense, whereas germline mutations were missense in 19 (67.9%) cases, frameshift in six (21.4%), and nonsense in three (10.7%). The majority of somatic mutations were located in the helicase C or C-terminal domain (n=18, 66.7%), whereas the majority of germline mutations were in the helicase ATP-binding or N-terminal domain (n=22, 78.6%; P=0.001) (Figure 2A). Of the germline DDX41 mutations, p.V152G (n=10, 35.7%) was the most common, followed by p.Y259C (n=8, 28.6%), p.A500fs (n=6, 21.4%), p.E7* (n=3, 10.7%), and p.L328R (n=1, 3.6%). Two germline variants (p.A500fs and p.E7*) were classified as pathogenic according to the ACMG guideline. The other germline variants (p.V152G, p.Y259C, and p.L328R) were classified as being of uncertain significance, but were considered causal when accompanied by somatic DDX41 mutations (Online Supplementary Table S3). Notably, four mutations (p.V152G, p.Y259C, p.A500fs and p.E7*) were found at a significantly higher frequency in the study patients than in healthy Koreans, as shown by high odds ratios (38.5, 17.3, 49.6 and 26.5, respectively) (Online Supplementary Table S4). Two germline mutations (p.V152G and p.Y259C) were only detected in ICUS/MDS (75.0%) and not in AML (0%), whereas p.A500fs and p.E7* were detected in both ICUS/MDS and AML groups (Online Supplementary Table S5). Of the 27 somatic DDX41 mutations, p.R525H (n=14, 51.9%) was the most common, followed by p.T227M (n=5, 18.5%), and the remaining eight somatic DDX41 mutations were detected in one (3.7%) patient each. The somatic p.R525H variant was haematologica | 2022; 107(2)

less frequently associated with the germline p.V152G variant (3 of 10) than with p.Y259C (6 of 8) or p.A500fs (4 of 6), whereas the somatic p.T227M variant tended to be more frequently associated with p.V152G (4 of 10) than with p.Y259C (1 of 8) or p.A500fs (0 of 6) (Online Supplementary Table S6). Twenty-two (78.6%) of the 28 patients with mutations in DDX41 had concurrent mutations in other genes. Genes mutated in over 10% of the patients were PHF6 and ASXL1 (5 patients [17.9%] each), followed by CBL and NF1 (4 patients [14.3%] each), and DNMT3A and TP53 (3 patients [10.7%] each) (Figure 2B; Online Supplementary Table S2). We observed six variants of the PHF6 gene in five patients with DDX41 germline mutations: p.M1T and p.R116* in one patient, and p.G248D, p.C20F, p.M1T and p.M1V in one patient each. Interestingly, PHF6 p.M1T/V variants were detected in only three patients harboring DDX41 germline mutations among the whole study population of 457 patients.

Clinical features and outcomes of the patients with DDX41 mutations There was a male predominance among the DDX41-mutated patients (96.4% vs. 57.1%; P<0.001), and the patients with this mutation tended to be older (median 66 vs. 57 years; P<0.001), and were more likely to have a normal karyotype (75.0% vs. 48.7%; P=0.007), lower white blood cell count (median 1.8 vs. 3.7×109/L; P=0.047), and lower marrow cellularity (median 30% vs. 60%; P<0.001) at diagnosis compared with the non-mutated patients (Table 2). Among patients with MDS, the DDX41 mutations were significantly more frequent in the MDS subtypes with excess blasts (EB)-1 and EB-2, compared to other categories with bone marrow blasts <5%, although the mutation frequencies were not significantly different between patients with lower risk or higher risk according to the IPSS-R (Table 2). Of 23 MDS or AML patients with causal germline DDX41 mutations, data regarding blood counts before diagnosis were available for 16 patients, and all 16 patients had a history of cytopenia at least 1 year prior to diagnosis. During the median follow-up of 25.5 months, 116 513

E.-J. Choi et al.

Table 2. Comparison of clinical features according to the presence of germline DDX41 mutations.

Sex, n(%) Male Female Median age (range), years Chromosome, n(%) Normal Abnormal WBC, × 109/ L, median (range) Hb, g/dL, median (range) Platelets, × 109/ L, median (range) BM cellularity, %, median (range) BM blasts, %, median (range) N. of mutated genes, median (range) MDS, n(%) MDS with SLD/MLD/del(5q)/U MDS with EB-1/EB-2 Unknown Risk stratification, n(%) MDS IPSS-R ≤ 3.5 (%) IPSS-R > 3.5 (%) AML Favorable Intermediate Adverse Unknown

DDX41 mutations (+) (n=28)

DDX41 mutations (-) (n=429)

27 (96.4) 1 (3.6) 66 (41-79)

245 (57.1) 184 (42.9) 57 (16-89)

< 0.001a

21 (75.0) 7 (25.0) 1.8 (1.0-3.3) 10.1 (5.2-13.2) 90 (13-174) 30 (5-60) 6.2 (0.8-65.2) 3 (2-6)

209 (48.7) 220 (51.3) 3.7 (0.7-313.1) 9.1 (2.3-16.4) 68 (3-638) 60 (3-100) 5.2 (0-98.8) 2 (0-12)

7 (4.8) 12 (19.0) 0

138 (95.2) 51 (81.0) 2 (100)

< 0.001c 0.007a 0.047c 0.113c 0.689c < 0.001c 0.016c 0.036c 0.001aa*

0.693a 6 (31.6) 13 (68.4)

69 (36.1) 122 (63.9) 0.215b

0 (0) 2 (50.0) 2 (50.0) 0 (0)

63 (37.5) 33 (19.6) 71 (42.3) 1 (0.6)

WBC: white blood cells, Hb: hemoglobin; BM, bone marrow; MDS: myelodysplastic syndrome; SLD: single lineage dysplasia; MLD: multilineage dysplasia; EB: excess blasts; IPSSR, International Prognostic Scoring System-Revised; AML: acute myeloid leukemia. aby the χ2 test; bby the Fisher exact test; cby t-test; *SLD/MLD/del(5q)/U vs. EB-1/EB-2.

patients (7 ICUS, 55 MDS, and 54 AML) died. The 5-year overall survival rate was 60.8% in the overall population and 84.6%, 62.2%, and 38.9% in patients with ICUS, MDS, and AML, respectively. There was no significant correlation between overall survival and the presence of DDX41 mutations in each disease category of ICUS, MDS, and AML (Figure 3) as well as in the total study population (Online Supplementary Figure S1). Online Supplementary Table S7 shows the clinical course of each patient with a DDX41 mutation. Clinical courses could be followed up in four of the five ICUS patients with probable germline DDX41 mutations, and notably, three of these four patients showed disease progression to MDS EB-1 (n=2; 77.9 and 17.6 months after ICUS diagnosis) or MDS EB-2 with a gain of PTPN11 mutation (n=1; 9 months after ICUS diagnosis) during the follow-up. Another ICUS patient with a germline DDX41 mutation had a son with Hodgkin lymphoma.

Discussion In our cohort of 457 patients with ICUS, MDS, or AML, 6.1% of the patients carried causal germline DDX41 mutations, which is a higher incidence than those found in previous studies which ranged between 0.8% and 3.9% in patients with myeloid malignancies (mostly MDS and 514

AML).5,6,10,11,22 In a study comparing the clinical and genetic characteristics of DDX41 mutations in AML and MDS patients between two ethnically distinct populations, germline DDX41 mutations were found in 3.9% of a Japanese cohort and in 0.8% of a Caucasian cohort.22 Therefore, there seems to be an ethnic difference in the incidence of DDX41 mutations in patients with myeloid neoplasms between Asian and Western patients. In contrast, the clinical features of our DDX41 -mutated patients, such as male predominance, old age at presentation,5,6,10,11,23 hypocellular marrow,3,4,6 leukopenia,6 and frequent normal cytogenetics3-6,11 were similar to those reported in other ethnic populations. The DDX41 mutations did not show significant associations with survival outcomes. There are several noteworthy findings in our study regarding the genetic characteristics of DDX41 mutations. First, the germline mutations were mostly N-terminal variants (78.6%), whereas somatic mutations were mostly C-terminal variants (66.7%). This finding is consistent with the observations in two recent studies.10,11 The N-terminal region of DDX41 has the helicase ATP binding domain,24,25 and the structural rearrangement in the N-terminal region may change the conformation of the ATP-binding site and eventually decrease ATP-binding ability.24 In contrast, the helicase C-terminal domain is involved in ATP hydrolysis.24,25 Therefore, genetic alterhaematologica | 2022; 107(2)

Ethnic features of DDX41 mutations in ICUS/MDS/AML

Figure 2. Distribution of DDX41 mutations and concurrent mutations in other genes. (A) Distribution of DDX41 mutations detected in the current study and two previous studies (Quesada et al.10 and Sebert et al.11). This figure shows the differences in positional distribution (N-terminal skewed vs. C-terminal skewed) and mutational effects (variable vs. missense-dominated) between germline and somatic mutations. The protein structure of DDX41 was based on the RefSeq accession number of NM_016222.3 and the UniProtKB entry of Q9UJV9: the 622 amino acid long protein comprises the helicase ATP-binding domain (position 212-396), the helicase C-terminal domain (position 407-567), and a zinc finger domain (position 580-597). Different colors indicate different effects of mutations: light blue, missense mutation; light green, inframe indel; purple, nonsense mutation; brown, splicing mutation; red, frameshift mutation; black, start codon loss. Different shapes represent the three studies: square, Sebert et al.11 diamond, Quesada et al.10 circle, current study. (B) Concurrent mutations of other genes identified in bone marrow samples from DDX41-mutated patients. The types of genetic alterations and diseases are presented in the legend.

haematologica | 2022; 107(2)

515

E.-J. Choi et al.

ations in the N-terminal region may cause greater susceptibility to protein hypofunction than those in the C-terminal region. Second, the patterns of germline DDX41 mutations in our Korean population were distinct from those in Western populations10,11 or even other Asian populations.22,23 The germline DDX41 mutations (p.V152G, p.Y259C, p.A500fs, p.E7*, and p.L328R) in our study are totally different from those reported in Western populations (p.M1I, p.D140fs, p.G173R, and Q41*). Korean and Japanese patients shared three major germline DDX41 variants (p.Y259C, p.A500fs, p.E7*),22 but p.V152G was only found in Koreans and not in Japanese or other ethnic populations. Third, we observed the exclusive presence of PHF6 p.M1T/V variants in three patients with probable germline DDX41 mutations. These variants potentially cause a complete lack of protein production as a consequence of start codon loss and are causative germline mutations of the Börjeson-Forssman-Lehmann syndrome.26-28 Thus, our findings suggest that the same genetic mutation can induce both hereditary diseases and sporadic cancer, as exemplified by mutations in ETV6.29 The possible association between PHF6 p.M1T/V variants and germline DDX41 mutations should be investigated further. Germline mutations that predispose an individual to MDS or AML may also contribute to the development of ICUS, but the genetic predisposition to ICUS has not been systematically investigated. In a recent study of germline DDX41 mutations in adult patients with MDS or AML, 45.5% of patients with pathogenic germline DDX41 mutations had a previous history of cytopenia before the diagnosis of MDS or AML, and the preexisting cytopenia might indicate the presence of ICUS in these patients.11 We also observed similar findings. Furthermore, five (6.7%) of 75 ICUS patients had causal germline DDX41 mutations, three of whom progressed to MDS. Our study shows that germline DDX41 mutations are not uncommon in ICUS patients. Our findings do not indicate that the germline DDX41 mutations contribute to the progression of ICUS to MDS, but instead do suggest that ICUS patients harboring such variants may be considered as having a hereditary myeloid neoplasm. Our observations highlight the potential oncogenic role of germline DDX41 mutations in the pathogenesis of ICUS/MDS in comparison with AML. First, the patients with ICUS/MDS carried germline DDX41 mutations more frequently than did AML patients and germline missense mutations were highly enriched in ICUS/MDS rather than in AML. These findings might support the notion that less disruptive variants are associated with a milder phenotype in the disease spectrum. Second, only one (3.6%) of the 28 patients with germline DDX41 mutations carried a mutation in the splicing factor gene. This finding is in line with previous observations that splicing factor gene mutations were largely mutually exclusive with DDX41 mutations.5,10 DDX41 interacts with core splicing proteins such as SF3B, U2 complex, PRPF8 scaffold protein, and U5 complex, indicating that spliceosomal proteins are the top functional group associated with DDX41.5,25 Genetic alterations of the splicing components affect the 3’-splice site recognition during pre-mRNA processing and are involved in the pathogenesis of myelodysplasia.30 This indicates that mutant DDX41 can have an oncogenic role in MDS via aberrant mRNA splicing with the assumption that mutations in these splicing factors have an impact on the 516

pathways of downstream oncogenes and tumor suppressor genes.3 Donor-derived leukemia has been reported in several families with germline DDX41 mutations; in all such cases, donors had the same type of germline DDX41 variants as the respective recipients.31,32 In our study, DDX41 mutations (germline p.E7* and somatic p.G228C) were found in a 60-year-old man with high-risk MDS (#12). No

Figure 3. Overall survival of patients with different hematologic disorders according to DDX41 mutation status. (A-C) Overall survival of patients with idiopathic cytopenia of undetermined significance (A), myelodysplastic syndrome (B) or acute myeloid leukemia (C) according to whether they had DDX41 mutations (red) or not (blue).

haematologica | 2022; 107(2)

Ethnic features of DDX41 mutations in ICUS/MDS/AML

HLA-matched sibling or unrelated donor was available, and his two adult offspring had the same DDX41 mutation (p.E7*). Fortunately, the patient’s HLA-haploidentical brother did not carry the DDX41 mutation, and the patient could undergo haploidentical hematopoietic stem cell transplantation from him. Considering that the risk of malignancy in DDX41 carriers is yet to be determined, we calculated the odds ratios of major germline DDX41 variants detected in our study (Online Supplementary Table S4). Nevertheless, an extensive population-based study is needed to obtain more reliable data that may be useful in establishing genetic counseling guidelines for germline DDX41 variants, which are currently available only for donor selection in allogeneic hematopoietic stem cell transplantation. NGS-based targeted genotyping for somatic mutations can identify patients who are at risk of hereditary hematopoietic malignancies. In a recent study, of 25 pathogenic or likely pathogenic variants with variant allele frequency >40% in 24 patients with germline tissues available, six variants (24%) were of germline origin – three DDX41 variants, two GATA2 variants, and one TP53 variant; DDX41 had a 100% diagnostic yield for pathogenic germline variants in that study.33 In another study, targeted NGS showed that 17 patients had putative germline DDX41 variants with a variant allele frequency >40%, all of which were of germline origin.11 We were also able to confirm germline origin in all of the 11 patients with probable germline DDX41 mutations. In cases in which germline samples are not available, NGSbased leukemia panels seem to predict germline DDX41 variants with high probability. However, it is worth mentioning that NGS-based panels may fail to detect deletions or gene rearrangements that are responsible for the predisposition syndrome. Our study has some limitations. The number of patients included in the study was relatively small, and this might have had an impact on the analysis for clinical associations of DDX41 mutations with clinical outcomes. Family history was not systematically collected in this study, although such information is helpful in pinpointing the pathogenicity of sequence variants. We did not perform functional studies to demonstrate that sequence variants detected in this study had a deleterious effect in vivo. Experimental data can be useful to support pathogenicity, particularly for missense variants of uncertain significance. We acknowledge that these limitations may hamper the precise variant classification based on the ACMG guideline. However, the concurrence of germline and somatic DDX41 mutations was a recurrent finding across recent

References 1. Rafei H, DiNardo CD. Hereditary myeloid malignancies. Best Pract Res Clin Haematol. 2019;32(2):163-176. 2. 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. 3. Cheah JJC, Hahn CN, Hiwase DK, Scott HS, Brown AL. Myeloid neoplasms with germline DDX41 mutation. Int J Hematol. 2017;106(2):163-174. 4. Maciejewski JP, Padgett RA, Brown AL,

haematologica | 2022; 107(2)

studies.5,6,10,11 This also has provided another illustration that germline alterations predispose to the acquisition of somatic mutations in the same genes which act as a second hit being associated with cancer development as demonstrated by JAK2, CEBPA, and RUNX1 mutations.34,35 Therefore, in cases harboring a germline DDX41 mutation, the acquisition of a somatic DDX41 mutation should be considered as strong evidence for causality. Lastly, our data may not reflect the whole Korean population, although study patients included in this study come from all across the Korean peninsula. In conclusion, our results delineate the unique ethnic features of DDX41 mutations in Korean patients, such as higher incidence and different patterns, compared with patients from Western countries or other Asian countries. Specifically, the most common germline mutation in our cohort was p.V152G, which was not found in previous studies in other ethnicities. Our results suggest that ICUS harboring germline DDX41 mutations may be regarded as a hereditary myeloid neoplasm. Germline DDX41 mutations may be predicted with a high probability by using clinical NGS-based leukemia panels based on variant allele frequency levels and public databases. Germline DDX41 mutations are not uncommon and should be explored when treating patients with myeloid malignancies. Disclosures No conflicts of interest to disclose. Contributions E-JC and Y-UC analyzed and interpreted the data; E-JC, YUC and J-HL contributed to the manuscript; E-HH performed experiments. All authors provided patients’ data, reviewed, and approved the final manuscript. Acknowledgments We thank Dr. Joon Seo Lim from the Scientific Publications Team at Asan Medical Center for his editorial assistance in preparing this manuscript. Funding This research was supported by a Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (NRF-2017R1E1A1A01074383). The biospecimens and data used in this study were provided by Asan Bio-Resource Center, Korea Biobank Network (2018-08). Data-sharing statement For original data, please contact imeunjeee@gmail.com.

Muller-Tidow C. DDX41-related myeloid neoplasia. Semin Hematol. 2017;54(2):9497. 5. Polprasert C, Schulze I, Sekeres MA, et al. Inherited and somatic defects in DDX41 in myeloid neoplasms. Cancer Cell. 2015; 27(5):658-670. 6. Lewinsohn M, Brown AL, Weinel LM, et al. Novel germ line DDX41 mutations define families with a lower age of MDS/AML onset and lymphoid malignancies. Blood. 2016;127(8):1017-1023. 7. Cardoso SR, Ryan G, Walne AJ, et al. Germline heterozygous DDX41 variants in a subset of familial myelodysplasia and

acute myeloid leukemia. Leukemia. 2016; 30(10):2083. 8. Li R, Sobreira N, Witmer PD, Pratz KW, Braunstein EM. Two novel germline DDX41 mutations in a family with inherited myelodysplasia/acute myeloid leukemia. Haematologica. 2016;101(6): e228. 9. Vairo FPE, Ferrer A, Cathcart-Rake E, et al. Novel germline missense DDX41 variant in a patient with an adult-onset myeloid neoplasm with excess blasts without dysplasia. Leuk Lymphoma. 2019;60(5):1337-1339. 10. Quesada AE, Routbort MJ, DiNardo CD, et al. DDX41 mutations in myeloid neoplasms are associated with male gender, TP53

517

E.-J. Choi et al. mutations and high-risk disease. Am J Hematol. 2019;94(7):757-766. 11. Sébert M, Passet M, Raimbault A, et al. Germline DDX41 mutations define a significant entity within adult MDS/AML patients. Blood. 2019;134(17):1441-1444. 12. Linder P. Dead-box proteins: a family affair-active and passive players in RNP-remodeling. Nucleic Acids Res. 2006;34(15):41684180. 13. DiNardo CD, Routbort MJ, Bannon SA, et al. Improving the detection of patients with inherited predispositions to hematologic malignancies using next-generation sequencing-based leukemia prognostication panels. Cancer. 2018;124(13):2704-2713. 14. Drazer MW, Kadri S, Sukhanova M, et al. Prognostic tumor sequencing panels frequently identify germ line variants associated with hereditary hematopoietic malignancies. Blood Adv. 2018;2(2):146-150. 15. Guidugli L, Johnson AK, Alkorta-Aranburu G, et al. Clinical utility of gene panel-based testing for hereditary myelodysplastic syndrome/acute leukemia predisposition syndromes. Leukemia. 2017;31(5):1226-1229. 16. Valent P, Horny H-P, Bennett JM, et al. Definitions and standards in the diagnosis and treatment of the myelodysplastic syndromes: consensus statements and report from a working conference. Leuk Res. 2007;31(6):727-736. 17. NCC Network. Myelodysplastic syndrome (Version 2.2020). https://www.nccn.org/ professionals/physician_gls/PDF/mds.pdf Accessed May 10, 2020. 18. Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the

518

Association for Molecular Pathology. Genet Med. 2015;17(5):405-423. 19. Padron E, Ball MC, Teer JK, et al. Germ line tissues for optimal detection of somatic variants in myelodysplastic syndromes. Blood. 2018;131(21):2402-2405. 20. Greenberg PL, Tuechler H, Schanz J, et al. Revised international prognostic scoring system for myelodysplastic syndromes. Blood. 2012;120(12):2454-2465. 21. Döhner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017;129 (4):424-447. 22. Takeda J, Yoshida K, Makishima H, et al. Genetic predispositions to sporadic myeloid neoplasms caused by germline DDX41 mutations in Asian and Caucasian populations. Haematologica. 2016; 101 (s1):66-67. 23. Polprasert C, Takeda J, Niparuck P, et al. Novel DDX41 variants in Thai patients with myeloid neoplasms. Int J Hematol. 2020;111(2):241-246. 24. Omura H, Oikawa D, Nakane T, et al. Structural and functional analysis of DDX41: a bispecific immune receptor for DNA and cyclic dinucleotide. Sci Rep. 2016;6(1):1-11. 25. Jiang Y, Zhu Y, Liu Z-J, Ouyang S. The emerging roles of the DDX41 protein in immunity and diseases. Protein Cell. 2017; 8(2):83-89. 26. Crawford J, Lower KM, Hennekam RC, et al. Mutation screening in BörjesonForssman-Lehmann syndrome: identification of a novel de novo PHF6 mutation in a female patient. J Med Genet. 2006; 43(3):238-243. 27. Ernst A, Le VQ, Højland AT, et al. The PHF6 mutation c. 1A> G; PM1V causes Börjeson-

Forsman-Lehmann syndrome in a family with four affected young boys. Mol Syndromol. 2015;6(4):181-186. 28. Todd MA, Ivanochko D, Picketts DJ. PHF6 degrees of separation: the multifaceted roles of a chromatin adaptor protein. Genes. 2015;6(2):325-352. 29. Feurstein S, Godley LA. Germline ETV6 mutations and predisposition to hematological malignancies. Int J Hematol. 2017; 106(2):189-195. 30. Yoshida K, Sanada M, Shiraishi Y, et al. Frequent pathway mutations of splicing machinery in myelodysplasia. Nature. 2011;478(7367):64-69. 31. Kobayashi S, Kobayashi A, Osawa Y, et al. Donor cell leukemia arising from preleukemic clones with a novel germline DDX41 mutation after allogenic hematopoietic stem cell transplantation. Leukemia. 2017;31(4):1020. 32. Berger G, van den Berg E, Sikkema-Raddatz B, et al. Re-emergence of acute myeloid leukemia in donor cells following allogeneic transplantation in a family with a germline DDX41 mutation. Leukemia. 2017;31(2): 520. 33. Drazer MW, Kadri S, Sukhanova M, et al. Prognostic tumor sequencing panels frequently identify germ line variants associated with hereditary hematopoietic malignancies. Blood Adv. 2018;2(2):146-150. 34. Kilpivaara O, Mukherjee S, Schram AM, et al. A germline JAK2 SNP is associated with predisposition to the development of JAK2 V617F-positive myeloproliferative neoplasms. Nat Genet. 2009;41(4):455-459. 35. Brown AL, Hahn CN, Scott HS. Secondary leukemia in patients with germline transcription factor mutations (RUNX1, GATA2, CEBPA). Blood. 2020;136(1):24-35.

haematologica | 2022; 107(2)

ARTICLE

Platelet Biology & its Disorders

Sequence-specific 2'-O-methoxyethyl antisense oligonucleotides activate human platelets through glycoprotein VI, triggering formation of platelet-leukocyte aggregates

Ferrata Storti Foundation

Martina H. Lundberg Slingsby,1,2 Prakrith Vijey,2 I-Ting Tsai,1,2 Harvey Roweth,2 Genevieve Couldwell,2 Adrian R. Wilkie,1,2 Hans Gaus,3 Jazana M. Goolsby,2 Ross Okazaki,2 Brooke E. Terkovich,2 John W. Semple,4,5 Jonathan N. Thon,2 Scott P. Henry,3 Padmakumar Narayanan3 and Joseph E. Italiano Jr.1,2 1

Vascular Biology Program, Department of Surgery, Boston Children’s Hospital, Boston, MA, USA; 2Division of Hematology, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA; 3Nonclinical Development, Ionis Pharmaceuticals Inc., Carlsbad, CA, USA; 4Departments of Pharmacology and Medicine, University of Toronto, Toronto, Ontario, Canada and 5Division of Hematology and Transfusion Medicine, Lund University, Lund, Sweden

Haematologica 2022 Volume 107(2):519-531

ABSTRACT

ntisense oligonucleotides (ASO) are DNA-based, disease-modifying drugs. Clinical trials with 2'-O-methoxyethyl (2’MOE) ASO have shown dose- and sequence-specific lowering of platelet counts according to two phenotypes. Phenotype 1 is a moderate (but not clinically severe) drop in platelet count. Phenotype 2 is rare, severe thrombocytopenia. This article focuses on the underlying cause of the more common phenotype 1, investigating the effects of ASO on platelet production and platelet function. Five phosphorothioate ASO were studied: three 2’MOE sequences; 487660 (no effects on platelet count), 104838 (associated with phenotype 1), and 501861 (effects unknown) and two CpG sequences; 120704 and ODN 2395 (known to activate platelets). Human cord bloodderived megakaryocytes were treated with these ASO to study their effects on proplatelet production. Platelet activation (determined by surface Pselectin) and platelet-leukocyte aggregates were analyzed in ASO-treated blood from healthy human volunteers. None of the ASO inhibited proplatelet production by human megakaryocytes. All the ASO were shown to bind to the platelet receptor glycoprotein VI (KD ~0.2-1.5 mM). CpG ASO had the highest affinity to glycoprotein VI, the most potent platelet-activating effects and led to the greatest formation of platelet-leukocyte aggregates. 2’MOE ASO 487660 had no detectable platelet effects, while 2’MOE ASOs 104838 and 501861 triggered moderate platelet activation and SYKdependent formation of platelet-leukocyte aggregates. Donors with higher platelet glycoprotein VI levels had greater ASO-induced platelet activation. Sequence-dependent ASO-induced platelet activation and platelet-leukocyte aggregates may explain phenotype 1 (moderate drops in platelet count). Platelet glycoprotein VI levels could be useful as a screening tool to identify patients at higher risk of ASO-induced platelet side effects.

Introduction Antisense oligonucleotides (ASO) are short, synthetic, single-stranded DNA molecules (between 8 and 50 nucleotides) that bind to specific mRNA segments through Watson-Crick base pairing.1 They are designed to modulate targeted mRNA by interfering with its function or promoting its degradation, ultimately altering protein expression.1,2 ASO offer therapeutic opportunities for treating rare genetic diseases such as spinal muscular atrophy (nusinersen),3,4 hereditary transthyretin amyloidosis (inotersen)5 and homozygous familial hypercholesterolemia (mipomersen).6 There has been a surge in ASO entering clinical trials for a wide range of diseases, which has been attributed to improvements in chemical modifications of ASO.7 Incorporating a phosphorothioate (PS)-containing backbone to the ASO increased the stability and resistance of the oligonucleotides to nucleolytic degradation.1 Second-

haematologica | 2022; 107(2)

Correspondence: MARTINA LUNDBERG SLINGSBY martina.slingsby@gmail.com Received: June 22, 2020. Accepted: February 2, 2021. Pre-published: February 11, 2021. https://doi.org/10.3324/haematol.2020.260059

519

M.H. Lundberg Slingsby et al.

generation ASO have the PS backbone and often include 2’O-methoxyethyl (MOE) modifications, which can further reduce ASO degradation and increase affinity for the target mRNA.1 While several trials with ASO drugs have shown 2’MOEcontaining ASO drugs to be well tolerated, there have been reports of adverse events including dose-dependent thrombocytopenia, with two phenotypes having been described.8,9 Phenotype 1 is a moderate dose-dependent drop in platelet counts that is reversible with cessation of drug treatment.8,10 The Ionis integrated safety database concluded that certain sequences (3 out of 16 2’MOE ASO) were associated with phenotype 1, with dose-dependent moderate (>30%) declines in platelet count without an effect on bleeding risk.8 Phenotype 1 has also been observed in monkeys treated with 40% of the evaluated 2’MOE ASO, in which it was manifested as a consistent, reproducible decline of platelets over 4-6 weeks, with the platelet counts decreasing moderately (by 30-50%) and then remaining steady.11 Phenotype 2 is a sporadic severe drop in platelet count (clinical thrombocytopenia) and is often not reproducible (for the same 2’MOE ASO), but appears to be dose-dependent.11 Treatment with the 2’MOE ASO inotersen induced phenotype 2 in a few individuals who were predisposed, in relation to their underlying disease, and was shown to be dependent on platelet antibodies.12 This paper will focus on the underlying cause of the more prevalent phenotype 1. Drug-induced thrombocytopenia can stem from decreased production of platelets in the bone marrow and/or increased destruction and clearance of platelets from peripheral blood.13 To examine the in vivo effects on platelet counts, monkeys were treated with the 2’MOE ASO 104838 for 12 weeks.10 Platelet counts decreased ~50% in all monkeys by day 30, and four out of five monkeys had decreases in platelet counts consistent with phenotype 1. The drops in platelet counts were attributed to a 60-80% increase in platelet sequestration in liver and spleen, caused by either increased phagocytosis of platelets or trapping of platelets on the reticuloendothelial surface of these organs.10 Thrombopoietin levels were not altered in the monkeys and bone marrow megakaryocyte morphology, cell density and maturation appeared normal. These findings suggested that the mild thrombocytopenia was due to increased platelet destruction or splenic sequestration and not to diminished platelet production).10 There have been few studies investigating the direct effects of ASO on human platelets. Flierl et al. showed that incubating human platelets with an oligonucleotide PSODN 2395 (without 2’MOE modifications), led to platelet activation (increased platelet P-selectin surface expression).14 The same study established that the PS backbone modification of ODN 2395 is a significant driver of this drug’s effects on platelets, through binding and activation of the platelet receptor glycoprotein VI (GPVI), by enhanced GPVI receptor clustering/dimerization.14 Sewing et al. supported these findings and demonstrated that ASO containing locked nucleic acid modifications had reduced binding to GPVI and platelet activation.15 These studies did not include 2’MOE-modified ASO. Considering that most ASO in clinical use and in the drug development pipeline are 2’MOE ASO, it is imperative to understand how 2’MOE modifications affect platelet responses in humans. Furthermore, it remains unknown whether ASO affect proplatelet production from megakaryocytes and to what 520

extent 2’MOE-modified PS-ASO share the platelet-activating effects of PS-ODN 2395 on human platelets. In the current study, we therefore focused on investigating the in vitro effects of 2’MOE ASO on proplatelet production from human cord blood-derived megakaryocytes and studied the direct effects of 2’MOE ASO on human platelets as well as interactions with immune cells. Specifically, we tested 2’MOE ASO 104838, which is known to cause phenotype 1 platelet count reductions.10,16 We also included 2’MOE ASO 501861, with unknown effects on human platelets, and 2’MOE ASO 487660, which has not been associated with reductions in platelet counts in monkeys. CpG ASO 2395 and another CpG 2’MOE ASO 120704 were included for comparison for their known platelet-activating effects.10,14

Methods Human and mouse megakaryocyte cultures and proplatelet analysis Human cord blood-derived primary CD34+ cells were cultured and mature megakaryocytes were purified by magnetic bead separation on day 11 as previously described.17 Mouse megakaryocytes were derived from fetal liver cultures extracted from CD-1 pregnant mice at day 13.5 of gestation.18 Human-derived, or mouse-derived megakaryocytes were plated on a 96-well half-area plate (Greiner Bio one 675101), followed by addition of ASO (5 mM) and imaged at hourly intervals for 24 h using an IncuCyte Live Cell Analysis System (IncuCyte Zoom). These images were analyzed for the percentage of megakaryocytes producing proplatelets, as well as the area containing proplatelets using Ilastik (version 1.3.0) and Cell Profiler (version 3.0.0) as described previously.19

Blood collection Blood was collected from healthy male and female human donors, after informed consent and institutional review board approval (2012P001526), in accordance with the Declaration of Helsinki. Washed platelets, platelet-rich plasma or whole blood was treated for 30 min (platelet studies) or 6 h (cytokine release experiments) with 1, 5 or 10 mM ASO. Previous studies have shown that ASO-induced activation of platelets is concentrationdependent.10,14,15 A therapeutically relevant dose of ASO, i.e., 300 mg, administered subcutaneously results in a Cmax (maximum concentration that a drug achieves after dosing) of 1-2 mM.20

Antisense oligonucleotides ASO were synthesized at Ionis Pharmaceuticals, Inc. (Carlsbad, CA, USA), purified by reverse-phase high-performance liquid chromatography and formulated in 10 mM HEPES. Five ASO (all with a PS backbone) were included: three with 2’-MOE sequences; 487660 (no effects on platelet count), 104838 (associated with phenotype 1),8,10 501861 (effects unknown) and two with CpG sequences; 120704 and ODN 2395 (known to activate platelets)10,14 (Table 1). More details of the Methods can be found in the Online Supplementary Material.

Results Proplatelet formation from human cord blood or murine fetal liver-derived megakaryocytes is not inhibited by 2’MOE or CpG ASO treatment To investigate whether the ASO induced thrombocyhaematologica | 2022; 107(2)

Effects of 2'MOE ASO on human platelets

Table 1. Antisense oligonucleotides included in this study.

Ionis ASO

Back bone

Length

PS load*

Platelet side effects

487660

2’MOE PS

No reported drops in platelet count

487660 Sequence (5’3’) 104838 104838 Sequence (5’3’) 501861 501861 Sequence (5’3’) 120704 120704 Sequence (5’3’) 818290 (ODN 2395) ODN 2395 Sequence (5’3’)

C*MOEC*MOEA*MOEG*MOEC*T*C*A*A*C*C*C*T*T*CMOET*MOET*MOET*MOEA*MOEA*

MOE

2’MOE PS

Phenotype 1-moderate drops in platelet count8,10

G*MOEC*MOET*MOEG*MOEA*T*T*A*G*A*G*A*G*A*G*MOEG*MOET*MOEC*MOEC*MOEC*

MOE

2’MOE PS

Unknown

T*MOEC*MOEA*MOEC*MOEA*G*A*A*T*T*A*T*C*A*G*MOEC*MOEA*MOEG*MOET*MOEA*

MOE

CpG PS

Platelet activation10

T*C*pG*T*C*pG*T*T*T*T*G*T*C*pG*T*T*T*T*G*T*C*pG*T*T* CpG PS

Platelet activation14

T*C*pG*T*C*pG*T*T*T*T*C*pG*G*C*G*C*G*C*G*C*C*pG*

All antisense oligonucleotides (ASO) were phosphorothioate (PS)-modified (locations indicated with *). MOE indicates the position of the 2'-O-methoxyethyl (2’MOE)-modified sugar residues with 2ʹ-deoxynucleotides in between. All cytosine residues were methylated at the five position. 818290 is ODN 2395 and is referred to as ODN 2395 in the paper, for ease of comparison to previous reports.14ODN 2395 and 120704 are non-MOE ASO with unmethylated CpG dinucleotide-rich motifs.

topenia by affecting proplatelet production from megakaryocytes, human cord blood-derived mature megakaryocytes were incubated with 5 mM ASO for 24 h. Puromycin, an inhibitor of protein synthesis known to impair proplatelet production,21,22 had an inhibitory effect on both human- and murine-derived megakaryocytes (Figure 1A, B). Compared to the vehicle, none of the ASO tested lowered proplatelet counts from the human-derived megakaryocytes (Figure 1A). Similarly, there was no decrease in percent proplatelet producing murine fetal liverderived megakaryocytes following incubation with any of the ASO (Figure 1B). The CpG ASO 120704 slightly increased proplatelet counts in the human-derived megakaryocytes after 22 h (117±5 vs. vehicle 79±7) (Figure 1A). There was also a small increase in proplatelet-producing murine megakaryocytes after 24 h (Figure 1B); vehicle (17±1%), 487660 (25±1%), 104838 (23±1%), 501861 (25±2%), 120704 (23±1%), and ODN 2395 (26±1%). Representative images of proplatelet production at 0, 8, 16 and 24 h by murine megakaryocytes treated with 2’MOE ASO 104838 or CpG ASO ODN 2395 appeared comparable to those of megakaryocytes treated with the vehicle (Figure 1C).

Human platelets internalize both 2’MOE and CpG ASO Given that the ASO did not appear to inhibit proplatelet production, we focused on examining the direct effects of ASO on human platelets. We used immune-electron microscopy to visualize how ASO (at 5 mM) interact with human washed platelets. ASO with a PS backbone are haematologica | 2022; 107(2)

hydrophilic, poly-anionic molecules with a high degree of plasma protein binding (typically >90%), and circulate transiently in the blood before interacting with cell-surface proteins and typically gaining entry into cells by endocytosis.23 The plasma-free conditions in the experiments shown in Figure 2 were intended to maximize visualization of ASO binding to platelets. Electron micrographs revealed that ASO immunogold staining was either localized to the platelet plasma membrane or, when internalized, to the cytoplasm with sporadic staining of internal membranes and granules (Figure 2B-F). The 2’MOE ASO 487660 appeared to stain less than the other ASO, especially at the plasma membrane (Figure 2B-F). To exclude any direct platelet cytotoxicity of ASO, we performed a lactate dehydrogenase leakage assay in washed platelets and confirmed that none of the ASO was cytotoxic at doses of 1 or 5 mM (Online Supplementary Figure S1).

2’MOE ASO (104838 and 501861) and CpG ASO increase platelet P-selectin expression in platelet-rich plasma and whole blood To establish whether the ASO activate human platelets directly, we investigated the effects on platelet surface Pselectin levels using flow cytometry (Figure 3). Plateletrich plasma treated with the positive control thrombin receptor activating peptide (TRAP) showed a significant increase in P-selectin surface expression, while the 2’MOE ASO 487660 (which does not affect platelet count) did not increase platelet P-selectin compared to vehicle (Figure 3A). Platelet activation nearly doubled after treatment 521

M.H. Lundberg Slingsby et al.

Figure 1. The effect of antisense oligonucleotides on proplatelet production by megakaryocytes derived from human cord blood and mouse fetal liver. Human and mouse megakaryocytes (MK) were treated with vehicle (HEPES, 10 mM), puromycin (100 nM, a known inhibitor of proplatelet production), or 5 mM of the 2’MOE antisense oligonucleotides (ASO) 104838 or 501861 or the CpG ASO 120704 or ODN 2395 for 24 h using the IncuCyte Live Cell Analysis System and analyzed with Ilastik and Cell Profiler. (A) Proplatelet count: individual count of proplatelet protrusions that emanate from the human cord blood-derived MK body, over the course of the 24 h incubation. (B) Percent of proplatelet-producing mouse fetal liver-derived MK treated with the above treatments as well as the 2’MOE ASO 487660. *P<0.05 by twoway analysis of variance with Bonferroni post-test (n=4 repeat runs). (C) Representative brightfield images at 20x magnification, of proplatelet-producing mouse fetal liver-derived MK at 0 h, 8 h, 16 h and 24 h treated with vehicle (HEPES), the 2’MOE ASO 104838 or the CpG ASO ODN 2395. Scale bars indicate 50 mm.

522

haematologica | 2022; 107(2)

Effects of 2'MOE ASO on human platelets

with 5 mM of the 2’MOE ASO 104838 and 501861 (19±3% and 20±3%, respectively), compared to vehicle (10±2%); the CpG ASO 120704 and ODN 2395 activated the platelets more potently (38±3% and 38±4%, respectively) (Figure 3A). We also studied platelet activation in whole blood, in which platelet P-selectin was elevated by the platelet agonists TRAP and collagen (Figure 3B). Pre-treatment with a Spleen tyrosine kinase (SYK) inhibitor (PRT-060318) decreased collagen, but not TRAP activation since SYK is downstream of the collagen receptor GPVI signaling in platelets (Figure 3B). The selective inhibitory effect on collagen signaling is demonstrated more clearly in Figure 6A. The 2’MOE ASO 104838 and 501861 had mild plateletactivating effects in whole blood (similar to that of collagen) whereas the CpG ASO 120704 and ODN 2395 had

stronger effects (comparable to the effect of TRAP) (Figure 3B). Pre-treatment with the SYK inhibitor blocked the ASO-induced P-selectin expression (Figure 3B).

Responsiveness to ASO treatment is strongly correlated to individual GPVI levels Since we noticed donor-to-donor variability in the responsiveness to ASO treatment (Figure 3B), we investigated whether this could be related to differential platelet surface expression of GPVI receptors, which we measured by flow cytometry. Basal platelet GPVI levels varied between donors and activating the platelet-rich plasma with TRAP reduced the platelet GPVI levels (Online Supplementary Figure S2), consistent with reports of GPVI shedding upon platelet activation.24 Pearson correlation analysis showed a strong positive correlation (correlation

Figure 2. Uptake and localization of antisense oligonucleotides in human washed platelets. Washed platelets were incubated with a therapeutically relevant concentration of the antisense oligonucleotides (ASO) (5 mM), then fixed and stained with an anti-ASO antibody and labeled with protein A-gold. Representative electron microscopy images at 15000x of ASO localization, using anti-ASO immunogold labeling (shown as black dots) of human washed platelets treated for 30 min with (A) vehicle (HEPES, 10 mM), (B) 2’MOE ASO 487660, (C) 2’MOE ASO 104838, (D) 2’MOE ASO 501861, (E) CpG ASO 120704, or (F) CpG ASO ODN 2395 (all at 5 mM). Black arrows indicate ASO localizing on the platelet membrane and white arrows indicate internalized ASO.

haematologica | 2022; 107(2)

523

M.H. Lundberg Slingsby et al.

Figure 3. Effect of antisense oligonucleotides on platelet activation marker P-selectin and stromal cell derived factor 1a release in human platelet-rich plasma and whole blood and correlation with individual platelet glycoprotein VI receptor levels. Platelet activation was identified through increases in P-selectin expression on the platelet surface after 30 min treatment with vehicle (HEPES, 10 mM) or thrombin receptor activating peptide (TRAP, 25 mM, to activate the platelets) or 5 mM of the antisense oligonucleotides (ASO): 104838 (2’MOE ASO), 501861 (2’MOE ASO), 120704 (CpG ASO) and ODN 2395 (CpG ASO), assessed by flow cytometry in (A) platelet-rich plasma (PRP) (from 9 human donors, the 2’MOE ASO 487660 was included in 3 of the experiments) and (B) whole blood (WB) (from 8 human donors) pretreated or not with the spleen tyrosine kinase (SYK) inhibitor PRT-060318 (10 mM) before addition of vehicle, TRAP (25 µM) or ASO (5 µM) that increased P-selectin in PRP (i.e., 104838, 501861, 120704 and ODN 2395). *P<0.05 compared to vehicle by one-way analysis of variance (ANOVA), with the Dunnett post-test. #P<0.05 paired Student t-test for the effect of the SYK inhibitor. (C, D) Individual donor platelet glycoprotein (GP)VI receptor levels (median fluorescence intensity, MFI) were correlated to the same individual platelet P-selectin levels after treatment with: (C) 2’MOE ASO 104838 or 501861, and (D) CpG ASO 120704 or ODN 2395, (7 human donors). P<0.05 by Pearson correlation analysis. (E, F) Stromal cell derived factor 1α (SDF1a), released from platelets upon activation, was measured by Mesoscale U-plex multiplex assay in (E) PRP and (F) WB blood treated for 30 min with vehicle (HEPES), TRAP (25 mM) or 1, 5 or 10 mM of the ASO in blood from four human donors. *P<0.05 compared to vehicle by one-way ANOVA, with Dunnett post-test.

524

haematologica | 2022; 107(2)

Effects of 2'MOE ASO on human platelets

vated platelets, we measured secreted plasma levels of stromal cell derived factor 1a (SDF1a) in ASO-treated platelet-rich plasma and whole blood. TRAP stimulated SDF1a release in both platelet-rich plasma and whole blood (Figure 3 E, F). Consistent with the pattern of the Pselectin effects, the CpG ASO 120704 and ODN 2395 triggered robust release of SDF1a in platelet-rich plasma and whole blood in a concentration-dependent manner: thus the effect could be seen with doses of 5 and 10 mM, but

coefficients: 104838 r=0.97, 501861 r=0.96, 120704 r=0.85, ODN 2395 r=0.86) between donors’ platelet GPVI levels and their platelet activation in response ASO treatment, with the highest ASO responders also having the highest platelet GPVI levels (Figure 3 C, D).

2’MOE ASO (104838 and 501861) and CpG ASO trigger SDF1a release from whole blood As an additional marker of a-granule release from acti-

Figure 4. Platelet aggregation in human platelet-rich plasma and whole blood treated with antisense oligonucleotides. (A, B) Ninety-six-well platelet aggregometry was used to generate full concentration-response curves to the platelet agonist collagen after incubating platelet-rich plasma (PRP) with vehicle or (A) 5 mM of the 2’MOE antisense oligonucleotides (ASO) 487660, 104838 and 501861 or (B) the CpG ASO 120704 or ODN 2395. *P<0.05 by two-way analysis of variance (ANOVA) with Bonferroni post-test, compared to vehicle (HEPES, 10 mM), n=6-8 human donors. (C) Platelet aggregation was also assessed using traditional light transmission aggregometry (LTA) by incubating PRP (n=8 human donors) with the ASO for 30 min at 1200 rpm stirring speed without stimulation (to detect spontaneous aggregation) or after stimulation with the platelet agonist thrombin receptor activating peptide (TRAP, 25 mM). (D) Impedance aggregometry was used to analyze platelet aggregation by incubating whole blood (WB) (n=7 human donors) with the ASO (at 1 and 5 mM) for 30 min at 1200 rpm stirring speed, without stimulation, or TRAP (25 mM) stimulated aggregation. (E) Platelet-platelet aggregates in WB (n=7 human donors) treated with vehicle (HEPES, 10 mM), collagen (20 mg/mL), TRAP (25 mM) or ASO (5 mM) were analyzed using flow cytometry. (F) Platelet-leukocyte aggregates (platelet marker CD41/61+ leukocyte marker CD14+) were analyzed in WB (n=6 human donors) treated with vehicle (HEPES, 10 mM), collagen (20 mg/mL), TRAP (25 mM), or ASO (5 mM), using flow cytometry. *P<0.05 compared to vehicle by one-way ANOVA, Dunnett post-test (C, D, E, F).

haematologica | 2022; 107(2)

525

M.H. Lundberg Slingsby et al.

Figure 5. Immunostimulatory effects after 30 minutes and 6 hours in antisense oligonucleotide-treated whole blood. Whole blood (WB) from four human donors was incubated with vehicle (HEPES, 10 mM), thrombin receptor activating peptide (TRAP, 25 mM), lipopolysaccharide (LPS 0.01 mg/mL) or 1, 5 or 10 mM of the antisense oligonucleotides (ASO). Interleukin 8 (IL-8) and monocyte chemotactic protein-1 (MCP-1) release was measured in plasma (pg/mL), using an MSD U-plex assay, after incubation for 30 min and 6 h at 37°C in 5% CO2. (A) IL-8 after 30 min. (B) MCP-1 after 30 min. (C) IL-8 after 6 h. (D) MCP-1 after 6 h. *P<0.05 compared to vehicle by one-way analysis of variance, Dunnett post-test. Concentration-response curves for (E) IL-8 release after 6 h and (F) MCP-1 release after 6 h.

not at a dose of 1 mM (Figure 3E, F). The 2’MOE ASO 104838 and 501861 triggered a small release at 5 mM in whole blood only (Figure 3F). The 2’MOE ASO 487660 did not trigger SDF1a release at any concentration tested, in either platelet-rich plasma or whole blood (Figure 3E, F).

2’MOE ASO 104838 increases platelet reactivity to collagen To investigate whether the platelet activation documented in Figure 3 would translate into an effect on platelet aggregation, we performed 96-well aggregometry and traditional light transmission aggregometry under both unstimulated and stimulated conditions (Figure 4AD). The 2’MOE ASO 104838 triggered a small potentiating effect on collagen-induced platelet aggregation, but this effect was not seen after treatment with the other two 2’MOE ASO (Figure 4A). The CpG ASO 120704 and ODN 2395 also increased platelet reactivity to collagen (Figure 4B), including ~40% spontaneous aggregation in two 526

donors with high platelet GPVI levels (Online Supplementary Figure S2). However, no significant spontaneous aggregation was detected by light transmission aggregometry in any of the ASO-treated platelet-rich plasma samples and stimulating the 5 mM ASO-treated samples with TRAP resulted in a similar light transmission aggregometry-measured aggregation response to that following stimulation with the vehicle (Figure 4C). Differences in the mechanical shear stress environment in 96-well aggregometry versus light transmission aggregometry25 may explain the discrepancy in results between these assays.

2’MOE ASO (104838 and 501861) and CpG ASO potentiate platelet reactivity to TRAP in whole blood impedance aggregometry When whole blood was incubated with 5 mM of the CpG ASO ODN 2395 in an impedance aggregometer, spontaneous aggregation was detected over the course of haematologica | 2022; 107(2)

Effects of 2'MOE ASO on human platelets

Figure 6. SYK inhibition of the formation of platelet-leukocyte aggregates in whole blood treated with antisense oligonucleotides. Whole blood (WB) from four human donors was pretreated or not with a spleen tyrosine kinase (SYK) inhibitor (PRT-060318, 10 mM) followed by vehicle (10 mM HEPES), thrombin receptor activating peptide (TRAP, 25 mM), collagen (20 mg/mL), or the 2’MOE non CpG antisense oligonucleotides (ASO): 104838 or 501861, or the CpG ASO: 120704 or ODN 2395 (all at 5 mM) and analyzed by flow cytometry for: (A) platelet-leukocyte aggregates (PLA, platelet marker CD41/61+, leukocyte marker CD14+). *P<0.05 compared to vehicle by one-way analysis of variance (ANOVA), Dunnett post-test, #P<0.05 paired Student t-test for the effect of the SYK inhibitor. (B) Representative confocal images of platelet-leukocyte aggregates in fixed and fluorescently labeled WB. Red represents CD45 (leukocyte marker), green represents CD41/61 (platelet marker). The white boxes indicate the location of the zoomed-in part of image shown to the right. Scale bar = 10 mm.

the 30 min incubation (12±6 Ω vs. 0.4±0.3 Ω with vehicle) (Figure 4D). When platelet-platelet interactions were studied in ASO-treated (but otherwise unstimulated) whole blood samples using flow cytometry, only the CpG ASO 120704 and ODN 2395 had significantly more plateletplatelet aggregates (13±6 and 14±6%, respectively, vs. 0.2±0.04% with vehicle) (Figure 4E), consistent with the impedance aggregometry results. After stimulation with TRAP, whole blood aggregation was potentiated to a similar level in both the 2’MOE ASO (104838 and 501861) and the CpG (120704 and ODN 2395)-treated samples at both 1 and 5 mM (Figure 4D).

2’MOE ASO (104838 and 501861) and CpG ASO trigger formation of platelet-leukocyte aggregates in unstimulated whole blood The impedance aggregometry results led us to hypothesize that the enhancement of whole blood aggregation haematologica | 2022; 107(2)

was perhaps not solely driven by homogenous plateletplatelet aggregates but could also contain heterogeneous platelet-leukocyte aggregates.26 We, therefore, analyzed ASO-treated (but otherwise unstimulated) whole blood for the formation of platelet-leukocyte aggregates by flow cytometry (Figure 4F). TRAP and collagen triggered substantial formation of platelet-leukocyte aggregates, whereas the 2’MOE ASO 487660 evoked a similar response to that produced by the vehicle (Figure 4F). However, whole blood treated with 2’MOE ASO 104838 and 501861 and the CpG 120704 and ODN 2395 had more platelet-leukocyte aggregates (33±8, 37±9, 69±4, and 46±4%, respectively) than vehicle-treated whole blood (12±1%) (Figure 4F).

Brief ASO treatment does not induce neutrophil or monocyte activation To further explore the interaction of ASO with mono527

M.H. Lundberg Slingsby et al.

Figure 7. The effect of antisense oligonucleotides on platelet-neutrophil, platelet-monocyte and glycoprotein VI interactions. Whole blood (WB) was incubated with vehicle (10 mM HEPES), thrombin receptor activating peptide (TRAP, 25 mM), or the 2’MOE antisense oligonucleotides (ASO): 104838 or 501861, or the CpG ASO: 120704 or ODN 2395 (all at 5 mM) and analyzed by flow cytometry for (A) platelet-neutrophil aggregates (platelet marker CD41/61+, neutrophil marker CD66b+), (B) platelet-monocyte aggregates (platelet marker CD41/61+, monocyte marker CD14+). (C) Surface P-selectin-positive (CD62p+) platelet-neutrophil aggregates (CD41/61+, CD66b+, CD62p+) (median fluorescence intensity, MFI). (D) Surface P-selectin-positive platelet-monocyte aggregates (CD41/61+, CD14+, CD62p+) (MFI) in WB from five to nine human donors. (E) Individual donor platelet glycoprotein (GP)VI levels were correlated to platelet-neutrophil aggregate formation after treatment with the ASO in six human donors. *P<0.05 by Pearson correlation analysis. (F) A fluorescence polarization assay was used to measure binding affinity of Alexa647-labeled ASO to human GPVI. Bmax is the total density of receptors in a sample and KD is the equilibrium dissociation constant. The smaller the KD, the greater the binding affinity of the ASO to human GPVI.

cytes and neutrophils, we assessed whether ASO increased surface expression of CD11b, a broad immune cell activation marker. Lipopolysaccharide was included as a positive control to increase expression of CD11b on the surface of neutrophils (Online Supplementary Figure S4A) or monocytes (Online Supplementary Figure S4B). SYK is also involved in leukocyte intracellular signaling,27,28 and there was an inhibitory effect on CD11b surface expression in SYK-treated samples exposed to vehicle and lipopolysaccharide (Online Supplementary Figure S4 A, B). None of the ASO tested had any effect on CD11b expression on either neutrophils or monocytes (after 30 min incubation); hence the ASO did not appear to activate these cells directly within this timeframe (Online 528

Supplementary Figure S4A,B). In support of these data, the proinflammatory chemokines interleukin-8 (IL-8) and monocyte chemoattractant protein-1 (MCP-1) were not released from whole blood incubated with ASO (1, 5 and 10 mM) for 30 min (Figure 5A, B).

Treatment with CpG ASO (but not 2’MOE ASO) leads to IL-8 and MCP-1 release CpG motifs have been shown to be immunostimulatory15 and proinflammatory effects of the CpG ASO 120704 and ODN 2395 were apparent after 6 h of incubation of whole blood, with 5 and 10 mM (but not 1 mM) resulting in robust IL-8 and MCP-1 release (Figure 5C, D). None of the 2’MOE ASO evoked a proinflammatory effect at any haematologica | 2022; 107(2)

Effects of 2'MOE ASO on human platelets

of the concentrations (Figure 5C-D). The concentrationdependent effects of 120704 and ODN 2395 are shown in Figure 5E, F.

104838, 501861, 120704 and ODN 2395-induced formation of platelet-leukocyte aggregrates in unstimulated whole blood is blocked by SYK pretreatment When whole blood was pretreated with a SYK inhibitor, collagen-induced platelet-leukocyte aggregates were markedly reduced, while TRAP-induced platelet-leukocyte aggregates were maintained, confirming the selectivity of the SYK inhibitor (Figure 6A). SYK pre-treatment completely reversed the ASO-induced formation of platelet-leukocyte aggregates (Figure 5A). Confocal imaging of the aggregates using a leukocyte marker (CD45) and a platelet marker (CD41/61), confirmed the presence of platelet-leukocyte aggregates (Figure 6B) but showed greater platelet-platelet aggregation in the platelet-leukocyte aggregates in whole blood treated with the CpG ASO (which was more similar to TRAP-treated samples) compared to the 2’MOE ASO, which mainly involved single platelets bound to the leukocytes (Figure 6B).

ASO increase levels of P-selectin-enriched platelet-neutrophil and platelet-monocyte aggregates, through a SYK-dependent mechanism, correlating to GPVI levels Further investigation into the types of immune cells that were driving the ASO-induced formation of plateletleukocyte aggregates revealed an increase in both platelet-neutrophil aggregates and platelet-monocyte aggregates (Figure 7A, B). The CpG ASO 120704 and ODN 2395 once again produced a more robust response than the 2’MOE ASO 104838 and 501861 (65±7 and 47±7 vs. 24±10 and 27±11 vs. 6±1 for vehicle). The platelet-neutrophil and platelet-monocyte aggregates that formed in the ASO-treated samples were enriched in P-selectin (Figure 7C, D) and the greatest increase was observed in the samples treated with 120704 and ODN 2395. However, platelet-neutrophil aggregates and plateletmonocyte aggregates were not enriched in CD11b (Online Supplementary Figure S4C, D), implying that the initial formation of these aggregates was driven by activated platelets (not activated leukocytes). Pearson correlation analysis between platelet GPVI levels and platelet neutrophil aggregates showed a strong positive correlation for all the ASO tested; higher platelet GPVI expression was associated with a stronger platelet-neutrophil aggregate response (Figure 7E).

CpG ASO bind to GPVI with a higher affinity compared to 2’MOE ASO Fluorescence polarization experiments revealed that all ASO bound to human GPVI, with equilibrium dissociation constant (KD) values in the low micromolar range (Figure 7F): the smaller the KD, the greater the binding affinity. All the ASO bound to GPVI with stronger affinity than to control human serum albumin (Online Supplementary Figure S5). The 2’MOE 487660 showed stronger affinity (KD 12.9 mM) to human serum albumin than the other two 2’MOE ASO (104838 KD 24 mM and 501861 KD 26 mM). The CpG ASO displayed higher affinity (lower KD) for GPVI compared to the 2’MOE ASO (Figure 7F). haematologica | 2022; 107(2)

Discussion The main findings from the current study are: (i) none of the ASO sequences investigated had an inhibitory effect on proplatelet production by either human- or mousederived megakaryocytes; (ii) all ASO demonstrated uptake into human platelets (2’MOE 487660 less than the others); (iii) a subset of 2’MOE ASO (104838 and 501861) and CpG ASO (120704 and ODN 2395) activated human platelets, triggering P-selectin and SDF1a release from platelet a-granules and platelet-leukocyte aggregate formation; (iv) the ASO-induced platelet activation and platelet-leukocyte aggregate formation were fully reversed by pre-treatment with a SYK inhibitor; (v) ASO-induced platelet effects appeared to be sequence-dependent, rather than 2’MOE-dependent, since the 2’MOE ASO 487660 did not affect platelet function; (vi) all the ASO were shown to bind to human GPVI (CpG ASO had the strongest affinity); (vii) only the CpG ASO had a concentration-dependent proinflammatory effect, triggering IL-8 and MCP-1 release in whole blood after incubation for 6 h; (viii) the responsiveness to the ASO-induced plateletactivating and platelet-leukocyte aggregate-forming effects varied between donors and showed a strong positive correlation to individual platelet GPVI surface expression. The novelty of this study concerns the effect of 2’MOE ASO on human platelets. Our focus was the 2’MOE ASO 104838, which has been shown to be representative of a subset of ASO sequences that produce phenotype 1 reductions in platelet count in monkeys and humans.8,10 We identified that the 2’MOE ASO 104838 binds to GPVI receptors on human platelets, increasing platelet surface P-selectin and prompting the formation of platelet-neutrophil aggregates and platelet-monocyte aggregates, most likely through an interaction between platelet P-selectin and leukocyte P-selectin glycoprotein ligand 1 (PSGL-1).29 This appears to be a platelet-driven interaction (at least when tested in vitro), as there were no signs of leukocyte activation (CD11b or IL-8/MCP-1 release) even after 10 µM of the 2’MOE ASO 104838. Mechanistically, we suggest that 2’MOE ASO 104838 lowers platelet count, producting a phenotype 1, by activating platelets, an effect which triggers platelet-leukocyte aggregates and subsequent clearance of platelets by leukocytes. Maugeri et al. showed that platelet-neutrophil interactions might contribute to platelet clearance through active phagocytosis of P-selectin-positive platelets by neutrophils.30 The internalization of platelets by neutrophils has been observed in patients with viral infection (associated with reductions in platelet count).31 Increased levels of P-selectin can also enhance sequestration of platelets on vascular endothelial surfaces.32 When monkeys were treated with 2’MOE ASO 104838, there was increased platelet sequestration in the densely vascularized liver and spleen.10 However, in vitro incubation of platelet-rich plasma from treatment-naïve monkeys with 2’MOE ASO 104838 did not lead to platelet activation10 (as we observed in human platelet-rich plasma and whole blood). This may be because the sample size used (n=3) did not capture the variability seen in ASO responsiveness. Monkeys treated with 2’MOE ASO 104838 also had higher plasma levels of IL-8 and MCP-1,10 which we did not see in our in vitro assay with 104838 in human blood (using doses up to 10 mM). This may reflect in vitro rather than in 529

M.H. Lundberg Slingsby et al.

vivo conditions or be due to species differences. Further investigation is warranted to see if in vivo treatment, or indeed clinical use of certain 2’MOE ASO sequences leads to increased platelet activation and formation of plateletleukocyte aggregates and to what degree this is paired with clinical reductions in platelet count. Our data showing that pre-treatment with a SYK inhibitor was able to fully reverse ASO-induced platelet activation and platelet-leukocyte aggregate formation, speculatively highlights the potential of using a clinically available SYK inhibitor, for instance fostamatinib,33 to treat ASO-induced platelet side effects. This would need to be investigated further. Flierl et al. were the first to identify that ODN 2395 binds to and activates GPVI on platelets.14 In the current study, we confirmed these findings and added the novel finding that 2’MOE ASO also bind to GPVI. Overall, the 2’MOE ASO had lower affinity to GPVI compared to the CpG ASO, which is consistent with their weaker platelet activating effects. In general it has been shown that the stronger affinity with which PS ASO bind, the larger the conformational change of the protein.34 Hence, CpG ASO, with their greater affinity than 2’MOE ASO for GPVI, may induce a more substantial conformational change in the GPVI protein, affecting for instance GPVI clustering/dimerization,14 explaining the greater potency of CpG ASO to activate platelets. Interestingly, the 2’MOE ASO 487660, which has not been shown to lower platelet counts in monkeys, did not affect any of the platelet function tests we performed in this study. Hence, the platelet effects observed with 104838 and 501861 appear to be sequence-dependent rather than due to their 2’MOE-backbone chemistry. 487660 did however still bind to human GPVI with a similar affinity as that of the other 2’MOE ASO, 104838 and 501861. This finding highlights the importance of functional in vitro platelet studies beyond ASO-GPVI binding assays in cell-free environments. The discrepancy by which 2’MOE ASO 487660 can bind GPVI but does not activate platelets may be explained by its greater affinity to bind human serum albumin, compared to the other 2’MOE ASO investigated, which may keep 487660 more bound to plasma proteins. 2’MOE ASO 487660 may also interact differently with the GPVI receptor as there appeared to be less surface and internalized 2’MOE ASO 487660 staining in platelets in the electron-microscopy images.

References 1. Bennett CF, Baker BF, Pham N, Swayze E, Geary RS. Pharmacology of antisense drugs. Annu Rev Pharmacol Toxicol. 2017;57:81-105. 2. Bennett CF, Swayze EE. RNA targeting therapeutics: molecular mechanisms of antisense oligonucleotides as a therapeutic platform. Annu Rev Pharmacol Toxicol. 2010;50:259-293. 3. Finkel RS, Mercuri E, Darras BT, et al. Nusinersen versus sham control in infantile-onset spinal muscular atrophy. N Engl J Med. 2017;377(18):1723-1732. 4. Mercuri E, Darras BT, Chiriboga CA, et al. Nusinersen versus sham control in lateronset spinal muscular atrophy. N Engl J Med. 2018;378(7):625-635.

530

There is a degree of subject variability in platelet count reductions following ASO treatment in monkeys and humans.8,10,16 We also noticed a high degree of variability in responsiveness to ASO in our in vitro studies of blood from healthy human donors. Although the sample was small (n=7), there was a strong positive correlation between an individual’s platelet GPVI levels and their platelet responsiveness to both 2’MOE (104838 and 501861) and CpG ASO-induced activation. Platelet GPVI levels have been shown to vary in healthy individuals35 and to be increased in different disease states such as obesity.36 Platelet GPVI levels could potentially be useful as a screening tool (used before commencement of treatment) to identify at-risk patients who may be more susceptible to platelet side effects of some sequence-specific GPVIactivating ASO. We have shown that 2’MOE ASO that have been associated with phenotype 1 may not simply be reducing platelet counts, but can also have direct effects on platelets, triggering interactions between platelets and immune cells. In summary, we have defined new mechanisms by which 2’MOE ASO-based drugs affect human platelets, which may yield new strategies to avoid ASO sequences with unfavorable platelet effects. Disclosures JEI has financial interest in and is a founder of Platelet BioGenesis, a company that aims to produce donor-independent human platelets from human-induced pluripotent stem cells at scale. The interests of JEI were reviewed and are managed by the Brigham and Women’s Hospital and Partners HealthCare. The remaining authors declare that they have no conflicts of interest. Contributions MHLS supervised the study, performed research, analyzed data and wrote the manuscript; PV, IT, HR, GC, AW, HG, JM, RO, BE, and JT performed research, analyzed data and edited the manuscript; JWS, SPH and PN analyzed data and edited the manuscript; JEI supervised the study, analyzed data and edited the manuscript. Funding This work was funded by a corporate sponsored research agreement between Ionis Pharmaceuticals Inc. and Brigham and Women’s Hospital and Partners HealthCare in accordance with their conflict-of-interest policies.

5. Mathew V, Wang AK. Inotersen: new promise for the treatment of hereditary transthyretin amyloidosis. Drug Des Devel Ther. 2019;13:1515-1525. 6. Geary RS, Baker BF, Crooke ST. Clinical and preclinical pharmacokinetics and pharmacodynamics of mipomersen (Kynamro®): a second-generation antisense oligonucleotide inhibitor of apolipoprotein B. Clin Pharmacokinet. 2015;54(2):133-146. 7. Schoch KM, Miller TM. Antisense oligonucleotides: translation from mouse models to human neurodegenerative diseases. Neuron. 2017;94(6):1056-1070. 8. Crooke ST, Baker BF, Witztum JL, et al. The effects of 2'-O-methoxyethyl containing antisense oligonucleotides on platelets in human clinical trials. Nucleic Acid Ther.

2017;27(3):121-129. 9. Chi X, Gatti P, Papoian T. Safety of antisense oligonucleotide and siRNA-based therapeutics. Drug Discov Today. 2017; 22(5):823-833. 10. Narayanan P, Shen L, Curtis BR, et al. Investigation into the mechanism(s) that leads to platelet decreases in Cynomolgus monkeys during administration of ISIS 104838, a 2'-MOE-modified antisense oligonucleotide. Toxicol Sci. 2018;164(2): 613-626. 11. Henry SP, Narayanan P, Shen L, Bhanot S, Younis HS, Burel SA. Assessment of the effects of 2'-methoxyethyl antisense oligonucleotides on platelet count in Cynomolgus nonhuman primates. Nucleic Acid Ther. 2017;27(4):197-208. 12. Narayanan P, Curtis BR, Shen L, et al.

haematologica | 2022; 107(2)

Effects of 2'MOE ASO on human platelets

Underlying immune disorder may predispose some transthyretin amyloidosis subjects to inotersen-mediated thrombocytopenia. Nucleic Acid Ther. 2020;30(2):94103. 13. Aster RH, Bougie DW. Drug-induced immune thrombocytopenia. N Engl J Med. 2007;357(6):580-587. 14. Flierl U, Nero TL, Lim B, et al. Phosphorothioate backbone modifications of nucleotide-based drugs are potent platelet activators. J Exp Med. 2015; 212(2):129-137. 15. Sewing S, Roth AB, Winter M, et al. Assessing single-stranded oligonucleotide drug-induced effects in vitro reveals key risk factors for thrombocytopenia. PLoS One. 2017;12(11):e0187574. 16. Sewell KL, Geary RS, Baker BF, et al. Phase I trial of ISIS 104838, a 2'-methoxyethyl modified antisense oligonucleotide targeting tumor necrosis factor-alpha. J Pharmacol Exp Ther. 2002;303(3):13341343. 17. Ferrer-Marin F, Stanworth S, Josephson C, Sola-Visner M. Distinct differences in platelet production and function between neonates and adults: implications for platelet transfusion practice. Transfusion. 2013;53(11):2814-2821. 18. Vijey P, Posorske B, Machlus KR. In vitro culture of murine megakaryocytes from fetal liver-derived hematopoietic stem cells. Platelets. 2018;29(6):583-588. 19. French SL, Vijey P, Karhohs KW, et al. High Content, Label-free analysis of proplatelet production from megakaryocytes. J Thromb Haemost. 2020;18(10):2701-2711. 20. Yu RZ, Grundy JS, Geary RS. Clinical pharmacokinetics of second generation antisense oligonucleotides. Expert Opin Drug

haematologica | 2022; 107(2)

Metab Toxicol. 2013;9(2):169-182. 21. Thon JN, Devine MT, Jurak Begonja A, Tibbitts J, Italiano JE Jr. High-content livecell imaging assay used to establish mechanism of trastuzumab emtansine (T-DM1)-mediated inhibition of platelet production. Blood. 2012;120(10):1975-1984. 22. Machlus KR, Wu SK, Stumpo DJ, et al. Synthesis and dephosphorylation of MARCKS in the late stages of megakaryocyte maturation drive proplatelet formation. Blood. 2016;127(11):1468-1480. 23. Gaus HJ, Gupta R, Chappell AE, Ostergaard ME, Swayze EE, Seth PP. Characterization of the interactions of chemically-modified therapeutic nucleic acids with plasma proteins using a fluorescence polarization assay. Nucleic Acids Res. 2019;47(3):1110-1122. 24. Rayes J, Watson SP, Nieswandt B. Functional significance of the platelet immune receptors GPVI and CLEC-2. J Clin Invest. 2019;129(1):12-23. 25. Chan MV, Leadbeater PD, Watson SP, Warner TD. Not all light transmission aggregation assays are created equal: qualitative differences between light transmission and 96-well plate aggregometry. Platelets. 2018;29(7):686-689. 26. Russell-Smith NC, Flower RJ, Cardinal DC. Measuring platelet and leucocyte aggregation/adhesion responses in very small volumes of whole blood. J Pharmacol Methods. 1981;6(4):315-333. 27. Berton G, Mocsai A, Lowell CA. Src and Syk kinases: key regulators of phagocytic cell activation. Trends Immunol. 2005; 26(4):208-214. 28. Miller YI, Choi SH, Wiesner P, Bae YS. The SYK side of TLR4: signalling mechanisms in response to LPS and minimally oxidized

LDL. Br J Pharmacol. 2012;167(5):990-999. 29. Kappelmayer J, Nagy B, Jr. The Interaction of selectins and PSGL-1 as a key component in thrombus formation and cancer progression. Biomed Res Int. 2017;2017: 6138145. 30. Maugeri N, Rovere-Querini P, Evangelista V, et al. Neutrophils phagocytose activated platelets in vivo: a phosphatidylserine, Pselectin, and b2 integrin-dependent cell clearance program. Blood. 2009;113(21): 5254-5265. 31. Koupenova M, Vitseva O, MacKay CR, et al. Platelet-TLR7 mediates host survival and platelet count during viral infection in the absence of platelet-dependent thrombosis. Blood. 2014;124(5):791-802. 32. Ed Rainger G, Chimen M, Harrison MJ, et al. The role of platelets in the recruitment of leukocytes during vascular disease. Platelets. 2015;26(6):507-520. 33. Connell NT, Berliner N. Fostamatinib for the treatment of chronic immune thrombocytopenia. Blood. 2019;133(19):2027-2030. 34. Crooke ST, Liang XH, Crooke RM, Baker BF, Geary RS. Antisense drug discovery and development technology considered in a pharmacological context. Biochem Pharmacol. 2020;114196. 35. Furihata K, Clemetson KJ, Deguchi H, Kunicki TJ. Variation in human platelet glycoprotein VI content modulates glycoprotein VI-specific prothrombinase activity. Arterioscler Thromb Vasc Biol. 2001; 21(11):1857-1863. 36. Barrachina MN, Sueiro AM, Izquierdo I, et al. GPVI surface expression and signalling pathway activation are increased in platelets from obese patients: elucidating potential anti-atherothrombotic targets in obesity. Atherosclerosis. 2019;281:62-70.

531

ARTICLE Ferrata Storti Foundation

Red Cell Biology & its Disorders

Sulfated non-anticoagulant heparin derivative modifies intracellular hemoglobin, inhibits cell sickling in vitro, and prolongs survival of sickle cell mice under hypoxia Osheiza Abdulmalik,1* Noureldien H. E. Darwish,2,3* Vandhana Muralidharan-Chari,2° Maii Abu Taleb2 and Shaker A. Mousa2,4 1

Haematologica 2022 Volume 107(2):532-540

Division of Hematology, the Children’s Hospital of Philadelphia, Philadelphia, PA, USA; The Pharmaceutical Research Institute, Albany College of Pharmacy and Health Sciences, Rensselaer, NY, USA; 3Clinical Pathology (Hematology Section), Faculty of Medicine, Mansoura University, Mansoura, Egypt and 4Vascular Vison Pharmaceuticals Co., Rensselaer, NY, USA 2

OA and NHED contributed equally as co-first authors.

Current address: College of Nanoscale Science and Engineering, SUNY Polytechnic Institute, Albany, NY, USA.

ABSTRACT

Correspondence: SHAKER A. MOUSA shaker.mousa@acphs.edu Received: September 15, 2020. Accepted: January 25, 2021. Pre-published: February 11, 2021. https://doi.org/10.3324/haematol.2020.272393

532

ickle cell disease (SCD) is an autosomal recessive genetic disease caused by a single point mutation, resulting in abnormal sickle hemoglobin (HbS). During hypoxia or dehydration, HbS polymerizes to form insoluble aggregates and induces sickling of red blood cells, which increases the adhesiveness of the cells, thereby altering the rheological properties of the blood, and triggers inflammatory responses, leading to hemolysis and vaso-occlusive crises. Unfractionated heparin and low-molecular weight heparins have been suggested as treatments to relieve coagulation complications in SCD. However, they are associated with bleeding complications after repeated dosing. An alternative sulfated non-anticoagulant heparin derivative (S-NACH) was previously reported to have no to low systemic anticoagulant activity and no bleeding side effects, and it interfered with P-selectin-dependent binding of sickle cells to endothelial cells, with concomitant decrease in the levels of adhesion biomarkers in SCD mice. S-NACH has been further engineered and structurally enhanced to bind with and modify HbS to inhibit sickling directly, thus employing a multimodal approach. Here, we show that S-NACH can: (i) directly engage in Schiff-base reactions with HbS to decrease red blood cell sickling under both normoxia and hypoxia in vitro, (ii) prolong the survival of SCD mice under hypoxia, and (iii) regulate the altered steady state levels of pro- and anti-inflammatory cytokines. Thus, our proof-of-concept, in vitro and in vivo preclinical studies demonstrate that the multimodal S-NACH is a highly promising candidate for development into an improved and optimized alternative to low-molecular weight heparins for the treatment of patients with SCD.

Introduction Sickle cell disease (SCD) is a hemoglobinopathy resulting from a mutation replacing the glutamic acid amino acid with the less polar valine amino acid at the sixth position of the b chain, converting normal adult hemoglobin (HbA) to sickle hemoglobin (HbS).1 Deoxygenated HbS polymerizes into long, rigid fibers, causing sickling of red blood cells (RBC).2 These characteristic sickled RBC impair blood flow through the microvasculature, leading to hemolysis, episodes of vaso-occlusion, and multi-organ damage.3-7 The loss of membrane phospholipid asymmetry on sickled RBC exposes phosphatidylserines8 that increase the adhesion of sickled RBC to neutrophils, monocytes, platelets, and endothelial cells to activate coagulation and inflammatory pathways,9-12 culminating in a ‘hypercoagulable’ state.13 Currently, four drugs have been approved by the Food and Drug Administration

haematologica | 2022; 107(2)

Sulfated non-anticoagulant heparin in SCD treatment

for the treatment of SCD in the USA. L-glutamine (Endari), approved in 2017, increases the amount of the reduced form of NADH in erythrocytes, which allows sickle RBC to maintain homeostasis more appropriately during oxidative stress, ultimately resulting in fewer painful vaso-occlusive crises and adverse events.14 Crizanlizumab (Adakveo)15 and voxelotor16 (Oxbryta, GBT440) were approved in 2019. Crizanlizumab is a monoclonal antibody that targets P-selectin to prevent pathological endothelial adhesion of sickle erythrocytes and leukocytes, leading to a reduction in the frequency of painful vaso-occlusive crises.17,18 The anti-sickling agent voxelotor is the first of a new class of aromatic aldehydes that target HbS polymerization by increasing Hb O2 affinity.19-21 Finally, hydroxyurea, which works by inducing the expression of fetal Hb (HbF), is the most proven therapeutic approach for SCD,22,23 as evidenced by its sustained clinical use for over two decades. However, a reported lack of response to hydroxyurea in up to 30% of patients, and supposed poor compliance, tend to limit its use.22 The reported limitations of hydroxyurea led to investigation of other modes of therapy, including the three more recently approved drugs. However, their true benefits will only manifest over time. Additionally, the inherently complex nature of SCD dictates the urgent need for a multimodal form of therapy. Antiplatelet molecules, anticoagulants, and heparin have been investigated to mitigate vaso-occlusive crises.24 Although heparin is beneficial, the associated risks of internal bleeding preclude its utility as a drug25 and the need for alternatives remains critical. We developed a sulfated non-anticoagulant heparin (S-NACH) with no to low systemic anticoagulant activity that can be safely administered in mice (at doses >300 mg/kg daily for 10 days; unpublished data) without causing internal bleeding.26,27 S-NACH does not bind antithrombin and thus does not inhibit systemic antithrombin-dependent clotting factors (activated factors X and II). Sulfation on S-NACH increases the drug’s affinity for endothelium to cause the release of endothelial tissue factor pathway inhibitor (TFPI).26,28 Furthermore, S-NACH interferes with P-selectin-dependent binding of cancer cells29 and RBC30 to endothelial cells and regulates plasma levels of adhesion biomarkers in SCD mice.30 Finally, S-NACH was further optimized to interact directly with HbS to exert desirable therapeutic benefits. In this study we tested our hypothesis that S-NACH can bind to HbS and directly prevent sickling and decrease inflammation in SCD due to the bidirectional relationship between inflammation and coagulation31 and investigated the effects of S-NACH on RBC morphology.

Methods Reagents S-NACH (average molecular weight 4,000 Da) was synthesized by Suzhou Ronnsi Pharma Co. Ltd. (Jiangsu Province, China). 5hydroxymethyl-2-furfural (5-MF) and other common reagents were purchased from Sigma (St. Louis, MO, USA).

Sickle blood samples Leftover blood samples from individuals with homozygous SS (SCD) were obtained and used, based on an approved Institutional Review Board protocol at the Children’s Hospital of Philadelphia,

haematologica | 2022; 107(2)

with informed consent. None of the subjects had been transfused within 4 months prior to their blood samples being used, and four of the five donors were on hydroxyurea therapy.

Anti-sickling, oxygen equilibrium and hemoglobin modification studies using human sickle blood The morphology of hypoxic sickled RBC was evaluated using a previously reported method.22,23 Blood samples from individual donors with SCD (n=5) were diluted using HEMOX buffer supplemented with glucose (10 mM) and bovine serum albumin (0.2%) to adjust the hematocrit of the suspensions to about 20%. We used standardized hematocrit for anti-sickling assays to normalize the ratio of RBC to drug for assay consistency and reproducibility. The suspensions were pre-incubated under air in the absence or presence of three concentrations (0.5, 1, and 2 mM) of S-NACH at 37°C for 1 h. The suspensions were subsequently incubated under a 2.5% O2/97.5% N2 gas mixture at 37°C for 2 h. Aliquots (5–20 mL) of each sample were collected without exposure to air into 2% glutaraldehyde solution for immediate fixation. Fixed cell suspensions were thereafter introduced into glass microslides (Fiber Optic Center, New Bedford, MA, USA)34 and subjected to microscopic morphological analysis of bright field images (at 40x magnification) of single layer cells on an Olympus BX40 microscope fitted with an Infinity 2 camera (Olympus, Waltham, MA, USA), with the coupled Image Capture software. The percentage of sickled cells for each condition was determined using blood with a computer-assisted image analysis system, as described previously.33,35 Untreated samples, as well as samples treated with GBT440/voxelotor, were used as controls. The residual samples were washed in phosphate-buffered saline (PBS) and hemolysed in hypotonic lysis buffer for subsequent analyses. For the oxygen equilibrium study, approximately 100 mL aliquot samples from clarified lysates obtained from the antisickling studies were mixed with 3 mL of 0.1 M potassium phosphate buffer, pH 7.0, in cuvettes, and subjected to hemoximetry analysis using a Hemox™ Analyzer (TCS Scientific Corp., New Hope, PA, USA) to assess P50 shifts.36-38 Degree of P50 shift (DP50) was expressed as percentage fractions of control dimethylsulfoxide-treated samples. Finally, for the Hb adduct formation study, clarified lysates, also from the above anti-sickling study, were subjected to cationexchange high performance liquid chromatography (Hitachi D-7000 Series, Hitachi Instruments, Inc., San Jose, CA, USA), using a weak cation-exchange column (Poly CAT A: 30 mm x 4.6 mm, Poly LC, Inc., Columbia, MD, USA). Hemoglobin isotype peaks were eluted with a linear gradient of mobile phase B from 0% to 80% at A410nm (mobile phase A: 40 mM Bis-Tris, 5 mM EDTA, pH 6.5; mobile phase B: 40 mM Bis-Tris, 5 mM EDTA, 0.15 M sodium chloride, pH 7.5).33,36 A commercial standard consisting of approximately equal amounts of composite HbF, HbA, HbS, and HbC (Helena Laboratories, Beaumont, TX, USA), was used as the reference isotypes. The areas of new peaks, representing HbS adducts, were obtained, calculated as percentage fractions of total Hb area, and reported as levels of modified Hb. All assays were conducted in five biological replicates on different days.

Animal studies C57/B mice aged 5-6 weeks were purchased from Harlan Laboratories (Indianapolis, IN, USA) and acclimatized for 5 days before initiating TFPI measurements after administration of SNACH. Townes SCD mice (stock # 013071) were purchased from The Jackson Laboratory (Bar Harbor, ME, USA), bred, genotyped, and used in experiments between 10 and 12 weeks of age. Animal studies were conducted at the animal research facility, Albany

533

O. Abdulmalik et al.

Figure 1. S-NACH binds to intracellular HbS (and HbF) and inhibits sickling of SS red blood cells under hypoxia. In this experiment, SS red blood cells (RBC) were incubated with or without sulfated non-anticoagulant heparin derivative S-NACH and subjected to hypoxia. (A) Cation-exchange high performance liquid chromatography analyses of aliquot samples demonstrated a concentration-dependent modification of intracellular HbS to the high-affinity adduct form. (B) Fixed SS RBC aliquots were subjected to microscopic image analysis and demonstrated a corresponding dose-dependent inhibition of sickling.

Table 1. Hemoglobin adduct formation, oxygen equilibrium, and anti-sickling studies using homozygous sickle red blood cells with a sulfated nonanticoagulant heparin derivative.

S-NACH 0.5 mM 1.0 mM 2.0 mM 1.0 mM GBT440

Modified Hb (%)a

DP50 (%)b

Sickling inhibition (%)c

20.5±8.2 44.5±13.0 69.7.5±5.5 ND

21.2±10.6 57.6±9.0 65.7±3.2 ND

33.1±5.3 58.6±14.3 85.8±4.7 92.7±4.7

All studies were conducted with SS cell suspensions (20% hematocrit) incubated with 0.5, 1, of 2 mM of sulfated non-anticoagulant heparin derivative (S-NACH). The results are mean values ± standard deviation for five separate experiments (biological replicates).aHbS adduct values obtained from high performance liquid chromatography elution patterns of hemolysate after incubation of compounds with SS cells. bP50 is the oxygen pressure at which hemolysates are 50% saturated with oxygen. DP50 (%) was determined as: DP50 (%) = P50 of lysates from untreated cells - P50 of lysates from treated cells x 100 P50 of lysates from untreated cells c

Anti-sickling studies with SS cells were conducted under hypoxia (2.5% O2/97.5% N2 gas mixture).

College of Pharmacy and Health Sciences (ACPHS; Albany, NY, USA) in accordance with and approved by the ACPHS Institutional Animal Care and Use Committee following institutional guidelines for humane animal treatment. Animals were maintained under standard climatic and light conditions with ad libitum access to food and water. For TFPI analysis, plasma was obtained from three groups of four C57/B mice each, via retroorbital bleeding 2 h after subcutaneous injections of PBS or S-NACH (100 mg/kg or 300 mg/kg). For normoxic studies, SCD mice were grouped into six groups of six mice each. Blood smears were made from tail snips before and after subcutaneous injection of S-NACH at various time points. Total plasma was harvested for cytokine analysis. Blood smears and plasma were obtained after 2 h from six untreated and 5-HMF-treated animals. For survival studies, SCD mice were treated with physiological PBS (n=6) or SNACH (n=8) by subcutaneous injection and subjected to hypoxia (5% O2) 30 min after the treatments and observed for 1.5 h. 534

Tissue factor pathway inhibitor and cytokine assays Plasma TFPI was measured using a kit from Neoscientific (Woburn, MA, USA). Cytokines in blood plasma were measured using commercial Bio-Plex beads in a Bio-Plex 200 system (BioRad Laboratories, Hercules, CA, USA), and analyzed using BioPlex manager software.

Morphological analysis Total blood was harvested from SCD animals in the presence of EDTA, treated with PBS, 5-HMF, or S-NACH and incubated under either normoxia or hypoxia (2% O2) at 37°C for 1 h. A blood smear from each sample was stained with Leishman stain and analyzed under an oil immersion light microscope.39

Statistical analyses Results are presented as the means ± standard deviation comparing experimental and control groups. A t-test was used for sta-

haematologica | 2022; 107(2)

Sulfated non-anticoagulant heparin in SCD treatment

Figure 2. Effect of S-NACH on HbS oxygen binding affinity. (A) The sulfated non-anticoagulant heparin derivative (S-NACH) increases hemoglobin oxygen affinity. Aliquots of hemolysates from the sickling assay were subjected to p50 analyses using the Hemox Analyzer. (B) Representative curves show a dose-dependent left shift, indicating an increase in oxygen affinity. Summarized data for biological replicates (n=5) are indicated in the graph. The findings confirm the primary direct anti-sickling mechanism of S-NACH.

tistical analyses, and results are considered statistically significant if P<0.05.

Results S-NACH modified intracellular HbS and reduced sickling of SS cells S-NACH was engineered to have an aldehyde moiety, which confers anti-sickling properties primarily due to specific interactions with HbS to increase its affinity for oxygen. We therefore tested the ability of S-NACH to modify HbS, increase oxygen affinity of HbS, and prevent RBC sickling. Our in vitro sickling assay under hypoxic conditions demonstrated that S-NACH, in a dose-dependent manner, significantly modified intracellular Hb (Figure 1A) and reduced the sickling of SS cells, with the maximum effect at the concentration of 2 mM, comparable to that of 1 mM GBT440 (Figure 1B; Table 1). This supports our hypothesis considering that two molecules were designed to target both N-terminal valine residues of the a globin in a tetrameric Hb molecule.

Levels of modified intracellular HbS translated into a dose-dependent increase in Hb oxygen affinity When aliquots of HbS-complex solution from the same studies were investigated in the oxygen equilibrium assay, we observed a similar concentration-dependent effect on increasing HbS affinity for oxygen, with a reduction in P50 values of about 55% at the highest concentration (65.7±3.2 at 2 mM) (Figure 2A; Table 1). These findings correlated linearly with the anti-sickling effects and degrees of HbS modifications, thus confirming this targeted mechanism of action (Figure 2B; Table 1).

S-NACH decreases in vivo red blood cell sickling and regulates inflammatory cytokines under normoxia When administered to C57/B mice, S-NACH caused an approximately 3-fold increase in plasma TFPI after 2 h of treatment (Figure 3A) at both doses tested. To determine the effect of S-NACH on RBC sickling, total blood from SCD mice was incubated at normoxia with S-NACH. Based on the lower effective dose with respect to TFPI release, the S-NACH dose for animal studies was set at 10 haematologica | 2022; 107(2)

mg/kg. 5-HMF (10 mg/kg) was used as a positive control because it decreases RBC sickling.33 Both S-NACH and 5-HMF moderately decreased the sickling of RBC by 35-50% (data not shown). Townes SCD mice treated with S-NACH showed a significant (*P<0.05) decrease in the percentage of circulating sickled RBC for up to 4 h, with a maximum decrease at 2 h after administration (50% to 35%) (Figure 3B, C) (n=6). Thus, S-NACH can decrease sickling of RBC under normoxia. Plasma samples (untreated, 5-HMF, 2 and 6 h after S-NACH treatment) were quantitatively analyzed for various pro-inflammatory mediators (interleukin [IL] 1b, IL-6, tumor necrosis factor-a [TNF-a]), anti-inflammatory mediators (IL-10, interferon g [IFN-g], monocyte chemoattractant protein 1 [MCP-1]), and growth factors (monocyte colony-stimulating factor [M-CSF], vascular endothelial growth factor [VEGF]) (Figure 4). Plasma levels of IL-1b, IL-6, IFN-g, MCP-1, TNF-a, M-CSF, and VEGF were increased in SCD untreated samples, whereas they were significantly decreased (P<0.0005) in S-NACH-treated samples, at both 2 and 6 h. Furthermore, S-NACH was able to increase the decreased levels of 1L-10. The regulatory effect of S-NACH was most effective at 2 h, similar to its effectiveness on sickled RBC morphology.

S-NACH decreases red blood cell sickling and prolongs survival of sickle cell disease mice under hypoxia In the Townes SCD mouse model, hypoxia increases RBC sickling, causing death within 15 min due to pulmonary sequestration of sickled RBC.26 We investigated the effect of S-NACH on RBC sickling and survival under hypoxia. Ex vivo deoxygenation was associated with increasing RBC sickling of up to 70%. In the presence of S-NACH, sickling was significantly (P<0.05) decreased to 30% (Figure 5). In the survival study, while all the untreated mice died within the first 15 min under hypoxia (5% O2), 50% of the S-NACH-treated mice were alive at 30 min, which was 1 h after S-NACH administration (Figure 6), a typical timespan used for investigating survival.33,40,41 Because S-NACH exhibited maximal effectiveness at 2 h, mice under hypoxia were observed for an additional 1 h, during which 37.5% of S-NACH-treated animals survived. Thus, SNACH increased the survival of SCD animals under hypoxia for up to 3 h. 535

O. Abdulmalik et al.

Figure 3. Effects of S-NACH under normoxia. (A) A sulfated non-anticoagulant heparin derivative (S-NACH) increases the release of endogenous tissue factor pathway inhibitor (TFPI). C57/B mice were treated with 100 and 300 mg/kg of free S-NACH, and plasma was obtained after 2 h. TFPI in plasma was measured in duplicate. TFPI levels were compared between S-NACH-treated samples and phosphate-buffered saline (PBS)-treated control samples, (n=4) (*P<0.05). (B) S-NACH treatment decreases sickling of red blood cells (RBC) in Townes sickle cell disease (SCD) mice. Blood smears were made from tail snips before and after subcutaneous injection of S-NACH (10 mg/kg) at the time points shown. 5-hydroxymethyl-2-furfural (5-MF) was used as a positive control (*P<0.05). (C) Morphology of the RBC from Townes SCD mice was examined in stained blood smears and expressed in percentage. RBC from four different fields or 120 cells were analyzed to calculate the percentage of sickled RBC. Blood from untreated samples contained higher percentages of sickled and distorted RBC (shown by arrows). S-NACH treatment decreased the presence of sickled RBC for up to 4 h with the greatest decrease seen at 2 h (n=6) (*P<0.05). SD: standard deviation.

Discussion We designed S-NACH, a modified low molecular weight heparin, to be devoid of anticoagulant properties, while acquiring new direct anti-sickling properties. In vitro, S-NACH directly modified intracellular HbS, increased oxygen affinity, and inhibited sickling of RBC under hypoxia. Additionally, S-NACH reduced the levels of circulating sickled cells in Townes SCD mice. We confirmed the in vitro release of endothelial-TFPI by S-NACH30 in C57/B mice, as demonstrated by a significant increase in plasma TFPI. Based on this, we speculate that S-NACH might exert local antithrombotic activity by increasing the concentration of endothelial TFPI in the vascular area. An in vivo increase in plasma TFPI levels after S-NACH administration confirms our earlier reported findings.26 According to Kemme et al., TFPI release increased by 3-fold (from 62.9 ng/mL to 237 ng/mL) after infusion of heparin.42 Kouta et al. reported a marked increase of TFPI release (~2.5-fold) within 20 min after intravenous administration of different types of heparins (bovine, ovine, and porcine) to non-human primates.43 Additionally, our observation with different species, including mice, rats, and rabbits (unpublished data) are consistent with these results. We demonstrated in vitro that S-NACH permeated RBC membranes to modify HbS and exert a significant antisickling effect by maintaining normal RBC morphology, protecting against the typical changes in RBC morphology 536

that occur in untreated samples from individuals with SCD. Although there was no prior evidence to indicate a relationship between RBC morphology and inflammatory mediators, the effect of both on decreased sickling and blood viscosity cannot be ruled out.6 Based on our previous studies, the observed reduction in the levels of proinflammatory cytokines was not unexpected. For example, in one previous study in an asthma-induced mouse model, S-NACH caused a robust reduction in airway eosinophilia, mucus production, and airway hyperresponsiveness even after chronic repeated challenges with allergen (ovalbumin).44 These effects were linked to suppression of Th2 cytokines IL-4/IL-5/IL-13/GM-CSF and upregulation of IL-10. The levels of these inflammatory cytokines increased around 2- to 8-fold (in both serum and bronchoalveolar lavage fluid) after induction with the allergen and decreased again to baseline after treatment with S-NACH. Similar observations were made for total white blood cell count, as well as eosinophil, macrophage, and lymphocyte counts, which were markedly elevated in the asthma-induced mouse model (6-, 4-, 1.5-, 1.5-, and 4-fold, respectively) after exposure to an allergen. SNACH also reduced lung fibrosis in mice that were chronically exposed to the allergen. As we showed in that study, the protective effects of S-NACH were attributable to modulation of the IL-4/JAK1 signal transduction pathway through inhibition of STAT6 phosphorylation and subsequent inhibition of GATA-3 and inducible nitric oxide synhaematologica | 2022; 107(2)

Sulfated non-anticoagulant heparin in SCD treatment

Figure 4. S-NACH treatment regulates the levels of inflammatory mediators. Total plasma from sickle cell disease (SCD) mice that were untreated, treated with sulfated non-anticoagulant heparin derivative (S-NACH; 10 mg/kg) or treated with 5-hydroxymethyl-2-furfural (5-MF) was harvested and frozen. Cytokines in blood plasma were measured in triplicate. S-NACH treatment significantly changed the plasma levels of the analytes (*P<0.0005). For most analytes, the effects of 5-HMF were comparable to those of S-NACH at 6 h. SD: standard deviation; IL: interleukin; TNF: tumor necrosis factor; IFN: interferon; MCP-1: monocyte chemoattractant protein 1; M-CSF: monocyte colony-stimulating factor; VEGF: vascular endothelial growth factor

thase expression. The protective effects of S-NACH treatment were associated with reductions of the basal expression of the two isoforms of arginase, ARG1 and ARG2, in lung epithelial cells.44 In another previous study, we measured the different biomarkers of inflammation in patients with SCD (35 patients with painful crises and 30 patients in steady state) in and 35 healthy donors. Plasma levels of several chemokines and cytokines including TNF-a, IL-1b, IL-6, IL-8, MCP-1, macrophage inflammatory protein 1a (MIP1a), and IFN-g in patients with SCD were distinctly and statistically significantly higher during painful crises and at steady state than in healthy donors (2- to 10-fold increases).45 The observed anti-sickling properties are in concordance with our expectations when compared to those produced by GBT440 (voxelotor, Oxybryta), which was recently approved for use by the Food and Drug Administration. There are some concerns that oxygen affinity-shifting strategies may be associated with different cerebrovascular risks,46 although this was adequately addressed by Estepp.47 Nonetheless, definitive reports on long-term use will provide conclusive information on this issue. With this in mind, our multimodal approach also incorporates polyanionic glycosaminoglycans such as heparins, which can be introduced into sickle RBC HbS by synthetic lipid vesicles. Once introduced, they would block sickling and also modulate ATPase activity and the charge of the RBC membrane in hypoxia.48 SCD occurs due to the replacement of an acidic, hydrophilic amino haematologica | 2022; 107(2)

acid (glutamic acid) in position 6 of the b chain of Hb with a non-polar, hydrophobic valine amino acid and this change causes a disturbance in Hb structure. We therefore speculate that S-NACH (polyanionic glycosaminoglycan) may reverse the polarity to make the HbS more soluble. This mechanism of action remains under investigation. We further speculate that S-NACH antagonizes hepcidin and might provide additional benefits in SCD by improving iron hemostasis, as suggested in a recent report.49 Additionally, our findings demonstrate that S-NACH plays a role similar to that of the non-anticoagulant heparin fractions from enoxaparin, which were shown to have an effect on inflammatory mediators.26,50 Furthermore, our thromboelastography assay results with S-NACH did not show any changes in platelet functions (data not shown). Indeed, S-NACH retains all the multimodal actions of the low molecular weight heparin tinzaparin but without systemic anticoagulation and its associated bleeding side effects. Tinzaparin demonstrated significant effects on the resolution of acute pain crises in patients with SCD in double-blind, randomized clinical trials.51 Overall, the effects of S-NACH on RBC sickling morphology, RBC adhesion, and regulation of inflammation resulted in increased survival of SCD mice under hypoxia. This study serves as a proof-of-concept that S-NACH is safe with respect to bleeding tendencies and argues for further detailed safety and efficacy studies in preclinical models of toxicity, the results of which would help guide and inform future human studies. 537

O. Abdulmalik et al.

Figure 5. S-NACH treatment decreases sickling of red blood cells ex vivo under hypoxia. Total blood harvested from sickle cell disease mice (n=8) was mixed with a sulfated non-anticoagulant heparin derivative (S-NACH) at the dose of 1, 5, or 10 mg/mL and incubated in 2% O2 at 37°C for 1 h. Blood smears were made, stained, and the morphology of red blood cells (RBC) was analyzed. Hypoxia increased the percentage of sickled RBC. S-NACH treatment decreased the sickling of RBC in a dose-dependent manner. Phosphate-buffered saline (PBS) was used as a negative control and 5-hydroxymethyl-2-furfural (5-MF) was used as a positive control. *P<0.05. SD: standard deviation

Figure 6. S-NACH treatment increases the survival of sickle cell disease mice under hypoxia. Sickle cell disease (SCD) mice were treated with phosphate-buffered saline (PBS, n=6) or 10 mg/kg sulfated non-anticoagulant heparin derivative (S-NACH, n=8). After 30 min, mice were incubated in a hypoxia chamber (5% O2), and the survival of animals was observed for 1.5 h. Surviving mice were euthanized, as per the guidelines. S-NACH treatment was associated with increased survival of mice.

S-NACH increased the levels of TFPI in plasma, decreased RBC sickling under normoxia and hypoxia, and reduced the levels of the pro-inflammatory mediators IL-1, IL-6, IFN-g, MCP-1, TNF-a, M-CSF, and VEGF while increasing anti-inflammatory factors such as IL-10, further establishing it as a promising bona fide multimodal candidate drug worthy of additional investigations for acute and chronic disease management in SCD patients. In sum538

mary, our data demonstrate direct and support pleiotropic effects of S-NACH in ameliorating the complex pathophysiological mechanisms involved in SCD. Development into an effective drug would lead to improved outcome in patients globally with SCD, considering the current limited therapeutic options, especially for the vast majority of patients with SCD who reside in underdeveloped areas of the world.52 haematologica | 2022; 107(2)

Sulfated non-anticoagulant heparin in SCD treatment

Disclosures SAM holds a US patent on S-NACH28 and other related US patents. None of the authors has any conflicts of interest.

Acknowledgments We appreciate Dr. Kelly A. Keating, Pharmaceutical Research Institute (PRI), for her excellent editing of this manuscript.

Contributions SAM designed the study and is the Principal Investigator; OA, NHED, VM-C and MAT conducted the experiment; OA did the data analysis; OA and NHED contributed equally to the manuscript write up and data interpretation.All authors have approved the final version of the manuscript.

Funding This project was funded by Vascular Vison Pharmaceuticals Co. to PRI and an NIH STTR Phase 1 grant (1R41HL14773701-A1, NIH/NHLB “Multi-modal Mechanisms of Novel Sulfated Non-Anticoagulant Heparin (S-NACH) in Sickle Cell Disease Management”) subaward to Children’s Hospital of Philadelphia (CHOP), university of Pennsylvania.

References 1. Ilesanmi OO. Pathological basis of symptoms and crises in sickle cell disorder: implications for counseling and psychotherapy. Hematol Rep. 2010;2(1):e2. 2. Lu L, Li X, Vekilov PG, Karniadakis GE. Probing the twisted structure of sickle hemoglobin fibers via particle simulations. Biophys J. 2016;110(9):2085-2093. 3. Telen MJ. Beyond hydroxyurea: new and old drugs in the pipeline for sickle cell disease. Blood. 2016;127(7):810-819. 4. Connes P, Alexy T, Detterich J, Romana M, Hardy-Dessources MD, Ballas SK. The role of blood rheology in sickle cell disease. Blood Rev. 2016;30(2):111-118. 5. Hebbel RP, Eaton JW, Steinberg MH, White JG. Erythrocyte/endothelial interactions and the vasocclusive severity of sickle cell disease. Progr Clin Biol Res. 1981;55:145162. 6. Kaul DK, Fabry ME, Costantini F, Rubin EM, Nagel RL. In vivo demonstration of red cell-endothelial interaction, sickling and altered microvascular response to oxygen in the sickle transgenic mouse. J Clin Invest. 1995;96(6):2845-2853. 7. Noguchi CT, Schechter AN, Rodgers GP. Sickle cell disease pathophysiology. Baillieres Clin Haematol. 1993;6(1):57-91. 8. de Jong K, Larkin SK, Styles LA, Bookchin RM, Kuypers FA. Characterization of the phosphatidylserine-exposing subpopulation of sickle cells. Blood. 2001;98(3):860867. 9. Wautier MP, Heron E, Picot J, Colin Y, Hermine O, Wautier JL. Red blood cell phosphatidylserine exposure is responsible for increased erythrocyte adhesion to endothelium in central retinal vein occlusion. J Thromb Haemost. 2011;9(5):1049-1055. 10. Setty BNY, Kulkarni S, Stuart MJ. Role of erythrocyte phosphatidylserine in sickle red cell–endothelial adhesion. Blood. 2002;99(5):1564-1571. 11. Proenca-Ferreira R, Brugnerotto AF, Garrido VT, et al. Endothelial activation by platelets from sickle cell anemia patients. PLoS One. 2014;9(2):e89012. 12. Belcher JD, Marker PH, Weber JP, Hebbel RP, Vercellotti GM. Activated monocytes in sickle cell disease: potential role in the activation of vascular endothelium and vasoocclusion. Blood. 2000;96(7):2451-2459. 13. Ataga KI, Key NS. Hypercoagulability in sickle cell disease: new approaches to an old problem. Hematology Am Soc Hematol Educ Program. 2007;91-96. 14. FDA approves new treatment for sickle cell disease. 2017 [accessed September 11, 2020]; Available from: https:// w w w. f d a . g o v / n e w s - e v e n t s / p r e s s announcements/fda-approves-new-treat-

haematologica | 2022; 107(2)

ment-sickle-cell-disease 15. Lee JO, Lee JY, Chun EJ, et al. Incidence and predictors of venous thromboembolism in medically ill hospitalized elderly cancer patients: a prospective observational study. Support Care Cancer. 2019;27(7):25072515. 16. FDA approves voxelotor for sickle cell disease. 2019 [accessed November 25, 2019]; Available from: https://www.fda.gov/ drugs/resources-information-approveddrugs/fda-approves-voxelotor-sickle-celldisease 17. Matte A, Zorzi F, Mazzi F, Federti E, Olivieri O, De Franceschi L. New therapeutic options for the treatment of sickle cell disease. Mediterr J Hematol Infect Dis. 2019;11(1):e2019002. 18. 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. 19. Oksenberg D, Dufu K, Patel MP, et al. GBT 440 increases haemoglobin oxygen affinity, reduces sickling and prolongs RBC half-life in a murine model of sickle cell disease. Br J Haematol. 2016;175(1):141-153. 20. Hutchaleelaha A, Patel M, Washington C, et al. Pharmacokinetics and pharmacodynamics of voxelotor (GBT440) in healthy adults and patients with sickle cell disease. Br J Clin Pharmacol. 2019;85(6):1290-1302. 21. Metcalf B, Chuang C, Dufu K, et al. Discovery of GBT440, an orally bioavailable R-state stabilizer of sickle cell hemoglobin. ACS Med Chem Lett. 2017; 8(3):321-326. 22. Khandros E, Huang P, Peslak SA, et al. Understanding heterogeneity of fetal hemoglobin induction through comparative analysis of F and A erythroblasts. Blood. 2020;135(22):1957-1968. 23. Green NS, Barral S. Emerging science of hydroxyurea therapy for pediatric sickle cell disease. Pediatr Res. 2014;75(1-2):196204. 24. Charneski L, Congdon HB. Effects of antiplatelet and anticoagulant medications on the vasoocclusive and thrombotic complications of sickle cell disease: a review of the literature. Am J Health Syst Pharm. 2010;67(11):895-900. 25. Garcia DA, Baglin TP, Weitz JI, Samama MM. Parenteral anticoagulants: antithrombotic therapy and prevention of thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest. 2012;141(2 Suppl):e24S43S. 26. Mousa SA, Linhardt R, Francis JL, Amirkhosravi A. Anti-metastatic effect of a non-anticoagulant low-molecular-weight heparin versus the standard low-molecularweight heparin, enoxaparin. Thromb

Haemost. 2006;96(6):816-821. 27. Alyahya R, Sudha T, Racz M, Stain SC, Mousa SA. Anti-metastasis efficacy and safety of non-anticoagulant heparin derivative versus low molecular weight heparin in surgical pancreatic cancer models. Int J Oncol. 2015;46(3):1225-1231. 28. Mousa SA, inventor Oxidized heparin fractions and their use in inhibiting angiogenesis. US patent no. 8,071,569. 2011 Dec 6. 29. Sudha T, Phillips P, Kanaan C, Linhardt RJ, Borsig L, Mousa SA. Inhibitory effect of non-anticoagulant heparin (S-NACH) on pancreatic cancer cell adhesion and metastasis in human umbilical cord vessel segment and in mouse model. Clin Exp Metastastis. 2012;29(5):431-439. 30. Alshaiban A, Muralidharan-Chari V, Nepo A, Mousa SA. Modulation of sickle red blood cell adhesion and its associated changes in biomarkers by sulfated nonanticoagulant heparin derivative. Clin Applied Thromb Hemost. 2016;22(3):230-238. 31. Petaja J. Inflammation and coagulation. An overview. Thromb Res. 2011;127(Suppl 2):S34-37. 32. Hijiya N, Horiuchi K, Asakura T. Morphology of sickle cells produced in solutions of varying osmolarities. J Lab Clin Med. 1991;117(1):60-66. 33. Abdulmalik O, Safo MK, Chen Q, et al. 5hydroxymethyl-2-furfural modifies intracellular sickle haemoglobin and inhibits sickling of red blood cells. Br J Haematol. 2005;128(4):552-561. 34. Asakura T, Mayberry J. Relationship between morphologic characteristics of sickle cells and method of deoxygenation. J Lab Clin Med. 1984;104(6):987-994. 35. Horiuchi K, Ohata J, Hirano Y, Asakura T. Morphologic studies of sickle erythrocytes by image analysis. J Lab Clin Med. 1990;115(5):613-620. 36. Abdulmalik O, Ghatge MS, Musayev FN, et al. Crystallographic analysis of human hemoglobin elucidates the structural basis of the potent and dual antisickling activity of pyridyl derivatives of vanillin. Acta Crystallogr D Biol Crystallogr. 2011;67(Pt 11):920-928. 37. Abdulmalik O, Safo MK, Lerner NB, et al. Characterization of hemoglobin bassett (a94Asp→ Ala), a variant with very low oxygen affinity. Am J Hematol. 2004; 77(3):268-276. 38. Abdulmalik O, Safo MK, Seeholzer SH, Asakura T, Hasbrouck NC, Russell JE. Hb Baden: structural and functional characterization. Am J Hematol. 2010;85(11):848852. 39. Leishman WB. Note on a simple and rapid method of producing Romanowsky staining in malarial and other blood films. Br Med J. 1901;2(2125):757-758.

539

O. Abdulmalik et al. 40. Iyamu EW, Turner EA, Asakura T. Niprisan (Nix-0699) improves the survival rates of transgenic sickle cell mice under acute severe hypoxic conditions. Br J Haematol. 2003;122(6):1001-1008. 41. Zhang C, Li X, Lian L, et al. Anti-sickling effect of MX-1520, a prodrug of vanillin: an in vivo study using rodents. Br J Haematol. 2004;125(6):788-795. 42. Kemme MJ, Burggraaf J, Schoemaker RC, Kluft C, Cohen AF. Quantification of heparin-induced TFPI release: a maximum release at low heparin dose. Br J Clin Pharmacol. 2002;54(6):627-634. 43. Kouta A, Hoppensteadt D, Bontekoe E, et al. Studies on tissue factor pathway inhibitor antigen release by bovine, ovine and porcine heparins following intravenous administration to non-human primates. Clin Appl Thromb Hemost. 2020; 26:1076029620951851.

540

44. Ghonim MA, Wang J, Ibba SV, et al. Sulfated non-anticoagulant heparin blocks Th2-induced asthma by modulating the IL4/signal transducer and activator of transcription 6/Janus kinase 1 pathway. J Trans Med. 2018;16(1):243. 45. Qari MH, Dier U, Mousa SA. Biomarkers of inflammation, growth factor, and coagulation activation in patients with sickle cell disease. Clin Appl Thromb Hemost. 2012;18(2):195-200. 46. Hebbel RP, Hedlund BE. Sickle hemoglobin oxygen affinity-shifting strategies have unequal cerebrovascular risks. Am J Hematol. 2018;93(3):321-325. 47. Estepp JH. Voxelotor (GBT440), a first-inclass hemoglobin oxygen-affinity modulator, has promising and reassuring preclinical and clinical data. Am J Hematol. 2018;93(3):326-329. 48. Winter WP, Seale WR, Yodh J. Interaction

of hemoglobin S with anionic polysaccharides. Am J Pediatr Hematol Oncol. Spring 1984;6(1):77-81. 49. Mohanty P, Jena RK, Sethy S. Variability of iron load in patients of sickle cell anaemia (HbSS): a study from Eastern India. J Clin Diagn Res. 2017;11(3):Ec19-ec22. 50. Shastri MD, Stewart N, Horne J, et al. Nonanticoagulant fractions of enoxaparin suppress inflammatory cytokine release from peripheral blood mononuclear cells of allergic asthmatic individuals. PLoS One. 2015;10(6):e0128803. 51. Qari MH, Aljaouni SK, Alardawi MS, et al. Reduction of painful vaso-occlusive crisis of sickle cell anaemia by tinzaparin in a double-blind randomized trial. Thromb Haemost. 2007;98(2):392-396. 52. Sankaran VG, Weiss MJ. Anemia: progress in molecular mechanisms and therapies. Nat Med. 2015;21(3):221-230.

haematologica | 2022; 107(2)

LETTERS TO THE EDITOR SARS-CoV-2 infection in aplastic anemia Coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus-2 (SARSCoV-2), was declared a pandemic by the World Health Organization in March 2020. Compared to patients with non-hematologic cancers, patients affected by hematologic disorders have increased mortality and more prolonged viral RNA persistence.1-3 Since the early phase of the pandemic, several groups have described thrombocytopenia or secondary hemophagocytic lymphohistiocytosis in patients infected by SARS-CoV-2, problems likely due to a cytokine storm and the potential cytotoxicity of the virus.4,5 Aplastic anemia (AA), a rare autoimmune disease with an incidence of two cases per million population, is characterized by cytopenia and bone marrow hypocellularity.6 It has been proposed that, in acquired AA, an initiating event provokes an aberrant immune response, triggering oligoclonal expansion of cytotoxic T cells that destroy hematopoietic stem cells. The consequences of SARS-CoV-2 infection in known cases of AA are not clear.7 Additionally, it is unknown whether this virus can trigger an aberrant immune response leading to depletion of the stem cell compartment and inducing bone marrow failure. Here we describe the features and clinical outcome of a group of patients affected with AA and SARS-CoV-2 infection between April 2020 and January 2021. A national survey was launched in April 2020 to assess the clinical features and outcome of patients with preexisting AA and new onset AA after SARS-CoV-2 infection. The criteria for diagnosing AA and classifying its severity were described previously by Camitta et al.8 The diagnosis of SARS-CoV-2 infection was confirmed by nasopharyngeal swab9 at the onset of symptoms or at access to the hematology department. The study population consisted of 23 patients with AA (30% with very severe AA, 26% with severe AA and 43% with non-severe AA) with a median age of 49 years (range, 20–77); there were seven females and 16 males. All cases were acquired, except one with Fanconi anemia. A subclinical paroxysmal nocturnal hemoglobinuria clone was present in five cases. None of the patients was vaccinated against SARS-CoV-2. At the onset of SARS-CoV-2 infection, 60% (14/23) of the patients were on active immunosuppressive therapy – six on high-dose cyclosporine maintenance treatment after horse antithymocyte globulin, one on eltrombopag and cyclosporine and seven on a combination of cyclosporine and mycophenolate mofetil – as part of graft-versus-host disease prophylaxis after a reducedintensity allograft. Table 1 summarizes the populations' demographic and allogeneic stem cell transplant details. The most common symptoms were fatigue, general malaise, fever, dry cough, shortness of breath, loss of smell, and diarrhea; 29% (7/23) of patients who developed a COVID-19-defining event (6 pneumonia and 1 hepatitis) were hospitalized (median 5 days; range, 3-12). Within this subgroup, three patients required oxygen supplementation, of whom two needed escalation to intensive care unit admission for high-flow oxygen and monitoring, but none required mechanical ventilation. At diagnosis of the infection, median blood count parameters showed pancytopenia: white blood cells 2.3x109/L (range, 0.42–5.1; interquartile range [IQR], 0.97), neutrophils 1.08x109/L (range, 0.14–2.56; IQR, 0.68), hemoglobin 93.5 g/L (range, 74–139; 25th perhaematologica | 2022; 107(2)

centile 82, 75th percentile 101) and platelets) 35x109/L (range, 2-121; IQR, 42). None developed evidence of secondary hemophagocytic lymphohistiocytosis. Upon review of blood results prior to the SARS-CoV-2 infection, it was possible to appreciate a progressive decline in all hematologic indices consistent with overt relapse (confirmed by bone marrow hypocellularity meeting diagnostic criteria) in two patients and, although not meeting relapse criteria, requiring treatment, intense monitoring, and transfusion support in 15 patients. Interestingly, three cases (12.5%) of idiopathic AA were diagnosed a few weeks after documented SARSCoV-2 infection. Blood counts performed in the immediate past for other medical reasons showed normal para meters in all these three patients, who developed severe or very severe AA with heavy transfusion dependency, and eventually required treatment with immunosuppressive therapy or hematopoietic stem cell transplantation (Table 1). At the time of reporting, all three patients are in remission with good hematologic response. Figure 1 shows the median values of white blood cell Table 1. Characteristics of the patients with aplastic anemia at the time of infection with severe acute respiratory syndrome coronavirus-2.

AA disease characteristics at SARS-Cov-2 infection Number of patients (n, %) Female Male Age in years, median [range] Disease category, n (%) Very severe AA Severe AA Non-severe AA Disease status (n, %) New onset/diagnosis In remission On treatment On CSA after hATG Eltrompopag Others* Post-HSCT** on IST

SARS-CoV2 features Severity Mild Moderate Severe Oxygen supplementation Intensive care admission AA status after SARS-CoV2 New onset AA Relapse of AA Decline in hematologic indices Outcome of AA Death New treatment IST HSCT

N (%) or median [range] 23 (100) 16 (70) 7 (30) 49 [20-77] 7 (30) 6 (26) 10 (43) 3 (13) 14 (60) 7 (30) 6 (26) 1 (4) 6 (26) 7 (30)

N (%) 13 (57) 7 (30) 3 (13) 3 (13) 2 (8) 3 (13) 1 (4) 15 (65) 1 (4) 4 (17) 3 (13) 1 (4)

*Others: included patients who never required treatment for aplastic anemia (AA) and also patients whose cyclosporine was successfully withdrawn after they had achieved remission of their AA. **Matched unrelated (n=3), matched sibling (n=2), mismatched unrelated (n=1), and haploidentical (n=1); this group includes patients who underwent transplantation either upfront or at failure of immunosuppressive therapy. AA: aplastic anemia; SARS-CoV-2: severe acute respiratory syndrome coronavirus-2; CSA: cyclosporine A; hATG: horse antithymocyte globulin; HSCT: hematopoietic stem cell transplant, IST: immunosuppressive therapy.

541

Figure 1. Hematologic parameters in patients with aplastic anemia and SARS-CoV-2 infection. Median values of white blood cell count (x109/L), hemoglobin (g/L), and platelet count (x109/L) at three different time-points (before, during and after SARS-CoV2 infection) in the four groups of the study population: (newly diagnosed aplastic anemia [AA], AA on active immunosuppressive therapy [IST], AA off IST and after hematopoietic stem cell transplantation [post-HSCT]). Note complete data for some of the blood parameters were not available in four cases, which were, therefore, excluded from the graphical illustration.

Letters to the Editor

542

haematologica | 2022; 107(2)

Letters to the Editor

count, hemoglobin concentration and platelet count at three different time-points (pre-infection, at infection, and post-infection) in the four groups of the study population: newly diagnosed, on active immunosuppressive therapy, off immunosuppressive therapy and after hematopoietic stem cell transplantation. For those patients affected by transfusion-independent non-severe AA (10/23), there was a new requirement of transfusion support in seven patients, but no cases of transition to severe/very severe AA were recorded. Despite profound neutropenia and being on immunosuppressive therapy, only 13% of patients (3/23) developed COVID-19. This may reflect some specific, favorable host factors (such as young age) or might be secondary to protective immune dysregulation known to be present in AA.10 Indeed, hypotheses on the role of hyperinflammation resulting in a more severe disease phenotype have resulted in proposals of trials to investigate the use of agents blocking these pathways for the treatment of severe SARS-CoV-2 infection in non-AA patients.11 Despite the lack of cytokine studies or viral polymerase chain reaction analysis of bone marrow aspirates in our study, it is reasonable to speculate a potential myelosuppressive effect of SARS-CoV-2: as demonstrated in Figure 1, patients had a clear decline in hematopoiesis, causing worsening of blood parameters and relapse of AA. However, the study does not clarify whether the virus has a direct cytotoxic effect on hematopoietic stem cells or acts through the cytokine storm or aberrant immune dysregulation following the infection and might have a bias due to non-reporting of milder cases. We demonstrate that SARS-CoV-2 infection is another factor that can jeopardize residual hematopoiesis during AA, as previously described for other viral infections (e.g., hepatitis). The kinetics of the deterioration in blood counts after SARS-CoV-2 infection mirrors the previously reported kinetics of AA diagnosis or relapse in pregnancy. Although a clear correlation between pregnancy and the onset or relapse of AA has never been demonstrated, several groups have described worsening of hematologic indices at the onset of pregnancy and subsequent recovery in the post-partum period.12,13 Our study does not enable clear conclusions to be drawn about the severity and long-term prognosis of SARS-CoV-2 infection in AA; despite the lack of COVID19 deaths, the viral infection was a risk factor for the onset of AA and for worsening of blood parameters in patients already with AA. This is the first report describing the outcome of AA following SARS-CoV-2 infection, and while it is encouraging to note that most patients (including transplanted cases) made a full recovery without the development of significant symptoms, this population needs to be considered at risk of complications of worsening cytopenias following COVID-19. Indeed, one patient died as a consequence of infectious complications due to relapsed AA. A possible temporal relationship between SARS-CoV2 infection and AA can be suggested in three cases in our series. Are these cases a casual association of SARS-CoV2 infection and AA or are they cases of secondary AA as a result of the viral insult? The detection of three new cases of AA within a total of 4.5 million cases of SARSCoV-2 infection in the UK is intriguing. Further studies that include measurement of cytokines and other factors such as regulatory T-cell subsets are needed to characterize the immune and inflammatory environment following SARS-CoV-2 infections in AA patients to help predict outcomes and prognosis. Furthermore, considering the availability of vaccines against SARS-CoV2 infection, it is haematologica | 2022; 107(2)

important to prevent cytopathic effects of the virus with a successful AA vaccination program, although close monitoring is required as vaccination-induced AA has been reported in the literature. Daniele Avenoso,1 Judith C.W. Marsh,1 Victoria Potter,1 Antonio Pagliuca,1 Simon Slade,1 Fiona Dignan,2 Eleni Tholouli,2 Sajjan Mittal,3 Bernard Davis,4 Sudhir Tauro,5 Rachel Kesse-Adu,6 Morag Griffin,7 Elspeth Payne,8 Shreyans Gandhi1 and Austin G. Kulasekararaj1 1 King’s College Hospital NHS Foundation Trust, London; 2Royal Manchester Infirmary, Manchester; 3Northampton General Hospital, Northampton; 4Whittington Hospital, London; 5University of Dundee, Dundee; 6Guys and St Thomas Hospital NHS Foundation Trust, London; 7St James University Hospitals, Leeds and 8University College London, London, UK Correspondence: DANIELE AVENOSO - d.avenoso@nhs.net doi:10.3324/haematol.2021.279928 Received: September 2, 2021. Accepted: October 13, 2021. Pre-published: October 21, 2021. Disclosures: no conflicts of interest to disclose. Contributions: all the authors were involved in the care of the patients and contributed equally to writing this manuscript; DA collected and analyzed the data; AK supervised the study. Acknowledgments: the authors thank all NHS staff for fighting the COVID-19 pandemic

References 1. Zhu N, Zhang D, Wang W, et al. A novel coronavirus from patients with pneumonia in China, 2019 N Engl J Med. 2020;382(8):727-733. 2. Yang X, Yu Y, Xu J, et al. Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: a singlecentered, retrospective, observational study. Lancet Respir Med. 2020;8(5):475-481. 3. Shah V, Ko Ko T, Zuckerman M, et al. Poor outcome and prolonged persistence of SARS-CoV-2 RNA in COVID-19 patients with haematological malignancies; King’s College Hospital experience. Br J Haematol. 2020;190(5):e279-e282. 4. Dewaele K, Claeys R. Hemophagocytic lymphohistiocytosis in SARS-CoV-2 infection. Blood. 2020;135(25):2323. 5. Hersby DS, Do TH, Gang AO, Nielsen TH. COVID-19-associated pancytopenia can be self-limiting and does not necessarily warrant bone marrow biopsy for the purposes of SARS-CoV-2 diagnostics. Ann Oncol. 2021;32(1):121-123. 6. Young NS. Aplastic anemia. N Engl J Med. 2018;379(17):1643-1656. 7. Paton C, Mathews L, Groarke EM, et al. COVID-19 infection in patients with severe aplastic anaemia. Br J Haematol. 2021;193(5):902-905. 8. Camitta BM, Storb R, Thomas ED. Aplastic anemia: pathogenesis, diagnosis, treatment, and prognosis. N Engl J Med. 1982;306(11):645-652. 9. Corman VM, Landt O, Kaiser M, et al. Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT-PCR. Eurosurveillance. 2020;25(3):2000045. 10. Dufour C, Ferretti E, Bagnasco F, et al. Changes in cytokine profile pre- and post-immunosuppression in acquired aplastic anemia. Haematologica. 2009;94(12):1743-1747. 11. Biran N, Ip A, Ahn J, et al. Tocilizumab among patients with COVID-19 in the intensive care unit: a multicentre observational study. Lancet Rheumatol. 2020;2(10):e603-e612. 12. Aitchison RGM, Marsh JCW, Hows JM, Russell NH, Gordon-Smith EC. Pregnancy associated aplastic anaemia: a report of five cases and review of current management. Br J Haematol. 1989;73(4):541-545. 13. Oosterkamp HM, Brand A, Kluin-Nelemans JC, Vandenbroucke JR. Pregnancy and severe aplastic anaemia: causal relation or coincidence? Br J Haematol. 1998;103(2):315-316.

543

Letters to the Editor

The insecticides permethrin and chlorpyrifos show limited genotoxicity and no leukemogenic potential in human and murine hematopoietic stem progenitor cells Epidemiological and clinical studies have revealed that maternal exposure to pesticides-insecticides during pregnancy is associated with an increased risk of infant or childhood acute leukemia.1-3 The household insecticides permethrin and chlorpyrifos, which are members of the pyrethroid and organophosphate families of pesticides, respectively, have been associated with both the induction of MLL rearrangements (MLLr) and the development of infant acute leukemia.4-9 Despite the epidemiological association between insecticides and increased risk of leukemia, whether such insecticides act as topo isomerase II DNA-damaging poisons remains unknown and studies testing the biological plausibility of such an association are lacking. Here, we assessed the genotoxicity, induction of MLLr and leukemogenic potential of permethrin and chlorpyrifos by taking advantage of cutting-edge in vitro and in vivo models using prenatal, neonatal and adult hematopoietic stem and progenitor cells (HSPC). Our data suggest that the insecticides tested show no cytotoxicity, limited genotoxicity and no leukemogenic potential in human and murine HSPC in vitro and in vivo. We initially assessed whether acute exposure to etoposide (1 µM), permethrin or chlorpyrifos (10 µM or 50 µM) induces MLL breaks in undifferentiated human embryonic stem cells (hESC) and CD34+ HSPC derived from hESC, cord blood or adult peripheral blood (Figure 1A). Acute exposure (24 h) to either permethrin or chlorpyrifos consistently induced MLL breaks in 3–7% of embryonic, neonatal and adult CD34+ cells in a doseindependent manner (Figure 1B, C). Embryonic and somatic CD34+ cells were sensitive to the three treatments (Figure 1C). Of note, whereas embryonic and neonatal CD34+ cells were slightly more sensitive to etoposide than to insecticides (Figure 1C), no differences were found for adult CD34+ cells, suggesting that the potential genotoxicity of the insecticides may have a more relevant etiological impact in adult MLLr acute leukemia. Chronic exposure to low doses of etoposide has been reported to induce apoptosis, MLL breaks and major chromosomal abnormalities in hESC.10 To test whether this occurred after exposure to insecticides, we treated hESC for 24 h with 10 µM of either permethrin or chlorpyrifos followed by a daily “booster” dose (2 µM) for 40 days (Figure 1D). After 5 days of recovery (without treatments), MLLr breaks and gross genomic abnormalities were assayed by interphase fluorescence in situ hybridization (iFISH), G-banding and comparative genomic hybridization (Figure 1E-G). In contrast to the frequency of MLL breaks (3–7%) observed after acute exposure (Figure 1C), MLL breaks were scarcely detectable upon chronic exposure to permethrin or chlorpyrifos, suggesting a legitimate repair of the DNA damage/double strand breaks (DSB) at the MLL locus (Figure 1E). Likewise, karyotyping and comparative genomic hybridization analysis revealed no numerical or structural chromosomal alterations (Figure 1F) or DNA gains or losses after chronic exposure (Figure 1G). It has been previously suggested that the chromosomal topology and chromatin structure resulting from early apoptosis may represent the underlying substrate for MLL chromosomal translocations to occur.11 Here, in contrast 544

to etoposide, which did cause significant cell death at doses as low as 1 µM, neither permethrin nor chlorpyrifos induced cell death/apoptosis at doses 50-fold higher (Online Supplementary Figure S1A), supporting the concept that the absence of high-grade DNA fragmentation may represent a chromatin physical impediment for MLL DSB to fuse in-frame with a partner gene and encode an oncogenic fusion protein. Inverse polymerase chain reaction assays confirmed the absence of in-frame MLL fusions (data not shown). Overall, these data indicate that chronic exposure to insecticides neither enriches for MLL breaks nor generates MLL fusion oncogenes or gross genomic instability. To test the bona fide ability of permethrin and chlorpyrifos to function as topoisomerase II poisons, and their ability to generate DNA-DSB we performed an in vivo complex of enzyme assays to analyze covalent genomic DNA/topoisomerase II complexes in live cells.12 Treatment of hESC with permethrin and chlorpyrifos at 10–500 µM induced only minimal poisoning of the topoisomerase II isoforms (a and b), whereas as little as 1 µM etoposide induced significant poisoning of both isoforms (Online Supplementary Figure S1B). Similarly, monitoring of g-H2AX by western blotting (Online Supplementary Figure S1C) or fluorescent activated cell sorting (FACS) analysis (Online Supplementary Figure S1D-F) revealed that the pesticides were unable to induce DSB in hESC, neonatal or adult CD34+ cells at relatively high concentrations, whereas etoposide potently induced DSB that were slightly repaired to some extent over time in somatic CD34+ cells but not in hESC (Online Supplementary Figure S1F). Overall, the data indicate that permethrin and chlorpyrifos are not topoisomerase II poisons and do not generate DNA-DSB in embryonic or somatic CD34+ cells. We next sought to determine whether prolonged in vivo exposure of human CD34+ HSPC to pesticides induces MLL breaks or initiates leukemia (Figure 2A). Immunodeficient (NSG) mice xenotransplanted with cord blood-derived human CD34+ HSPC were exposed for 12 weeks to permethrin or chlorpyrifos in the drinking water to mimic the environmental exposure in humans. Our target concentration for pesticide exposure was 10 mg/kg/day, and 5 mg/kg/day for etoposide. As the actual concentration to which mice are exposed will depend on their weight and their consumption of drinking water over the experimental period, we measured both parameters weekly. Results revealed an actual intake of permethrin, chlorpyrifos and etoposide only 10–20% lower than the theoretical concentrations (8.5, 9 and 4.2 mg/kg/day, respectively) (Online Supplementary Figure S2A-C). Importantly, the permethrin and chlorpyrifos metabolites 3-BPA and TCPy were readily detected by gas chromatography-mass spectrometry in serum and urine as soon as 48 h after pesticide exposure, confirming consistent exposure through administration of drinking water (Online Supplementary Figure S2D). iFISH analysis at sacrifice revealed a small but significant increase in the frequency of bone marrow-engrafted human CD45+ cells harboring MLL breaks in etoposidetreated mice, but not in permethrin- or chlorpyrifostreated mice (Figure 2B). FACS analysis at sacrifice revealed similar levels of human graft and normal multilineage (immature, myeloid and B-cell lymphoid) engraftment in the bone marriow and peripheral blood of etoposide-, permethrin- and chlorpyrifos-treated mice (Figure 2C, Online Supplementary Figure S2E), with no evidence of splenomegaly (Figure 2D). These findings indicate that chronic in vivo exposure to the indicated haematologica | 2022; 107(2)

Letters to the Editor

Figure 1. In vitro induction of MLL rearrangements in embryonic, neonatal and adult human CD34+ hematopoietic stem and progenitor cells following acute and chronic exposure to etoposide, permethrin and chlorpyrifos. (A) Experimental design to assess the induction of MLL breaks in human undifferentiated embryonic stem cells (hESC) and CD34+ hematopoietic stem and progenitor cells (HSPC) after 24 h single-pulse exposure to the indicated treatments (etoposide, ETO; permethrin, PER; and chlorpyrifos, CPF). (B) Representative interphase fluorescence in situ hybridization (iFISH) images showing MLL germline and rearranged (MLLr) human CD34+ cells. (C) Frequency of MLL breaks in undifferentiated hESC and embryonic, neonatal and adult CD34+ HSPC after 24 h single-pulse exposure to the indicated treatments (n=3 independent experiments for each cell type). Asterisks indicate statistically significant differences of a given treatment as compared with dimethylsulfoxide (DMSO, vehicle treatment) *P<0.05, **P<0.01. Dotted lines in the graphs show the percentage of MLL breaks in the DMSO-treated control groups. A minimum of 500 nuclei were analyzed except in some samples for which 80-400 nuclei were analyzed. (D) Experimental design to analyze the frequency of MLL breaks and gross chromosomal damage in hESC after continuous exposure to PER and CPF. (E) Frequency of MLL breaks detected by iFISH 45 days after chronic treatment with either PER or CPF (n=3 independent experiments). A minimum of 400 nuclei was analyzed per experiment. (F) Representative image of a G-banding karyotype 45 days after chronic treatment with either PER, CPF or DMSO (n=3). (G) Representative image of DNA copy number variation profiling by comparative genomic hybridization array analysis 45 days after chronic treatment with either PER or CPF. CGH: comparative genomic hybridization; AMP: amplification; DEL: deletion; LOH: loss of heterozygosity.

haematologica | 2022; 107(2)

545

Letters to the Editor

doses of etoposide or the pesticides fails to induce MLL breaks or to initiate leukemia in NSG-reconstituting human CD34+ HSPC. We next assessed whether chronic exposure to pesticides during embryonic development induces Mll breaks in developing murine HSPC. To do this, CD1 male and female mice were mated and pregnant CD1 females were exposed to etoposide (10 mg/kg/day) or to pesti-

cides (20 mg/kg/day) in drinking water from gestational day 0.5 to day 21 (Figure 3A). No significant differences were found between treatment groups for the number of pups per litter at birth (range, 11–18) or sex distribution (Online Supplementary Figure S3A, B). The mothers and one-half of the litter were euthanized at weaning to analyze the impact of etoposide and pesticides on the Mll locus and on hematopoietic homeostasis, and the

C D

Figure 2. Continuous exposure to permethrin or chlorpyrifos fails to induce MLL breaks or leukemia in NSG-reconstituting human CD34+ hematopoietic stem and progenitor cells. (A) Experimental design to determine whether prolonged in vivo exposure of human CD34+ hematopoietic stem and progenitor cells (HSPC) to insecticides induces MLL breaks or can initiate leukemia. In brief, 32 NSG mice were irradiated and cord blood-derived human CD34+ cells were transplanted into the bone marrow 6–8 h later. Four days later, mice were divided into four groups (8 mice/group) to initiate exposure in drinking water to etoposide (ETO, 5 mg/kg/day), permethrin (PER, 10 mg/kg/day), and chlorpyrifos (CPF, 10 mg/kg/day) or 0.1% dimethylsulfoxide (DMSO). The presence of the PER and CPF metabolites 3-BPA and TCPy in urine and serum was analyzed 48 h later by gas chromatography-mass spectometry. Mice were sacrificed for interphase fluorescence in situ hybridization (iFISH) and fluorescence activated cell sorting (FACS) analysis after 12 weeks of continuous treatment. (B) Left, scheme depicting the human chromosome 11 and the 11q23 region where the MLL probe hybridizes. Middle, representative iFISH image showing human cells with germline MLL or MLL rearrangement (MLLr). Right, percentage of human CD45+ cells harboring MLL breaks detected by iFISH at sacrifice. A minimum of 500 nuclei per sample was analyzed, except in one mouse from the DMSO group, for which only 216 nuclei could be analyzed. (C) Upper panels show the percentage of human engraftment (CD45+HLA-ABC+ cells) in bone marrow and peripheral blood. Lower panels show the relative proportion of immature (CD34+), myeloid (CD33+) and B-cell (CD19+) populations within human engraftment in bone marrow and peripheral blood. (D) Representative macroscopic images of spleens at sacrifice for each experimental group. GC-MS: gas chromatography-mass spectrometry.

546

haematologica | 2022; 107(2)

Letters to the Editor

remaining half of the litter was maintained for 32 weeks for analysis at adulthood (Figure 3A). iFISH analysis revealed that exposure to permethrin or chlorpyrifos during pregnancy failed to induce Mll breaks in bone marrow Lin-Kit+ progenitors from mothers, pups or adult offspring, whereas etoposide exposure induced Mll breaks in a small but significant proportion of Lin-Kit+

progenitors in the mothers and adult offspring, but not in the pups (Figure 3B, C, Online Supplementary Figure S3C). Finally, we analyzed whether prenatal exposure to etoposide or pesticides affects the hematopoietic homeostasis of mothers and pups (at weaning) and adult offspring. Analysis at sacrifice revealed no significant differ-

Figure 3. Legend on following page

haematologica | 2022; 107(2)

547

Letters to the Editor

Figure 3. Continuous exposure to permethrin or chlorpyrifos during pregnancy fails to induce Mll breaks in bone marrow progenitors or alterations in the hematopoietic homeostasis of mothers, pups or adult offspring in mice. (A) Experimental design to determine whether prenatal exposure to etoposide (ETO), permethrin (PER) or chlorpyrifos (CPF) induces Mll breaks or hematopoietic alterations in mothers, pups or adult offspring in mice. In brief, pregnant CD1 female mice were exposed to PER (20 mg/kg/day), CPF (20 mg/kg/day), ETO (10mg/kg/day) or 0.1 % dimethylsulfoxide (DMSO) from day 0 to day 21 of gestation. The number of pups per litter and their sex distribution were analyzed at birth. The mothers and one-half of the litter were analyzed at weaning while the remaining one-half of the offspring were maintained for 32 weeks for analysis in adulthood. (B) Upper panel, scheme depicting mouse chromosome 9 and the region where mouse Mll probes hybridize (UCSC, GRCm38/mm10). Lower left panel, a representative fluorescence in situ hybridization (FISH) image of a mouse metaphase and presence of fluorescence signals in both Mll alleles. Lower right panel, a zoom image of mouse chromosome 9 by DAPI banding, with and without the BAC fluorescence signal revealing the Mll gene localization. (C) Percentage of mouse bone marrow (BM) LK cells with Mll breaks detected by interphase FISH at sacrifice. DMSO and ETO were used as negative and positive controls, respectively. The numbers in bars indicate the number of mice analyzed. (D) Fluorescence activated cells sorting (FACS) BM analysis of the Lin- Sca-1+ Kit+ (LSK) subpopulation (upper panels), hematopoietic stem and progenitor cell subsets (middle panels), and mature cells (lower panels) in mothers, pups and adult offspring. (E) Representative macroscopic images of spleens (upper panel) and livers (lower panel) at sacrifice of mothers, pups and adult offspring exposed to the indicated treatments. HSC: hematopoietic stem cells; MPP: multipotent progenitors; HPC: hematopoietic progenitor cells.

ences in the proportions of mature cells (myeloid, T and B) in peripheral blood or total Lin-Sca+Kit+ progenitors, hematopoietic stem cells, multipotent progenitors or hematopoietic progenitor cells in the bone marrow between treatments in any group (Figure 3D, Online Supplementary Figure S3D). Similarly, prenatal exposure to etoposide or pesticides did not affect the hematopoietic homeostasis in peripheral blood, as determined by absolute numbers of white blood cells, red blood cells and platelets (Online Supplementary Figure S3E). Lastly, no evidence of splenomegaly or hepatomegaly was observed in mothers, pups or adult offspring (Figure 3E). Our results thus suggest that chronic exposure to permethrin or chlorpyrifos during pregnancy does not induce Mll breaks in bone marrow progenitors or alterations in the hematopoietic homeostasis of mothers, pups or adult offspring. A unique strength of the present study is the cuttingedge in vivo models employed to assess the genotoxicity and leukemogenesis potential of etoposide, permethrin or chlorpyrifos. The NSG mice model was established to mimic the adult exposure associated with occupational risk, whereas the CD1 mice model attempted to mimic prenatal exposure to topoisomerase II poisons and insecticides suggested to be involved in the etiology of infant leukemia. Continuous exposure to permethrin or chlorpyrifos in both models failed to induce MLL breaks or alterations in hematopoietic homeostasis, confirming the in vitro results of limited genotoxicity and no leukemogenic potential of permethrin or chlorpyrifos in human and murine HSPC after chronic exposure. The fact that MLL breaks are acutely induced by permethrin or chlorpyrifos but are not sustained upon long-term chronic exposure in vitro or in vivo indicates a legitimate repair of the DNA damage/DSB in the MLL locus. Of note, although long-term in vivo exposure to etoposide did induce MLLr in some hematopoietic progenitors, it failed to initiate leukemia in either in vivo models, in line with a previous study confirming that in utero exposure to etoposide did not trigger the development of leukemia in either Atm+/+ or Atm-/- mice.13 The eventual development of overt leukemia might depend on the survival and proliferative advantage of minor MLLr pre-leukemic clones, targeting the right cell-of-origin, on stromal bone marrow interactions and also on the acquisition of secondary cooperating oncogenic alterations. The clearance and lack of selection of MLLr clones is consistent with the development of MLLr treatment-related acute leukemia in adults or infant leukemia in only a rare subset of patients exposed to topoisomerase II poisons. Our results clearly suggest that permethrin and chlorpyrifos induce MLL breaks in human HSPC across ontogeny. However, such insecticide-induced DNA-DSB are successfully repaired, and do not involve chromoso-

548

mal translocations encoding MLL fusions. This explains their limited genotoxicity and no leukemogenic potential and reinforces why these compounds are still considered as non-classifiable carcinogens. Linking environmental or genotoxic exposure to causal and/or functional MLL chromosomal translocations has long been controversial,11 and the cutting-edge in vitro and in vivo cellular models employed in the present study also have obvious limitations. For instance, site-specific cleavage within the MLL break cluster region (bcr) has been shown to be induced by either topoisomerase II poisons but also genotoxic chemotherapeutic agents which do not target topoisomerase II and even by non-genotoxic stimuli of apoptotic cell death. In addition, MLL chromosomal translocations have been linked to higher-order chromatin fragmentation that occurs during the initial stages of apoptosis, suggesting that the generation of MLL chromosomal translocations (and likely others) are part of a generalized acute apoptotic response-mediated higher-order chromatin fragmentation which ultimately renders a chromosome topology and chromatin structure prone to chromosomal DNA exchanges.14 This is further supported by the ambiguity of MLL translocations partnering with a large number of different chromosomal loci. Virginia C. Rodriguez-Cortez,1 María Pilar NavarreteMeneses,2* Oscar Molina,1* Talia Velasco-Hernandez,1 Jessica Gonzalez,3 Paola Romecin,1 Francisco GutierrezAgüera,1 Heleia Roca-Ho,1 Meritxell Vinyoles,1 Eric Kowarz,4 Pedro Marin,5 Sandra Rodriguez-Perales,6 Carlos Gomez-Marin,7 Patricia Perez-Vera,2 Felipe CortesLedesma,7 Anna Bigas,1,3,8 Andrea Terron,9 Clara Bueno1,8 and Pablo Menendez1,8,10 1 Josep Carreras Leukemia Research Institute. Department of Biomedicine. School of Medicine, University of Barcelona. Barcelona, Spain; 2Laboratorio de Genética y Cáncer, Departamento de Genética Humana, Instituto Nacional de Pediatría, Ciudad de México, México; 3 Cancer Research Program, Institut Hospital del Mar d'Investigacions Mèdiques, Hospital del Mar, Barcelona, Spain; 4Institute of Pharmaceutical Biology/DCAL, Goethe-University of Frankfurt, Frankfurt/Main, Germany; 5Hematology Department. Hospital Clínic de Barcelona, Barcelona, Spain; 6Molecular Cytogenetics Group, Human Cancer Genetics Program, Spanish National Cancer Research Center (CNIO), Madrid, Spain; 7Topology and DNA Breaks Group, Spanish National Cancer Center (CNIO), Madrid, Spain; 8Centro de Investigación Biomedica en Red-Oncología (CIBERONC), Madrid, Spain; 9European Food and Safety Authority. Parma. Italy and 10 Instituciò Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Spain. *MPNM and OM contributed equally. Correspondence: PABLO MENÉNDEZ - pmenendez@carrerasresearch.org haematologica | 2022; 107(2)

Letters to the Editor

VIRGINIA RODRÍGUEZ-CORTEZ - vrodriguez@carrerasresearch.org doi:10.3324/haematol.2021.279047 Received: April 21, 2021. Accepted: October 19, 2021. Pre-published: October 28, 2021. Disclosures: no conflicts of interest to disclose. Contributions:VR-C and MPN-M: conceptualization, methodology, formal analysis and investigation, writing and preparation of the original draft; OM and TV-H: methodology, formal analysis and investigation; JG, PR, FG-A, HR-H, EK, MV, SR-P and CG-M: methodology; PMa, AB, AT, PP-V and FC-L: resourses; CB and Pme: conceptualization, writing, review and editing, supervision. Acknowledgments:the authors would like to thank the staff of the mass spectrometry unit of the scientific and technological services of the University of Barcelona for pesticide metabolite analysis. We are also grateful to the staff at the animal facility of the scientific park of Barcelona for urine collection in metabolic cages. Funding: this work was supported by CERCA/Generalitat de Catalunya and Fundació Josep Carreras-Obra Social la Caixa through institutional support. Pablo Menendez’s laboratory was supported by the Spanish Ministry of Science and Innovation (PID2019-108160RB-I00 / AEI / 10.13039/501100011033), Retos Colaboración with Banc de Sang i Teixits (RTC-2017-63671), the European Food and Safety Authority (EFSA.PRAS.2018.04-CT1), and the European Research Council. VR-C and MV were supported by Juan de la Cierva fellowships from the Spanish Ministry of Science and Innovation (FJCI-201524303 and IJCI-2017-33172). CB was supported by the Health Institute Carlos III (ISCIII/FEDER PI17/01028 and PI20/00822). PP-V and PN-M were supported by Consejo Nacional de Ciencia y Tecnología (CB-2012-01-183467) and by Fondos del Presupuesto Federal para la Investigación (project 001/2013, Instituto Nacional de Pediatría). JG was supported by Centro de Investigación Biomédica en Red en Cáncer (PID2019-104695RBI00). TV-H was supported by a Maria Sklodowska-Marie Curie fellowship (792923). OM was supported by a Beatriu de Pinós postdoctoral fellowship (BP2016-00048) from the Generalitat de Catalunya, a Lady Tata award from the Lady Tata Memorial Trust, and an investigator fellowship from the Spanish Cancer Research Association (AECC INVES211226MOLI). SR-P is supported by grants from the Spanish National Research and Development Plan, Instituto de Salud Carlos III, and FEDER (PI17/02303, PI20/01837 and DTS19/00111); AEI/MICIU

haematologica | 2022; 107(2)

EXPLORA Project BIO2017-91272-EXP and AECC_Lab_2020 Project. PM is an investigator of the Spanish Cell Therapy cooperative network (TERCEL).

References 1. Infante-Rivard C, Labuda D, Krajinovic M, Sinnett D. Risk of childhood leukemia associated with exposure to pesticides and with gene polymorphisms. Epidemiology. 1999;10(5):481-487. 2. Ma X, Buffler PA, Gunier RB, et al. Critical windows of exposure to household pesticides and risk of childhood leukemia. Environ Health Perspect. 2002;110(9):955-960. 3. Turner MC, Wigle DT, Krewski D. Residential pesticides and childhood leukemia: a systematic review and meta-analysis. Environ Health Perspect. 2010;118(1):33-41. 4. Navarrete-Meneses MP, Pedraza-Meléndez AI, Salas-Labadía C, Moreno-Lorenzana D, Pérez-Vera P. Low concentrations of permethrin and malathion induce numerical and structural abnormalities in KMT2A and IGH genes in vitro. J Appl Toxicol. 2018;38(9):12621270. 5. Navarrete-Meneses MDP, Pérez-Vera P. Pyrethroid pesticide exposure and hematological cancer: epidemiological, biological and molecular evidence. Rev Environ Health. 2019;34(2):197-210. 6. Lu C, Liu X, Liu C, et al. Chlorpyrifos induces MLL translocations through caspase 3-dependent genomic instability and topoisomerase II inhibition in human fetal liver hematopoietic stem cells. Toxicol Sci. 2015;147(2):588-606. 7. Ferreira JD, Couto AC, Pombo-de-Oliveira MS, Koifman S. In utero pesticide exposure and leukemia in Brazilian children < 2 years of age. Environ Health Perspect. 2013;121(2):269-275. 8. Ding G, Shi R, Gao Y, et al. Pyrethroid pesticide exposure and risk of childhood acute lymphocytic leukemia in Shanghai. Environ Sci Technol. 2012;46(24):13480-13487. 9. Borkhardt A, Wilda M, Fuchs U, Gortner L, Reiss I. Congenital leukaemia after heavy abuse of permethrin during pregnancy. Arch Dis Child Fetal Neonatal Ed. 2003;88(5):436-437. 10. Bueno C, Catalina P, Melen GJ, et al. Etoposide induces MLL rearrangements and other chromosomal abnormalities in human embryonic stem cells. Carcinogenesis. 2009;30(9):1628-1637. 11. Stanulla M, Wang J, Chervinsky DS, Thandla S, Aplan PD. DNA cleavage within the MLL breakpoint cluster region is a specific event which occurs as part of higher-order chromatin fragmentation during the initial stages of apoptosis. Mol Cell Biol. 1997;17(7):4070. 12. Schellenberg MJ, Lieberman JA, Herrero-Ruiz A, et al. ZATT (ZNF451)–mediated resolution of topoisomerase 2 DNA-protein cross-links. Science. 2017;357(6358):1412-1416. 13. Nanya M, Sato M, Tanimoto K, Tozuka M, Mizutani S, Takagi M. Dysregulation of the DNA damage response and KMT2A rearrangement in fetal liver hematopoietic cells. PLoS One. 2015;10(12):1-18. 14. Aplan PD. Chromosomal translocations involving the MLL gene: molecular mechanisms. DNA Repair (Amst). 2006;5(9-10):12651272.

549

Letters to the Editor

Somatic STAT3 mutations in CD8+ T cells of healthy blood donors carrying human T-cell leukemia virus type 2 Chronic viral antigen stimulation may underlie CD8+ T-cell expansion and T-cell large granular lymphocyte leukemia (T-LGLL).1 In T-LGLL, CD8+ T-cell expansions are associated with somatic mutations that solidify clonal dominance, such as in the case of activating STAT3 mutations which are found in 40% of patients.2 However, whether chronic exposure to viral antigens are associated with somatic mutations in expanding CD8+ T cells among individuals without clinically detectable lymphoproliferations is currently not known. Human T-cell leukemia virus type 2 (HTLV-2) preferentially targets CD8+ T cells, causing strong expansion of the infected CD8+ T-cell clones.3 Whereas HTLV-1 is the causative agent of adult T-cell leukemia/ lymphomas,4 the etiologic role of HTLV-2 in lymphoproliferative diseases is less clear. In 1992 an incidental case of LGL leukemia with HTLV-2 seropositivity was described.5 Later, Thomas et al.6 reported anti-HTLV antibody positivity in 44% of T-LGLL patients. While some cross-reactivity cannot be excluded, 7.5% (4 of 53) of T-LGLL patients tested positive for HTLV-2 by both western blotting and polymerase chain reaction (PCR),6 suggesting that HTLV-2 may participate in T-LGLL pathogenesis in a minority of cases. In this study, we examined whether CD8+ T cells from healthy, asymptomatic blood donors with chronic HTLV-2 infection harbor somatic mutations in STAT3 or other immune-associated genes, potentially identifying invidiuals at risk of subsequent lymphoproliferative diseases. This study was conducted with samples from the HTLV Outcomes Study (HOST) which includes samples from subjects recruited from five major US blood donation centers.7 HTLV status was analyzed with an ezymelinked immunosorbant (ELISA) assay, followed by western blot confirmation and HTLV-1 versus HTLV-2 typing by either real-time PCR or a type specific ELISA.7 At each visit, cohort participants were interviewed in detail for symptoms, followed by a physical and neurological examination. Informed consent was obtained from all participants. The study and sample collection were approved by the University of California San Francisco committee on human research and other Institutional Review Boards. We obtained frozen peripheral blood mononuclear cells (PBMC) of 30 HTLV-2 infected and 35 HTLV-2 uninfected blood donors from University of California San Francisco and Vitalant Research Institute (CA, USA). The PBMC samples collected between 2000 and 2008 were randomly selected from HOST. We separated CD4+ and CD8+ T cells from PBMC samples of HTLV-2 positive (n=30) and negative (n=35) healthy blood donors. The presence of STAT3 mutations in the sorted fractions was analyzed by ultra-deep targeted amplicon sequencing, covering hotspot regions of STAT3 gene (median coverage =7,472, sensitivity =0.5% variant allele frequency [VAF]). The sequencing was performed with Illumina Miseq System (Online Supplementary Figure S1A), and variant calling was performed as previously reported.8 Somatic nonsynonymous STAT3 mutations were discovered in CD8+ T cells from four of 30 (13.3%) HTLV-2 positive subjects, whereas no STAT3 mutations were discovered in HTLV2 negative subjects using deep amplicon sequencing of STAT3 (Figure 1A; Fisher’s exact test P=0.04). Furthermore, no STAT3 mutations were discovered in 550

CD4+ T cells, indicating that the mutations were specific for the CD8+ T-cell subset. Among the four STAT3 mutations detected, three were missense STAT3 mutations (Y640F, N647I and D661Y) and one was a nonframeshift insertion (Y657_K658insY), with VAF of 11.9%, 0.5%, 4.9%, and 1.2%, respectively (Figure 1B). All the mutations identified in CD8+ T cells were located in the SH2 domain of STAT3 and have been previously reported in T-LGLL (Online Supplementary Figure S1A).2 The proportion of differentiated, putatively cytotoxic CD57+, CD16+ CD8+ T cells were higher in STAT3 mutated compared to STAT3 unmuted HTLV-2 positive individuals (Figure 1C). In addition, higher level of the cytotoxic marker perforin was noted in CD8+ T cells of STAT3 mutated, HTLV-2 positive individuals (Online Supplementary Figure S1B and C). TCRb deep sequencing9 of sorted CD8+ T cells revealed higher clonality index in the STAT3 mutated compared to STAT3 unmuted individuals (Figure 1D). The VAF of the STAT3 Y640F mutation was consistent with clonal event in the largest TCRBV03-01 clone; the other three STAT3 mutations with smaller VAF likely occurred as subclonal events or in the smaller CD8+ T-cell clones (Figure 1E). The age distribution was similar between HTLV-2 negative subjects (median age 52 year), HTLV-2 positive subjects without STAT3 mutations (median age 53 years), and HTLV-2 positive subjects with STAT3 mutations (median age 58.5 years) (P-value between HTLV-2 positive subjects with and without STAT3 mutations, P=0.50; Mann–Whitney U test) (Figure 1F). No statistical difference in viral load was detected between STAT3 mutated (median =0.149) and unmuted HTLV-2 positive subjects (median =0.0005) (P=0.5; Mann–Whitney U test) (Figure 1G). No serial HTLV-2 viral load measurements were available; however, HTLV-2 viral load has been reported to be stable over time.10 There was no difference in total white blood cell and lymphocyte counts between STAT3 mutated and unmuted cases (Figure 1H and I). In order to characterize a larger spectrum of somatic variants in genes linked to immune regulation, we analyzed CD8+ T cells from 28 HTLV-2 positive subjects using a custom next generation sequencing panel covering the coding regions of 2,533 immune-related genes.11 Samples were sequenced with Illumina HiSeq or NovaSeq 6000 system (Online Supplementary Figure S2), and somatic variant calling followed a previously described approach.11 Variants were filtered using population based filtering, MuTect2 filters and against CD4+ and CD8+ panels of normals from 21 healthy controls (Online Supplementary Figure S2). Variants only found in CD8+ T cells, using sorted CD4+ T cells as matched normals, were evaluated further. Sequencing coverages are presented in the Online Supplementary Figure S2. Two STAT3 mutations (Y640F and D661Y) were detected in two of four STAT3 mutant cases. N647I and Y657_K658insY variants did not pass filtering due to lower sequencing coverage compared to amplicon sequencing; their presence was confirmed with visual inspection of the sequencing data using Integrative Genomics Viewer.12 In addition to STAT3 mutations, we identified a total of 66 coding somatic variants in 61 genes in CD8+ T cells (Figure 2A). Nineteen (68%) of the subjects had at least one variant, and the median number of variants was one per subject. Eight subjects (29%) harbored variants in the genes previously discovered in LGLL (STAT3, KMT2D, TYRO3, DIDO1, BCL11B, CACNB2, KRAS, LRBA and FANCA),13 and five subjects (18%) harbored genes involved in JAKhaematologica | 2022; 107(2)

Letters to the Editor

Figure 1. STAT3 mutations discovered in CD8+ T cells of HTLV-2 positive subjects. (A) Prevalence of STAT3 mutations in CD4+ and CD8+ T cells of human Tcell leukemia virus type 2 (HTLV-2) positive subjects (n=30) and HTLV-2 negative subjects (n=35). Four (13.3%; Fisher’s exact test P=0.04) of 30 HTLV-2 positive individuals had STAT3 mutations in CD8+ T cells. (B) STAT3 mutations found in HTLV-2 positive subjects by amplicon sequencing. STAT3 mutations (insertion, N647I, Y640F and D661Y) were discovered in CD8+ T cells from 4 HTLV-2 positive subjects. (C) Flow cytometry based immunophenotyping was performed to identify the proportion of differentiated, putatively cytotoxic (CD56+, CD57+, and CD16+) CD8+ T cells in 9 HTLV-2 positive blood donors. For the immunophenotyping, anti-CD3 APC, -CD45 V500, -CD4 APC-H7, -CD8 PE-Cy7, -CD16 PerCP-Cy5.5, -CD56 FITC, and -CD57 PE, were used. Each dot represents one individual, and horizontal lines indicate median values. Statistically significant difference was evaluated using Mann-Whitney U test. (D) The CD8+ T-cell clonality index by STAT3 mutation status in 8 HTLV-2 positive blood donors. The clonality index was calculated using ImmnoSEQ Analyzer software (Adaptive Biotechnologies, WA, USA) as 1 minus Shannon entropy normalized by the logarithm of the number of productive T-cell receptor (TCR) sequences. Each dot represents 1 individual, and horizontal lines indicate median values. P-values were evaluated using Mann-Whitney U test. (E) CD8+ T-cell repertoire analyzed with TCRb deep sequencing (Adaptive Biotechnologies). Sorted CD8+ T cells of HTLV-2 positive cases bearing STAT3 mutations (n=4) and without STAT3 mutations (n=4) were used. Variant allele frequency (VAF) was analyzed by amplicon sequencing. The graph shows top 3 TCR clones in each sample. (F-I) (F) Age distribution, (G) HTLV-2 proviral load in copies per peripheral blood mononuclear cells (PBMC), (H) white blood cells count and (I) lymphocytes count within HTLV-2 negative subjects (HTLV2-), HTLV-2 positive subjects without STAT3 mutations (No STAT3Mut) and HTLV-2 positive subjects harboring STAT3 mutations (STAT3Mut). Each dot represents one individual. P-values were calculated using Mann–Whitney U test (No STAT3Mut vs. STAT3Mut). Horizontal lines indicate median values. Ref: reference base; Var: variant base.

haematologica | 2022; 107(2)

551

Letters to the Editor

Table 1. Clinical characteristics of HTLV-2 positive subjects analyzed in immunogene panel sequencing.

Age Median (range) Sex Female Male Race Black Hispanic White White blood cells (109/L) Median (range) Hemoglobin (g/dL) Median (range) Hematocrit (%) Median (range) Platelets (109/L) Median (range) Lymphocytes (109/L) Median (range) Eosinophils (109/L) Median (range) Neutrophils (109/L) Median (range) Monocytes (109/L) Median (range) Basophils (109/L) Median (range)

Total n = 28 (%)

0 n = 10 (36)

Mutation Number 1-2 n = 8 (29)

>3 n = 10 (36)

P-value

53 (43 - 70)

52 (46 - 70)

54 (43 - 70)

60 (49 - 68)

0.04†

19 (68) 9 (32)

7 (70) 3 (30)

6 (75) 2 (25)

6 (60) 4 (40)

0.63^

9 (32) 5 (18) 14 (50) 6.2 (3.5-10) 13.9 (4.8-17.4) 40.8 (14-50.5) 231 (101-435) 2.378 (0.984-4.122) 0.130 (0.028-0.319) 3.207 (1.404-6.232) 0.410 (0.042-1.140) 0.0 (0.0-0.2)

1 (10) 2 (20) 7 (70) 6.2 (3.5-8.2) 13.4 (4.8-16.7) 40.3 (14-49.2) 222 (144-435) 1.906 (0.984-3.854) 0.117 (0.028-0.246) 3.418 (1.582-6.232) 0.378 (0.042-0.711) 0.0 (0.0-0.2)

4 (50) 1 (12) 3 (38) 6.6 (6.0-9.1) 13.6 (12.5-17.4) 39.9 (38.7-50.5) 239 (200-340) 2.907 (1.365-4.122) 0.185 (0.130-0.319) 3.393 (2.500-4.940) 0.434 (0.195-0.919) 0.0 (0.0-0.1)

4 (40) 2 (20) 4 (40) 5.8 (3.9-10) 14.2 (12.8-16.2) 42.7 (38.1-48.9) 240 (101-350) 2.378 (1.566-3.610) 0.078 (0.057-0.200) 2.944 (1.404-5.700) 0.480 (0.235-1.140) 0.0 (0.0-0.2)

0.41^

0.85† 0.27† 0.15† 0.80† 0.24† 0.21† 0.51† 0.24† 0.99†

The data are at sample collection. P-values are calculated by Cuzick’s trend test (†), or Kruskal-Wallis trend test (^).

STAT signaling pathway (NFKBIA, PIK3R5, MAPK14, EP300, MPL, IFNAR1, IL6ST and IL20RA) according to Uniprot identifier. Three genes had more than one variant: VWF (3 mutations), SMAD7 and MXRA5 (2 mutations each) (median VAF: 7%, 2% and 4%, respectively). Subject 13 who harbored a STAT3 Y640F mutation also had mutations in KMT2D, NFKBIA, PIK3R5, CTCF, and VWF. In this subject with multiple somatic mutations, STAT3 had the highest VAF (16.2%, Online Supplementary Table S1), suggesting that other variants may be subclonal. Subject 12 with a STAT3 N647I mutation also harbored variants in INPP5D, FSCN1 and PLA2R1, CD248, and P4HTM. Subject 11 with STAT3 Y657_K658insY harbored variants in MTA1, NCOR2, BCL6, ADCY8, RPS6KA3. Subject 27 with a STAT3 D661Y mutation had no additional variants. The overall number of mutations was higher in HTLV-2 positive blood donors harboring STAT3 mutations (median =6) compared to HTLV-2 positive blood donors without STAT3 mutations (median =1; P=0.061; Mann–Whitney U test), although the difference was not statistically significant. The complete list of variants and VAF can be found in the Online Supplementary Table S1. Somatic mutations can accumulate in tissues with aging.14 Accordingly, the total number of coding variants was associated with older age among HTLV-2 positive blood donors (no variants, median age 51 years; 1-2 variants, 53 years; 3 or more variants, 59 years; P=0.04) (Table 1; Figure 2B). The most frequent single nucleotide transition was C>T involved in age-associated mutational signature 1 (Figure 2C). The high prevalence of signature 1, 552

revealed by mutational signature analysis (Figure 2D), further supported age-dependent accumulation of somatic variants in CD8+ T cells among HTLV-2 positive subjects. No association between peripheral blood counts and number of variants was observed (Table 1). In summary, our results highlight the presence of STAT3 mutations in CD8+ T cells of healthy blood donors harboring HTLV-2 without clinical history of lymphoproliferative disease. HTLV-2 positive subjects with STAT3 mutations showed variable clonal expansion of CD8+ T cells, suggesting that HTLV-2 infection may promote lymphoproliferation and STAT3 mutagenesis outside the clinical context of T-LGLL. We identified additional mutations in CD8+ T cells of HTLV-2 positive subjects in genes involved in JAKSTAT signaling, immune regulation and lymphoproliferation. In addition to T-LGLL, somatic STAT3 mutations have been detected in LGLL associated diseases such as aplastic anemia, hypoplastic myelodysplastic syndrome and Felty’s syndrome, and in some patients with multiple sclerosis and rheumatoid arthritis.15 Although CD8+ T cell expansions can also be detected in other diseases such as in rheumatoid arthritis, somatic STAT3 mutations are not common in these conditions.8 STAT3 mutations and a history of HTLV-2 infection may highlight a subset of blood donors who are at risk of subsequent diagnosis of lymphoproliferative diseases. However, no clinical follow-up information is available from our study participants, and future studies are needed to elucidate whether HTLV-2 positive subjects carrying STAT3 and other mutations are at increased risk of T-LGLL or other lymphoproliferative diseases. haematologica | 2022; 107(2)

Letters to the Editor

Figure 2. Somatic mutations identified in CD8+ T cells of HTLV-2 positive subjects. (A) Mutation landscape in CD8+ T cells of human T-cell leukemia virus type 2 (HTLV-2) positive subjects (n=28). Coding variants identified by immunogene panel sequencing are presented, together with age, sex, number of mutations and mutation types. * (white asterisk), STAT3 variants detected in the deep amplicon sequencing but filtered out in the immunogene panel sequencing due to lower coverage but confirmed with visual inspection with the Integrative Genomics Viewer (Broad Institute, USA). The average expression of mutated genes in healthy controls are shown on the right, presented as counts per millions reads mapped (CPM) with mean ± standard deviation (n =5). The complete list of variants and variant allele frequencies (VAF) can be found in the Online Supplementary Table S1. (B) Correlation plot of age vs. mutation number in HTLV-2 positive blood donors. (C) Percentages of somatic base substitutions and indel identified by immunogene panel sequencing in CD8+ T cells of HTLV-2 positive blood donors. (D) Normalized weights of COSMIC signatures contributions. Signature 1 (weight: 0.362) was highly related signature.

Daehong Kim,1,2 Mikko Myllymäki,1,2 Matti Kankainen1-4 Timo Järvinen,1,2 Giljun Park,1,2 Roberta Bruhn,5,6 Edward L. Murphy5,6 and Satu Mustjoki1,2,4 1 Hematology Research Unit Helsinki, Helsinki University Hospital Comprehensive Cancer Center, Helsinki, Finland; 2Translational Immunology Research Program and Department of Clinical Chemistry and Hematology, University of Helsinki, Helsinki, Finland; 3 Department of Medical and Clinical Genetics, University of Helsinki and Helsinki University Hospital, Helsinki, Finland; 4iCAN Digital

haematologica | 2022; 107(2)

Precision Cancer Medicine Flagship, University of Helsinki and Helsinki University Hospital, Helsinki, Finland; 5Vitalant Research Institute, San Francisco, CA, USA and 6University of California San Francisco, San Francisco, USA Correspondence: SATU MUSTJOKI - satu.mustjoki@helsinki.fi doi:10.3324/haematol.2021.279140 Received: May 10, 2021. 553

Letters to the Editor

Accepted: October 20, 2021. Pre-published: October 28, 2021. Disclosures: SM has received honoraria and research funding from Novartis, Pfizer and Bristol-Myers Squibb (not related to this study). MM has received honoraria from Celgene and Sanofi. All other authors declare have no conflicts of interest to disclose. Contributions: DK, MM and SM designed the study and experiments; DK performed sample preparation and analyzed data; MK performed immunogene panel sequencing analysis; TJ performed coverage analysis of immunogene panel sequencing; GP performed sample preparation; RB and ELM provided the clinical data and study materials; DK, MM and SM wrote the manuscript with the input of other authors; SM supervised the study. All authors read and approved the final manuscript. Acknowledgment: Amplicon sequencing and Immunogene panel sequencing were performed at the Institute for Molecular Medicine Finland (FIMM) Technology Center, which is supported by Biocenter Finland. The authors also acknowledge IT Center for Science Ltd for computational resources. Funding: the research was funded by European Research Council (M-IMM project), ERAPerMed consortium JAKSTAT-TARGET, Academy of Finland, Finnish special governmental subsidy for health sciences, research and training, Sigrid Juselius Foundation, Helsinki Institute of Life Sciences Fellow funding, and Cancer Foundation Finland. The HOST study was supported by the US National Heart, Lung and Blood Institute research grant 2R01-HL62235 and Signe and Ane Gyllenberg Fundation. Data sharing statement: all data are available in the Online Supplementary Appendix.

References 1. Lamy T, Moignet A, Loughran TP, Jr. LGL leukemia: from pathogenesis to treatment. Blood. 2017;129(9):1082-1094.

554

2. Koskela HL, Eldfors S, Ellonen P, et al. Somatic STAT3 mutations in large granular lymphocytic leukemia. N Engl J Med. 2012;366(20):1905-1913. 3. Melamed A, Witkover AD, Laydon DJ, et al. Clonality of HTLV-2 in natural infection. PLoS Pathog. 2014;10(3):e1004006. 4. Kataoka K, Nagata Y, Kitanaka A, et al. Integrated molecular analysis of adult T cell leukemia/lymphoma. Nat Genet. 2015;47(11):1304-1315. 5. Loughran TP, Jr., Coyle T, Sherman MP, et al. Detection of human T-cell leukemia/lymphoma virus, type II, in a patient with large granular lymphocyte leukemia. Blood. 1992;80(5):1116-1119. 6. Thomas A, Perzova R, Abbott L, et al. LGL leukemia and HTLV. AIDS Res Hum Retroviruses. 2010;26(1):33-40. 7. Lee TH, Chafets DM, Busch MP, Murphy EL. Quantitation of HTLV-I and II proviral load using real-time quantitative PCR with SYBR Green chemistry. J Clin Virol. 2004;31(4):275-282. 8. Savola P, Kelkka T, Rajala HL, et al. Somatic mutations in clonally expanded cytotoxic T lymphocytes in patients with newly diagnosed rheumatoid arthritis. Nat Commun. 2017;8:15869. 9. Kim D, Park G, Huuhtanen J, et al. Somatic mTOR mutation in clonally expanded T lymphocytes associated with chronic graft versus host disease. Nat Commun. 2020;11(1):2246. 10. Kwaan N, Lee TH, Chafets DM, et al. Long-term variations in human T lymphotropic virus (HTLV)-I and HTLV-II proviral loads and association with clinical data. J Infect Dis. 2006;194(11):15571564. 11. Savola P, Martelius T, Kankainen M, et al. Somatic mutations and T-cell clonality in patients with immunodeficiency. Haematologica. 2020;105(12):2757-2768. 12. Robinson JT, Thorvaldsdottir H, Winckler W, et al. Integrative genomics viewer. Nat Biotechnol. 2011;29(1):24-26. 13. Coppe A, Andersson EI, Binatti A, et al. Genomic landscape characterization of large granular lymphocyte leukemia with a systems genetics approach. Leukemia. 2017;31(5):1243-1246. 14. Blokzijl F, de Ligt J, Jager M, et al. Tissue-specific mutation accumulation in human adult stem cells during life. Nature. 2016;538(7624):260-264. 15. Mustjoki S, Young NS. Somatic mutations in "benign" disease. N Engl J Med. 2021;384(21):2039-2052.

haematologica | 2022; 107(2)

Letters to the Editor

Incidence and outcome of SARS-CoV-2 infection in patients with monoclonal gammopathy of undetermined significance: a case-control study In March 2020, "Coronavirus Disease 2019" (COVID19) outbreak due to SARS-CoV-2 was declared a pandemic by the World Health Organization. Clinical manifestations of COVID-19 are variable, ranging from complete absence of symptoms, to severe pneumonia, multiorgan failure, and death. Main risk factors for poor outcome of COVID-19 are advanced age and comorbidities, conditions that often recur in patients with monoclonal gammopathies. In this setting, several papers have reported more frequent and severe COVID-19, as well as higher fatality rate, in patients with multiple myeloma (MM), particularly in those older than 60 years, with high risk, active/progressive disease and/or renal failure, respect to general population.1-3 By contrast, very few data are available about patients with monoclonal gammopathies of undetermined significance (MGUS).4 In our retrospective study, we investigated the incidence of SARS-CoV-2 infection and COVID-19 outcomes in MGUS patients. Overall, we found that subjects with MGUS neither have an increased risk of contracting SARS-CoV-2, nor show poorer COVID-19 outcomes compared to controls. Patients with MGUS are frequently asymptomatic and diagnosed by chance; therefore, differently from MM, one would not expect an increased incidence of SARSCoV-2 infection or a worse outcome of COVID-19 respect to the normal population in these subjects. Notwithstanding, according to well recognized risk factors, they may have different risks of developing MM, as well as clinical findings, including older age, presence of medical comorbidities, and impaired humoral/cellular immunity, which could still play a role when assessing their risk during the COVID-19 pandemic.5,6 Notably, in epidemiological studies, people with MGUS were shown to have an increased risk of developing both venous and arterial thrombosis, bacterial and viral infections, as well as an excess mortality risk due to bacterial infections as compared to age and sex-matched healthy controls.7,8 On this basis, the presence of MGUS could possibly increase susceptibility to SARS-CoV-2 infection and severity of COVID-19, and might theoretically account for the increased mortality due to COVID-19 observed in the elderly population.9 However, a retrospective chart review aiming to study the vulnerability of 228 patients with MGUS (3 of whom resulted infected by SARS-CoV2, with 1 death) and their clinical outcomes during the COVID-19 pandemic, concluded that there were neither significant differences in the mean age or survival of the MGUS patients not infected by SARS-COV-2 who died before versus after the pandemic onset, nor an increase in venous thrombotic events.10 Furthermore, in a small case series of seven MGUS patients experiencing COVID-19, 71% were hospitalized, but none of the patients required mechanical ventilation or ICU (intensive care unit) management.4 Patients had an age range between 59 and 92 years and all had underlying high-risk comorbidities. One patient with acute kidney injury recovered after hemodialysis. The only death was a male patient with advanced age, nursing home residency, multiple comorbidities and elevated D-dimer. This small case series would suggest that MGUS does not pose additional risks for poor outcome in COVID-19 patients. The aim of our observational, retrospective, single haematologica | 2022; 107(2)

center study was to formally investigate the incidence of SARS-CoV-2 infection, as well as the characteristics and the clinical outcome of COVID-19 in a larger cohort of MGUS patients. The study was conducted within the context of the clinicaltrials gov. Identifier: NCT04352556. Between March 1, 2020 and April 30, 2021, we collected, among 1.454 MGUS patients followed at our center, clinical data from 91 patients with SARS-CoV2 infection, diagnosed by RT-PCR on nasopharyngeal swabs. Data were mainly extracted from “GIAVA-COVID-19”, a regional platform where authorized medical health workers can view the results of the nasopharyngeal swabs for SARS-CoV-2 performed, along with other information. In MGUS patients a review of medical records was also carried out. Clinical data collected regarded age, cardiovascular, pulmonary, diabetic and neoplastic comorbidities, presence of symptoms (in detail: fever or chills, cough, shortness of breath or difficulty breathing, fatigue, muscle or body aches, headache, loss of taste or smell, sore throat, congestion or runny nose, nausea or vomiting, diarrhea), hospitalization, hospitalization in ICU, and outcome (alive/dead). Patients with monoclonal gammopathies of clinical (renal, dermatological or neurological) significance (MGCS) were excluded. The characteristics of COVID-19 in MGUS patients were compared with those of 182 age- and sex-matched normal controls infected by SARS-CoV-2. Furthermore, the incidence of SARS-CoV-2 infection in MGUS patients was compared to that of the entire Apulian population (Apulia is a region of Southern Italy with about 4 million inhabitants). Wilcoxon test, chi-square tests and multivariate logistic regression were applied, as appropriate, by using STATA software MP17. Table 1 summarizes the main characteristics of SARSCoV-2 infected MGUS patients and controls. The two groups were comparable for age, sex, and presence of comorbidities. Among the MGUS group, 68 patients showed a non-IgM subtype and nine an IgM subtype: this information was missing in five patients. Nine patients had a double M-component. Immunoparesis (at least one uninvolved immunoglobulin below reference levels) was present in 19% of 68 evaluable patients. Regarding MGUS risk-stratification according to Mayo Clinic model, most (94%) of 47 patients with complete available dataset scored as low or low/intermediate risk. Sixty-two patients showed the presence of at least one co-existing, potentially clinically relevant comorbidity (cardiovascular disease 40.6%, diabetes 11%, nonhematological cancer 8.8%, pulmonary disease 6.6%). As shown in Table 1, rates of COVID-19-related symptoms, hospitalization, hospitalization in ICU and deaths due to COVID-19 were slightly higher in the MGUS group than in the control group, but these differences were not statistically significant. In MGUS patients, sex, immunoparesis, presence/number of comorbidities and IgM versus non-IgM isotype did not significantly influence the risk of death, while a statistically significant association was observed with older age; importantly, the risk of death was not correlated with the presence of MGUS (Table 2). Lastly, incidence of SARS-CoV2 infection in MGUS patients (91/1.454, 6.2%) was not statistically different from that observed in the entire population of the Apulia region (227.761/3.926.931, 5.8%) during the same period (Table 1). Thus, in our study, patients with MGUS, contrarily to what is seen in MM, did not show an increased incidence 555

Letters to the Editor

of SARS-CoV-2 infection compared to the general population. Furthermore, although rates of symptoms, hospitalization, hospitalization in ICU and deaths were slightly higher than in controls, MGUS did not appear to represent a risk for a poorer COVID-19 outcome. The only factor associated with an increased risk of death was older age; however this was likely not related to the presence of MGUS, but rather to the well known predictive power of this clinical parameter for a worse outcome in the general population infected by SARS-CoV-2.9 Although, to the best of our knowledge, we conducted the largest study of SARS-CoV-2 infection in patients with MGUS, several limitations are present in our analysis. First, as with any observational retrospective study,

there could be unintentional patient selection bias: MGUS may be misclassified and, conversely, individuals may be unaware of its presence. Second, not all MGUS patients had a complete dataset and some laboratory findings were lacking. Finally, long-term outcomes of MGUS and COVID-19 infection should also be explored. Therefore, further data are needed to achieve greater generalizability of our findings. Other questions, particularly the possibility of a suboptimal response in people with MGUS to anti-SARSCoV-2 vaccine (as observed in MM), possibly due to age and MGUS-related immune dysfunction, will be probably soon addressed by ongoing studies. In this setting, preliminary data show that MGUS patients receiving

Table 1. Characteristics of patients with monoclonal gammopathies of undetermined significance versus controls and COVID-19 outcome. Total, n. Mean Age, years +/- SD (range) Sex, n. (%) Male Female Comorbidities, n. (%) * no 1 2 ≥3 Mean number, n. +/- SD (range) Incidence SARS-CoV-2 infection (%) COVID-19 outcome Presence of symptoms, n. (%; 95% CI) Hospitalization, n. (%; 95% CI) Hospitalization in ICU, n. (%; 95% CI) Death due to COVID-19, n. (%; 95% CI) MGUS subtype (available in 86 patients), n. (%) IgG IgA IgM Biclonal LC only Immunoparesis (available in 68 patients) 0/1/2, n. ≥1 subclass (%) MGUS risk (available in 47 patients)°, n. (%) 0 Low 1 Low-intermediate 2 High-intermediate

MGUS

Controls

91 65.6 +/- 13.3 (29-89)

182 65.2 +/- 13.4 (29-89)

42 (46.2) 49(53.8)

80 (44) 102 (56)

29 (31.9) 30 (32.9) 19 (20.9) 13 (14.3) 1.3 +/-1.3 (0-5) 6.2 **

50 (27.5) 59 (32.5) 64 (35) 9 (5) 1.2 +/- 0.9 (0-3) 5.8 §

0.684 1.000 0.125 0.148 0.844 0.454

54 (59.3; 48.5-69.5) 19 (20.9; 13.1-30.7) 10 (11.0; 5.4-19.3) 8 (8.8; 3.9-16.6)

102 (56.0; 48.5-63.4) 26 (14.3; 9.5-20.2) 16 (8.8; 5.1-13.9) 10 (5.5; 2.7-10.0) NA

0.604 0.166 0.560 0.301

63 (73.2) 5 (6.0) 9 (10.4) 8 (9.3) 1 (1.1) 55/10/3 13/68 (19.1) 22 (46.8) 22 (46.8) 3 (6.4)

P-value 0.734 0.796

*Comorbidities evaluated included cardiovascular disease, pulmonary disease, diabetes, and non-hematological cancers; ** Incidence among 1,454 MGUS followed at our Institution; § Incidence among the entire population of the Apulia region (227.761 cases of SARS-CoV-2 infection in 3.926.931 inhabitants); ° Rajkumar et al. Blood 2005;106(3):812-7. MGUS: monoclonal gammopathies of undetermined significance; NA: not applicable; ICU: intensive care unit; LC: light chain; CI: confidence interval; SD: standard deviation

Table 2. Risk for death in COVID-19 monoclonal gammopathies of undetermined significance patients, adjusted for different clinical parameters.

Determinants

aOR

95% CI

P-value

Group (MGUS vs. controls) Age (years) Gender (Male vs. Female) Presence of comorbidities (Yes vs. No) Number of comorbidities (1-2 vs. ≥3) IgM vs non-IgM Immunoparesis (0 vs. ≥1)

0.76 0.88 0.27 2.16 2.01 0.28 0.45

0.20 - 3.00 0.78 - 0.99 0.04 - 1.91 0.26 - 17.7 0.11 - 37.8 0.31 - 2.56 0.05 - 3.98

0.700 0.035 0.193 0.474 0.641 0.262 0.473

MGUS: monoclonal gammopathies of undetermined significance; aOR: adjusted odds ratio; CI: confidence interval; IgM: immunoglobulin M.

556

haematologica | 2022; 107(2)

Letters to the Editor

anti-SARS-CoV-2 vaccines could have a better antibody response than those with MM.11 Anyway, in patients with MGUS, as in those with MM, full vaccination against SARS-CoV-2 should be strongly recommended to reduce the possibility of infection rate and severity of the illness.12-14 Last, but not least, the psychosocial impact of the pandemic on MGUS patients and their management in the long term follow-up also would warrant to be further and specifically investigated.15 Nicola Sgherza,1 Paola Curci,1 Rita Rizzi,1,2 Vanda Strafella,2 Daniela Di Gennaro,2 Angelantonio Vitucci,1 Antonio Palma,1 Antonella Vita Russo Rossi,1 Francesco Albano,2 Pasquale Stefanizzi,3 Silvio Tafuri3 and Pellegrino Musto1,2 1 Hematology and Bone Marrow Transplantation Unit, AOUC Policlinico di Bari; 2Department of Emergency and Organ Transplantation, “Aldo Moro” University School of Medicine and 3 Department of Biomedical Science and Human Oncology, “Aldo Moro” University School of Medicine, Bari, Italy Correspondence: PELLEGRINO MUSTO- pellegrino.musto@uniba.it doi:10.3324/haematol.2021.279895 Received: September 1, 2021. Accepted: October 28, 2021. Pre-published: November 4, 2021. Disclosures: no conflicts of interest to disclose. Contributions: PM and NS conceived and led the project; NS conducted database building, extraction and coding; NS, PM, PS and ST queried and analyzed the data; PM and NS wrote the main manuscript text and created all tables. All authors made a substantial intellectual contribution to the study, interpreted the data, discussed the results and reviewed, edited and approved the final version of the manuscript.

References 1. Chari A, Samur MK, Martinez-Lopez J, et al. Clinical features associated with COVID-19 outcome in multiple myeloma: first results from the International Myeloma Society data set. Blood. 2020;136(26):3033-3040. 2. Martínez-López J, Mateos MV, Encinas C, et al. Multiple myeloma and SARS-CoV-2 infection: clinical characteristics and prognostic

haematologica | 2022; 107(2)

factors of inpatient mortality. Blood Cancer J. 2020;10(10):103. 3. Engelhardt M, Shoumariyeh K, Rösner A, et al. Clinical characteristics and outcome of multiple myeloma patients with concomitant COVID-19 at Comprehensive Cancer Centers in Germany. Haematologica. 2020;105(12):2872-2878. 4. Gonzalez-Lugo JD, Bachier-Rodriguez L, Goldfinger M, et al. A case series of monoclonal gammopathy of undetermined significance and COVID-19. Br J Haematol. 2020;190(3):e130-e133. 5. Kyle RA, Larson DR, Therneau TM, et al. Long-term follow-up of monoclonal gammopathy of undetermined significance. N Engl J Med. 2018;378(3):241-249. 6. van de Donk NW, Palumbo A, Johnsen HE, et al. The clinical relevance and management of monoclonal gammopathy of undetermined significance and related disorders: recommendations from the European Myeloma Network. Haematologica. 2014;99(6):984996. 7. Kristinsson SY, Björkholm M, Andersson TM, et al. Patterns of survival and causes of death following a diagnosis of monoclonal gammopathy of undetermined significance: a population-based study. Haematologica. 2009;94(12):1714-1720. 8. Kristinsson SY,Tang M, Pfeiffer RM, et al. Monoclonal gammopathy of undetermined significance and risk of infections: a populationbased study. Haematologica. 2012;97(6):854-858. 9. G. Onder, G. Rezza, S. Brusaferro. Case-fatality rate and characteristics of patients dying in relation to COVID-19 in Italy. JAMA. 2020;323(18):1775-1776. 10. Lee H, Tay J, Street L, Duggan P, Jiménez-Zepeda VH. Monoclonal gammopathy of undetermined significance clinic during the coronavirus disease-19 pandemic: caring for the vulnerable in an academic medical center. Rev Invest Clin. 2021;73(4):259-264. 11. Terpos E, Gavriatopoulou M, Ntanasis-Stathopoulos I, et al. The neutralizing antibody response post COVID-19 vaccination in patients with myeloma is highly dependent on the type of antimyeloma treatment. Blood Cancer J. 2021;11(8):138. 12. https://cms.cws.net/content/beta.myelomasociety.org/files/PM%2 0COVID%20vaccination%20in%20MM%20guidelines%20The% 20Final.pdf 13. Ludwig H, Meckl A, Engelhardt M. Compliance with vaccination recommendations among patients with multiple myeloma: a real world experience. Hemasphere. 2021;5(7):e597. 14. Gavriatopoulou M, Ntanasis-Stathopoulos I, Korompoki E, Terpos E, Dimopoulos MA. SARS-CoV-2 vaccines in patients with multiple myeloma. Hemasphere. 2021;5(3):e547. 15. Quinn SJ, Anderson LA, Lohfeld L, McShane CM. The psychosocial impact of the COVID-19 pandemic on patients with monoclonal gammopathy of undermined significance, smouldering and active myeloma: findings from an international survey. Br J Haematol. 2021;194(2):294-297.

557

Letters to the Editor

Interferon a-induced SAMHD1 regulates human cultured megakaryocyte apoptosis and proplatelet formation Megakaryocyte (MK) growth, differentiation and maturation are required for thrombopoiesis and platelet production. Most studies of megakaryocytopoiesis have utilized in vitro culture systems expected to model a healthy human condition. However, consistent with the ability of MK to respond to inflammatory mediators, chronic inflammatory conditions often induce thrombocytosis, whereas acute inflammation can result in thrombocytopenia. Furthermore, there is an increasing awareness of the role MK play in innate and adaptive immunity.1 Type 1 interferons (IFN-1), including IFNa, IFNb and INFω are a family of cytokines that bind to the IFN-1 receptor and trigger transcription of diverse genes. IFNinducible genes regulate resistance to viral infections, enhance innate and adaptive immunity, and modulate

normal and tumor cell survival and death.2 MK express the IFN-1 receptor that signals through Janus kinases/signal transducer and activator of transcription proteins (JAK/STAT) pathway in response to IFN-1 cytokines.3 IFNa, an IFN-1 cytokine, has been effectively utilized in the treatment of myeloproliferative neoplasms and viral hepatitis. Thrombocytopenia is a common adverse effect of IFNa therapy that can require dose reduction. Although there are inconsistent reports regarding IFNa suppression of colony forming units of megakaryocyte progenitors (CFU-MK) in cultures of human CD34+ cells, there are consistent findings to support a mechanism of decreased platelet production rather than reduced platelet life span.4-7 However, the molecular mechanisms regulating IFNa-induced decrease in platelet production and peripheral blood thrombocytopenia are poorly understood. The major conclusions in this report are (i) using genome-wide gene expression profiling we show that IFNa upregulates the

Figure 1. Interferona regulates platelet production. (A to E) CD34+ hematopoietic stem cells and progenitor cells were isolated by immunomagnetic separation from human umbilical cord blood. Cells were cultured for 13 days in stem cells expansion media supplemented with thrombopoietin that promotes megakaryocyte (MK) differentiation. 1,000 units/milliliter (U/mL) of human interferon α (INFα) and phosphate-buffered saline (PBS) (used as a negative control) was added at day 9 and further incubated until day 13. All assays mentioned in panels A to E were performed at day 13. (A) MK proplatelet formation (PPF) was counted blinded as to the IFNa treatment. At least 200 cells were counted per culture (n=4). (B) Representative images of day 13 cultured MK treated with PBS or IFNa. Treated MK were plated on fibrinogen on day 12 overnight, and fixed with 4% paraformaldehyde, stained with Alexa Fluor 488 Phalloidin (green) and a nuclear stain, DAPI (blue). Images were taken by a confocal microscope at 40X oil objective lens. (C) Platelet-like particles (PLP) were collected from IFNa or PBS-treated MK cultures and stained with APC labeled anti-CD41a antibody at 37°C for 10 minute and measured by flow cytometry (n=3). PLP were gated based on human peripheral blood platelets. (D) MK were stained with APC-labeled CD41a and PE-labeled CD42a antibodies, and CD41a+ CD42a+ MK (a marker for MK maturation) were assessed by flow cytometry (n=3). (E) Cultured MK treated with IFNa or PBS were stained with APC-labeled CD41a and propidium iodide, and ploidy was assessed by flow cytometry (n=3). The quantification of the ploidy distribution is shown on the y-axis by calculating the percentage of cells with 2n, 4n, 8n and 16n. Apoptotic population were gated out. Statistical significance was determined by two-tailed paired t-test (A to E). Error bars represent mean ± standard error of mean. (F) 25,000 units of murine IFNa or PBS (negative control) were administered intraperitoneally in wild-type mice for consecutive three days (n=5 per group). On day 4, mice blood was harvested by cardiac puncture and platelet count was measured by Hemavet. Statistical significance was determined by two-tailed unpaired t-test with Welch’s correction. Error bars represent mean ± standard error of mean.

558

haematologica | 2022; 107(2)

Letters to the Editor

expression of MK sterile a motif (SAM) and histidineaspartate (HD) domain containing deoxynucleoside triphosphate triphophohydrolase 1 (SAMHD1) and (ii) SAMHD1 expression inhibits cultured human MK proplatelet formation (PPF) and promotes apoptosis. This is the first identification of SAMHD1 in human MK and report of a dNTP hydrolase regulating platelet production. In order to pursue studies on the effects of inflammation on megakaryocytopoiesis, we used CD34+ hematopoietic stem cells derived from human umbilical vein cord blood. IFNa significantly decreased day 13 PPF and platelet-like particles (Figure 1A to C), but did not affect the percentages of MK or polyploidy (Figure 1D to E). Importantly, we also showed that exogenous IFNa induces thrombocytopenia in wild-type mice (Figure 1F), consistent with studies in immunodeficient mice.4 In order to begin to understand how IFNa regulates latestage megakaryopoiesis and platelet production, we used an unbiased, transcriptome-wide approach and performed RNA sequencing (RNA-seq) on CD61-purified, day 13 cultured MK stimulated with IFNa. Our analyses identified 201 transcripts that were differentially expressed at a nominal significance threshold (P<0.05). Adjusting for multiple comparisons and setting a false discovery rate (FDR) threshold of <0.05, we found that 66 of the 201 transcripts were upregulated by IFNa (Online Supplementary Table S1). Increased mRNA expression in response to IFNa was validated by realtime polymerase chain reaction (PCR) analysis for all five genes tested (SAMHD1, PHF11, ISG20, IFITM3 and TAP2) (Online Supplementary Figure S1). Gene ontology analysis indicated that the differentially expressed genes were associated with the type 1 interferon signaling pathway, defense response to virus, and negative regulation of viral genome replication. Subsequent studies focused on SAMHD1, whose abundance increased more than 16-fold with IFNa induction (FDR=2.0x10-18)

(Figure 2A). Figure 2B and C shows that IFNa treatment of cultured MK greatly increased the abundance of SAMHD1 mRNA and protein (n=3 independent biological replicates). SAMHD1 is a hydrolase, the activated form of which degrades the intracellular pool of deoxynucleoside triphosphates (dNTPase) into deoxynucleosides and inorganic triphosphates, and is known to restrict viral replication of the human immunodeficiency virus type-1.8 In addition to viral restriction, SAMHD1 is required for cellular functions including replication fork progression, cell proliferation, apoptosis and DNA damage repair.9 IFNa stimulation induces SAMHD1 expression in human monocytes,10 astrocytes, microglia,11 HEK293T and HeLa cells,12 but there are no prior reports of SAMHD1 expression and/or function in MK or platelets. Platelet RNA And eXpression 1 (PRAX1) data13 demonstrated that SAMHD1 transcript levels are negatively associated with platelet count in healthy human subjects (Figure 3A), suggesting a possible inhibitory role of SAMHD1 in platelet production. Since SAMHD1 modulates the intracellular levels of dNTP, we hypothesized that an increase in the abundance SAMHD1 upon IFNa stimulation leads to decreased MK proliferation, maturation and DNA synthesis (MK polyploidy). However, deletion of SAMHD1 by CRISPR/Cas9 gene editing in cultures promoting unilineage MK differentiation (Figure 3B) did not affect MK maturation (Online Supplementary Figure 2A and B) or ploidy (Online Supplementary Figure 2C). This suggests SAMHD1 effects thrombopoiesis rather than megakaryocytopoiesis. Similar to Figure 1, IFNa stimulation caused a significant decrease in MK PPF MK without CRISPR modification (Figure 3C, first 2 bars). The effect of IFNa on PPF was abolished when SAMHD1 was deleted (Figure 3C, second 2 bars). Lastly, IFNa is well-established as pro-apoptotic.2 MK must restrain apoptosis to survive and progress safely through PPF and platelet shedding.14,15

Figure 2. Interferon α upregulates. dNTP hydrolase enzyme SAMHD1 in cultured megakaryocytes. CD34+ hematopoietic stem cells and progenitor cells were isolated by immunomagnetic separation from human umbilical cord blood. Cells were cultured in megakaryocyte (MK) promoting conditions by supplementation with thrombopoietin for 13 days. 1,000 units per milliliter (U/mL) of interferon α (INFα) and phosphate-buffered saline (PBS) (negative control) was added in 3 independent cultures at day 13 for 8 hours. (Dose and time course studies were performed in Meg-01 to select optimal concentrations and times, data not shown). CD61+ MK were separated by immunomagnetic beads and total RNA was obtained by mirVana kit. RNA samples with RIN score of >8 were used for RNA sequencing. The library preparation was done using TruSeq standard total RNA sample preparation kit with RNA depletion (Illumina Inc.). RNA sequensing was performed using NextSeq500 instrument with read length of 50 basepair single-end read. On average 68.9 million reads were obtained from each sample and genome mapping was on average 94.5% for all samples. (A) Volcano plot shows the relationship between the P-values and the fold change in normalized expression for IFNa or PBS treated cultured human MK. Differentially expressed genes (false discovery rate [FDR] <0.05) on IFNa treatment was plotted. SAMHD1 is shown in red. (B) Real-time polymerase chain reaction analysis validates the upregulation of SAMHD1 in IFNa-treated CD61+ MK compared to PBS control. Actin was used as housekeeping gene. Log2 fold change of SAMHD1 mRNA levels is plotted for IFNa vs. PBS control (n=3). Statistical significance was determined by one sample t-test. (C) Representative immunoblot shows increase in SAMHD1 protein in IFNa-treated MK compared to PBS control. Actin is used as loading control.

haematologica | 2022; 107(2)

559

Letters to the Editor

Figure 3. Interferon α-induced SAMHD1 expression regulates human megakaryocyte (MK) proplatelet formation and MK apoptosis. (A) Plot of Pearson correlation (R) between platelet count and SAMHD1 mRNA levels in 154 healthy donors. Dotted lines represent 95% confidence intervals. (B to D) CRISPR/Cas9 knock-down of SAMHD1 in CD34+ derived human umbilical cord cells were performed at day 3. Cells were cultured in MK differentiation promoting conditions, and treated with 1,000 U/mL interferon α (INFα) and phosphate-buffered saline (PBS) control. On day 13, MK treated with IFNa or PBS were assayed as mentioned below. (B) Representative immunoblot of SAMHD1 after CRISPR-Cas9 knock-down (denoted as “SAMHD1 cr” throughout the figure) in day 13 human CD34+ derived cultured MK with or without IFNa. Guide RNA not targeting known genes are used as negative control with or without IFNa (“neg cr” throughout the figure). Fold changes of densitometric quantification of SAMHD1 immunoblots, normalized to actin is shown on right (n=5). (C) The percentage of proplatelet formation (PPF) MK, scored blinded, for SAMHD1 cr vs neg cr, with or without IFNa is shown (n=6). (D) The mean fluorescence intensity (MFI) of annexin V binding (a marker of apoptosis-induced phosphatidylserine expression) on IFNa stimulation in SAMHD1 cr vs. neg cr MK, with or without IFNa is plotted (n=5). Statistical significance was determined by paired t-test (B to D). Error bars represent mean ± standard error of mean.

Although viral infections induce MK apoptosis,15 we are not aware of in vitro studies assessing IFNa-induced MK apoptosis. Under the culture conditions described in Figure 3C, we assessed annexin V binding as a measure of MK apoptosis, and observed a significant increase in response to IFNa stimulation (Figure 3D, first 2 bars). Next, because SAMHD1 also promotes apoptosis,9 we tested the effects of SAMHD1 deletion on IFNa-induced MK apoptosis. Deletion of SAMHD1 caused a modest reduction in annexin V binding (Figure 3D, bar 1 vs. bar 3), and significantly reduced IFNa-induced MK annexin V binding (Figure 3D, compare bars 2 and 4), supporting a role for SAMHD1 as a mediator of IFNa-induced MK apoptosis. In summary, our study indicates that IFNa leads to reduced platelet production and thrombocytopenia through apoptosis, and that IFNa-induced SAMHD1 is at least partially responsible for these effects on late560

stage platelet production by MK. Prior work has shown that expression of three candidate MK transcription factors is inhibited by IFNa4, and perhaps SAMHD1 is also regulated at a transcriptional level in MK. Post-transcriptional mechanisms may also be at play, since the enzymatic ability of SAMHD1 to maintain dNTP homeostasis in other cells requires protein phosphorylation. Future studies in MK will be need to address these issues. Seema Bhatlekar,1 Shancy Jacob,1 Emilia Tugolukova,1 Bhanu K. Manne,1 Yasuhiro Kosaka,1 Phillipe Loher,2 Ryan M. O’Connell,3 Vicente Planelles,4 Matthew T. Rondina,1,5 Isidore Rigoutsos5 and Paul F. Bray1,6 1 Program in Molecular Medicine and Department of Internal Medicine, University of Utah, Salt Lake City, UT; 2Computational Medicine Center, Thomas Jefferson University, Philadelphia, PA; 3 Division of Microbiology and Immunology, Department of Pathology, haematologica | 2022; 107(2)

Letters to the Editor

and Huntsman Cancer Institute, University of Utah Health Sciences Center, University of Utah, Salt Lake City, UT; 4Division of Microbiology & Immunology, Department of Pathology, University of Utah, Salt Lake City, UT; 5Department of Pathology, University of Utah, Salt Lake City, UT and the George E. Wahlen VAMC Department of Medicine and George E. Wahlen VAMC GRECC, Salt Lake City, UT and 6Division of Hematology and Hematologic Malignancies, Department of Internal Medicine, University of Utah, Salt Lake City, UT, USA. Correspondence: PAUL F. BRAY- Paul.bray@hsc.utah.edu doi:10.3324/haematol.2021.279864 Received: August 24, 2021 Accepted: October 28, 2021. Pre-published: November 11, 2021. Disclosures: no conflicts of interest to disclose. Contributions: SB, SJ, ET, BK, YK performed research; PL, IR performed RNA-Seq analysis; SB, PB wrote the manuscript; RO, VP, MR, IR, PF provided scientific expertise. Funding: this study was supported by grants from the National Institutes of Health National Heart, Lung and Blood Institute (HL116713, HL142804, HL145237, and HL130541), the National Institute of Allergy, Immunology and Infectious Diseases (AI143567), and the Division of Hematology and Hematologic Malignancies at the University of Utah. This work was also supported by Merit Review Award Number I01 CX001696 from the United States (U.S.) Department of Veterans Affairs Clinical Sciences R&D (CSRD). This material is the result of work supported with resources at the George E. Wahlen VA Medical Center, Salt Lake City, Utah. The contents do not represent the views of the U.S. Department of Veterans Affairs or the U.S. Government. The authors thank the University of Utah Flow Cytometry Facility in addition to the National Cancer Institute through Award Number 5P30CA042014-24.

References 1. Cunin P, Nigrovic PA. Megakaryocytes as immune cells. J Leukoc Biol. 2019;105(6):1111-1121.

haematologica | 2022; 107(2)

2. Borden EC, Sen GC, Uze G, et al. Interferons at age 50: past, current and future impact on biomedicine. Nat Rev Drug Discov. 2007;6(12):975-990. 3. Negrotto S, De Giusti CJ, Lapponi MJ, et al. Expression and functionality of type I interferon receptor in the megakaryocytic lineage. J Thromb Haemost. 2011;9(12):2477-2485. 4. Yamane A, Nakamura T, Suzuki H, Ito M, Ohnishi Y, Ikeda Y, Miyakawa Y. Interferon-alpha 2b-induced thrombocytopenia is caused by inhibition of platelet production but not proliferation and endomitosis in human megakaryocytes. Blood. 2008;112(3): 542-550. 5. Ganser A, Carlo-Stella C, Greher J, Volkers B, Hoelzer D. Effect of recombinant interferons alpha and gamma on human bone marrow-derived megakaryocytic progenitor cells. Blood. 1987;70(4): 1173-1179. 6. Mazur EM, Richtsmeier WJ, South K. Alpha-interferon: differential suppression of colony growth from human erythroid, myeloid, and megakaryocytic hematopoietic progenitor cells. J Interferon Res. 1986;6(3):199-206. 7. Wadenvik H, Kutti J, Ridell B, et al. The effect of alpha-interferon on bone marrow megakaryocytes and platelet production rate in essential thrombocythemia. Blood. 1991;77(10):2103-2108. 8. Laguette N, Sobhian B, Casartelli N, et al. SAMHD1 is the dendritic- and myeloid-cell-specific HIV-1 restriction factor counteracted by Vpx. Nature. 2011;474(7353):654-657. 9. Coggins SA, Mahboubi B, Schinazi RF, Kim B. SAMHD1 functions and human diseases. Viruses. 2020;12(4):382. 10. Berger A, Sommer AF, Zwarg J, et al. SAMHD1-deficient CD14+ cells from individuals with Aicardi-Goutieres syndrome are highly susceptible to HIV-1 infection. PLoS Pathog. 2011;7(12): e1002425. 11. Jin C, Peng X, Liu F, et al. Interferon-induced sterile alpha motif and histidine/aspartic acid domain-containing protein 1 expression in astrocytes and microglia is mediated by microRNA-181a. AIDS. 2016;30(13):2053-2064. 12. St Gelais C, de Silva S, Amie SM, et al. SAMHD1 restricts HIV-1 infection in dendritic cells (DCs) by dNTP depletion, but its expression in DCs and primary CD4+ T-lymphocytes cannot be upregulated by interferons. Retrovirology. 2012;9:105. 13. Edelstein LC, Simon LM, Montoya RT, et al. Racial differences in human platelet PAR4 reactivity reflect expression of PCTP and miR-376c. Nat Med. 2013;19(12):1609-1616. 14. Bhatlekar S, Basak I, Edelstein LC, et al. Anti-apoptotic BCL2L2 increases megakaryocyte proplatelet formation in cultures of human cord blood. Haematologica. 2019;104(10):2075-2083. 15. Josefsson EC, James C, Henley KJ, et al. Megakaryocytes possess a functional intrinsic apoptosis pathway that must be restrained to survive and produce platelets. J Exp Med. 2011;208(10):20172031.

561

Letters to the Editor

Detection of ABL1 kinase domain mutations in therapy-naïve BCR-ABL1-positive acute lymphoblastic leukemia Mutations in the ABL1 kinase domain are the main mechanism of resistance to tyrosine kinase inhibitors (TKI) in Philadelphia chromosome-positive (Ph+) leukemia. In acute lymphoblastic leukemia (ALL), a very early acquisition of mutations can be observed and reports have described that mutations may exist already before TKI treatment.1-4 Initial data came from cloning or allele-specific polymerase chain reaction assays,5 which are labor-intensive, do not detect all possible mutations and especially come with the doubt of detecting amplification artefacts. Despite the first description over 18 years ago, ABL1 mutation screening is not a standard test for ALL patients before initiation of TKI treatment. The current broad availability of highly sensitive next-generation sequencing approaches enables the initial limitations to be overcome. In this study, we used next-generation sequencing to screen 91 BCR-ABL1-positive ALL patients for mutations before therapy and identified five (5.5%) with ABL1 kinase domain mutations at initial diagnosis. We studied the diagnostic samples of 39 female and 52 male ALL patients diagnosed between April 2007 and November 2016. The median age was 61 years (range, 18-84), and treatment included first- to third-generation TKI (Online Supplementary Table S1). Thirty-five patients had the M-bcr (p210) transcript and 56 the m-bcr (p190) transcript. The study was approved by the Internal Review Board and by the Bavarian Ethics Commitee, the Bavarian State Medical Association (Bayerische Landesärztekammer) with number 05117. The study adhered to the tenets of the Declaration of Helsinki. RNA for BCR-ABL1 detection and mutation analysis was isolated from bone marrow (n=56) or peripheral blood (n=35) with standard protocols and cDNA synthesis was performed using SuperScript™ IV VILO™ (Thermo Fisher Scientific, Waltham, MA, USA). Six amplicons for sequencing on the MiSeq (Illumina, San Diego, CA, USA) covered amino acids 184-510 of ABL1 and were generated from pre-amplified BCR-ABL1. A 1% detection limit was used for analysis with SeqNext (JSI Medical Systems, Kippenheim, Germany). Using next-generation sequencing, we found wellcharacterized ABL1 kinase domain mutations in five of the 91 (5.5%) diagnostic samples (Table 1, Online Supplementary Table S1). Mutations were confirmed in a second independent sequencing run to exclude potential artefacts from random polymerase chain reaction errors. Four patients with a mutation at diagnosis had the m-bcr

transcript (4/56 [7%]) and one was found in a patient with M-bcr (1/35 [3%]). The mutations are present in subclones (<10% of BCR-ABL1-positive cells) in patients #2, #4 and #5 (Table 1). For patient #1, the Q252H (cag>cat) mutation had expanded to 97% of BCR-ABL1 transcripts at relapse after 5 months of treatment with the GMALL elderly protocol including imatinib, to which the mutation causes resistance (Figure 1). Patients #2 and #3 were both transplanted as part of standard of care at the respective time. Patient #3 (T315I) had achieved complete remission; however, the T315I mutation would be particularly severe because it causes resistance to four TKI (imatinib, nilotinib, dasatinib and bosutinib).6 Follow-up data for patients #4 and #5 were not available. Next, we focused on patients who had a molecular follow-up of at least 6 months (n=35) for longitudinal mutation testing. This excluded patients who died within the first 6 months or those from whom no sample was available for BCR-ABL1 expression and ABL1 mutation testing. In 19 of 35 cases, mutation analysis was considered relevant during follow-up due to failure of primary therapy or relapse (defined as a BCR-ABL1/ABL1 ratio of at least ~1%) (Online Supplementary Table S1). Mutation testing was performed between 4 and 71 months after diagnosis (median: 12 months). One or more known resistance mutations were found in 15 of 19 (79%) Ph+ ALL cases with relapse or increased BCR-ABL1/ABL1 expression despite treatment (Online Supplementary Table S1). This is in line with the previously published ABL1 mutation frequency of 70-80% for relapsed Ph+ ALL.7 Except for the Q252H mutation in patient #1 (Table 1), we did not detect any of the other resistance mutations from the relapse sample at the diagnostic time point with a 1% sensitivity cutoff. For comparison, Soverini et al. used a relapse-to-diagnosis backtracking approach to study resistance mutations and found one out of 34 patients with a mutation (Y253H) already at the initial time point.8 While selective pressure of TKI treatment makes the outgrowth of clones with resistance mutations very plausible, the outgrowth in therapy-naïve cells requires a fitness advantage. In previously published biochemical or cell culture studies, the effect of mutations without TKI was evaluated. D276G increased catalytic activity,9 and E255K exhibited increased transformation potency.10 Although T315I showed reduced kinase activity compared to unmutated controls, cell culture assays suggested increased oncogenic potency via different pathways (e.g., phosphorylation).11 In a recent study of pediatric Ph-like ALL (with EBF1-PDGFRB rearrangement), Tran and colleagues back-tracked the gatekeeper T681I mutation in PDGFRB, which is the analog of the T315I muta-

Table 1. Characteristics of patients with mutations at initial diagnosis.

Patient Gender Age Blasts Transcript Mutation Treatment Outcome

male 72 years 91% p190 Q252H (G>T): 46% Q252H (G>C): 8% GMALL elderly remission followed by relapse

female 67 years 88% p190 D276G: 3%

female 52 years 12% p190 T315I: 14%

male 79 years 85% p210 M244V: 2%

male 76 years 85% p190 E255K: 1%

alloSCT NA

alloSCT CR

NA NA

GMALL: protocol of the German Multicenter Study Group on Adult Acute Lymphoblastic Leukemia; alloSCT: allogeneic stem cell transplantation; CR: complete remission; NA: data not available.

562

haematologica | 2022; 107(2)

Letters to the Editor

Constance Baer, Manja Meggendorfer, Claudia Haferlach, Wolfgang Kern and Torsten Haferlach MLL Munich Leukemia Laboratory, Munich, Germany Corresponding author: CONSTANCE BAER - constance.baer@mll.com doi:10.3324/haematol.2021.279807 Received: August 17, 2021. Accepted: November 3, 2021. Pre-published: November 11, 2021. Disclosures: CB and MM are employees in MLL Munich Leukemia Laboratory; WK, CH and TH own equity in MLL Munich Leukemia Laboratory. Contributions: CB and MM analyzed the data; CB, CH, WK and TH designed the study; CB and TH wrote the manuscript.

References Figure 1. Molecular monitoring of BCR-ABL1 transcripts and ABL1 point mutations in a patient with Philadelphia chromosome-positive acute lymphoblastic leukemia. (A) Patient #1 was treated according to the GMALL elderly protocol including imatinib. BCR-ABL1 expression (BCR-ABL/ABL1 in %) was measured by quantitative polymerase chain reaction. (B) The variant allele frequency (VAF in %) is given for two follow-up time-points.

tion in ABL1. They identified the mutation in three of 23 patients prior to treatment.12 We speculate that mutations which confer resistance but reduce kinase activity or transformative capacity would not outgrow before TKI treatment and therefore only a subset of resistance mutations can exist in therapy-naïve patients. The high turnover rate of ALL cells should allow a much faster selection of mutated clones with a relative growth advantage if compared to the situation in chronic myeloid leukemia. In an initial dataset of treatmentnaïve patients with chronic phase chronic myeloid leukemia, only minor subclones (<1%) were identified, and showed no correlation with endpoints.13 In summary, known resistance mutations in the ABL1 kinase domain were detected in five of 91 (5.5%) therapy-naïve patients with BCR-ABL1-positive ALL. For patient #1 (Figure 1) we can show that the mutated clone expands rapidly if treatment with an insensitive TKI is chosen. At present, we see no significant difference in outcome for the five patients with mutations, but the heterogeneity of the cohort and the small number of cases with mutations need to be considered. Further studies will be necessary, especially involving the full spectrum of TKI, which are now increasingly a backbone and guarantor of success of improving outcomes in Ph+ disease. Next-generation sequencing is a direct and sensitive (1%) strategy to identify patients at risk of resistance before any TKI therapy is started. However, 93% (14/15 patients) of mutations found in relapse samples were most likely acquired under the selective pressure of (TKI) treatment. Therefore, testing at initial diagnosis should only be considered in addition to established mutation testing in refractory/relapsed disease.

haematologica | 2022; 107(2)

1. Soverini S, Vitale A, Poerio A, et al. Philadelphia-positive acute lymphoblastic leukemia patients already harbor BCR-ABL kinase domain mutations at low levels at the time of diagnosis. Haematologica. 2011;96(4):552-557. 2. Pfeifer H, Wassmann B, Pavlova A, et al. Kinase domain mutations of BCR-ABL frequently precede imatinib-based therapy and give rise to relapse in patients with de novo Philadelphia-positive acute lymphoblastic leukemia (Ph+ ALL). Blood. 2007;110(2):727-734. 3. Hofmann WK, Komor M, Wassmann B, et al. Presence of the BCRABL mutation Glu255Lys prior to STI571 (imatinib) treatment in patients with Ph+ acute lymphoblastic leukemia. Blood. 2003;102(2):659-661. 4. Hato T, Yamanouchi J, Tamura T, et al. Existence of leukemic clones resistant to both imatinib mesylate and rituximab before drug therapies in a patient with Philadelphia chromosome-positive acute lymphocytic leukemia. Int J Hematol. 2004;80(1):62-66. 5. Willis SG, Lange T, Demehri S, et al. High-sensitivity detection of BCR-ABL kinase domain mutations in imatinib-naive patients: correlation with clonal cytogenetic evolution but not response to therapy. Blood. 2005;106(6):2128-2137. 6. Cortes JE, Kantarjian H, Shah NP, et al. Ponatinib in refractory Philadelphia chromosome-positive leukemias. N Engl J Med. 2012;367(22):2075-2088. 7. Soverini S, De Benedittis C, Papayannidis C, et al. Drug resistance and BCR-ABL kinase domain mutations in Philadelphia chromosome-positive acute lymphoblastic leukemia from the imatinib to the second-generation tyrosine kinase inhibitor era: the main changes are in the type of mutations, but not in the frequency of mutation involvement. Cancer. 2014;120(7):1002-1009. 8. Soverini S, De Benedittis C, Papayannidis C, et al. Clinical impact of low-burden BCR-ABL1 mutations detectable by amplicon deep sequencing in Philadelphia-positive acute lymphoblastic leukemia patients. Leukemia. 2016;30(7):1615-1619. 9. Piazza RG, Magistroni V, Gasser M, et al. Evidence for D276G and L364I Bcr-Abl mutations in Ph+ leukaemic cells obtained from patients resistant to imatinib. Leukemia. 2005;19(1):132-134. 10. Griswold IJ, MacPartlin M, Bumm T, et al. Kinase domain mutants of Bcr-Abl exhibit altered transformation potency, kinase activity, and substrate utilization, irrespective of sensitivity to imatinib. Mol Cell Biol. 2006;26(16):6082-6093. 11. Skaggs BJ, Gorre ME, Ryvkin A, et al. Phosphorylation of the ATPbinding loop directs oncogenicity of drug-resistant BCR-ABL mutants. Proc Natl Acad Sci U S A. 2006;103(51):19466-19471. 12. Tran TH, Nguyen JV, Stecula A, et al. The EBF1-PDGFRB T681I mutation is highly resistant to imatinib and dasatinib in vitro and detectable in clinical samples prior to treatment. Haematologica. 2021;106(8):2242-2245. 13. Franke G-N, Maier J, Wildenberger K, et al. Incidence of low level mutations in newly diagnosed CML patients: a substudy of the German Tiger Trial. Blood. 2017;130(Suppl 1):252

563

Haematologica, Volume 107, Issue 2

Articles inside

Integrating genetic and epigenetic factors in chronic myeloid leukemia risk assessment: toward gene expression-based biomarkers

Comprehensive analysis of dose intensity of acute lymphoblastic leukemia chemotherapy

Too much and not enough: revisiting maintenance rituximab in indolent lymphomas

Are clinical pharmacology studies still needed in childhood acute lymphoblastic leukemia?

An exciting RXRA mutant revives interest in retinoids for acute myeloid leukemia