Haematologica, Volume 103, issue 2 by Haematologica

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation Editor-in-Chief Luca Malcovati (Pavia)

Managing Director Antonio Majocchi (Pavia)

Associate Editors Hélène Cavé (Paris), Pavan Reddy (Ann Arbor), Andreas Rosenwald (Wuerzburg), Monika Engelhardt (Freiburg), Wyndham Wilson (Bethesda), Paul Kyrle (Vienna), Swee Lay Thein (Bethesda), Pieter Sonneveld (Rotterdam)

Assistant Editors Anne Freckleton (English Editor), Cristiana Pascutto (Statistical Consultant), Rachel Stenner (English Editor), Kate O’Donohoe (English Editor), Ziggy Kennell (English Editor)

Editorial Board Omar I. Abdel-Wahab (New York); Jeremy Abramson (Boston); Paolo Arosio (Brescia); Raphael Bejar (San Diego); Erik Berntorp (Malmö); Dominique Bonnet (London); Jean-Pierre Bourquin (Zurich); Suzanne Cannegieter (Leiden); Francisco Cervantes (Barcelona); Nicholas Chiorazzi (Manhasset); Oliver Cornely (Köln); Michel Delforge (Leuven); Ruud Delwel (Rotterdam); Meletios A. Dimopoulos (Athens); Inderjeet Dokal (London); Hervé Dombret (Paris); Peter Dreger (Hamburg); Martin Dreyling (München); Kieron Dunleavy (Bethesda); Dimitar Efremov (Rome); Sabine Eichinger (Vienna); Jean Feuillard (Limoges); Carlo Gambacorti-Passerini (Monza); Guillermo Garcia Manero (Houston); Christian Geisler (Copenhagen); Piero Giordano (Leiden); Christian Gisselbrecht (Paris); Andreas Greinacher (Greifswals); Hildegard Greinix (Vienna); Paolo Gresele (Perugia); Thomas M. Habermann (Rochester); Claudia Haferlach (München); Oliver Hantschel (Lausanne); Christine Harrison (Southampton); Brian Huntly (Cambridge); Ulrich Jaeger (Vienna); Elaine Jaffe (Bethesda); Arnon Kater (Amsterdam); Gregory Kato (Pittsburg); Christoph Klein (Munich); Steven Knapper (Cardiff); Seiji Kojima (Nagoya); John Koreth (Boston); Robert Kralovics (Vienna); Ralf Küppers (Essen); Ola Landgren (New York); Peter Lenting (Le Kremlin-Bicetre); Per Ljungman (Stockholm); Francesco Lo Coco (Rome); Henk M. Lokhorst (Utrecht); John Mascarenhas (New York); Maria-Victoria Mateos (Salamanca); Simon Mendez-Ferrer (Madrid); Giampaolo Merlini (Pavia); Anna Rita Migliaccio (New York); Mohamad Mohty (Nantes); Martina Muckenthaler (Heidelberg); Ann Mullally (Boston); Stephen Mulligan (Sydney); German Ott (Stuttgart); Jakob Passweg (Basel); Melanie Percy (Ireland); Rob Pieters (Utrecht); Stefano Pileri (Milan); Miguel Piris (Madrid); Andreas Reiter (Mannheim); Jose-Maria Ribera (Barcelona); Stefano Rivella (New York); Francesco Rodeghiero (Vicenza); Richard Rosenquist (Uppsala); Simon Rule (Plymouth); Claudia Scholl (Heidelberg); Martin Schrappe (Kiel); Radek C. Skoda (Basel); Gérard Socié (Paris); Kostas Stamatopoulos (Thessaloniki); David P. Steensma (Rochester); Martin H. Steinberg (Boston); Ali Taher (Beirut); Evangelos Terpos (Athens); Takanori Teshima (Sapporo); Pieter Van Vlierberghe (Gent); Alessandro M. Vannucchi (Firenze); George Vassiliou (Cambridge); Edo Vellenga (Groningen); Umberto Vitolo (Torino); Guenter Weiss (Innsbruck).

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

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

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation

Information for readers, authors and subscribers Haematologica (print edition, pISSN 0390-6078, eISSN 1592-8721) publishes peer-reviewed papers on all areas of experimental and clinical hematology. The journal is owned by a non-profit organization, the Ferrata Storti Foundation, and serves the scientific community following the recommendations of the World Association of Medical Editors (www.wame.org) and the International Committee of Medical Journal Editors (www.icmje.org). Haematologica publishes editorials, research articles, review articles, guideline articles and letters. Manuscripts should be prepared according to our guidelines (www.haematologica.org/information-for-authors), and the Uniform Requirements for Manuscripts Submitted to Biomedical Journals, prepared by the International Committee of Medical Journal Editors (www.icmje.org). Manuscripts should be submitted online at http://www.haematologica.org/. Conflict of interests. According to the International Committee of Medical Journal Editors (http://www.icmje.org/#conflicts), “Public trust in the peer review process and the credibility of published articles depend in part on how well conflict of interest is handled during writing, peer review, and editorial decision making”. The ad hoc journal’s policy is reported in detail online (www.haematologica.org/content/policies). Transfer of Copyright and Permission to Reproduce Parts of Published Papers. Authors will grant copyright of their articles to the Ferrata Storti Foundation. No formal permission will be required to reproduce parts (tables or illustrations) of published papers, provided the source is quoted appropriately and reproduction has no commercial intent. Reproductions with commercial intent will require written permission and payment of royalties. Detailed information about subscriptions is available online at www.haematologica.org. Haematologica is an open access journal. Access to the online journal is free. Use of the Haematologica App (available on the App Store and on Google Play) is free. For subscriptions to the printed issue of the journal, please contact: Haematologica Office, via Giuseppe Belli 4, 27100 Pavia, Italy (phone +39.0382.27129, fax +39.0382.394705, E-mail: info@haematologica.org). Rates of the International edition for the year 2018 are as following: Print edition

Institutional Euro 600

Personal Euro 150

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. Edoardo Ascari; 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)

haematologica calendar of events

Journal of the European Hematology Association Published by the Ferrata Storti Foundation The 11th Annual Congress of the European Association for Haemophilia and Allied Disorders 2018 European Association for Haemophilia and Allied Disorders Chairs: C Hermans, M Makris, V Jimenez Juste Madrid, Spain February 7-9, 2018 EHA-SWG Scientific Meeting on Integrated Diagnosis Strategies in Oncohematology for the management of cytopenias and leukocytosis Chair: MC Béné February 8-10, 2018 Barcelona, Spain EHA-ISHBT Hematology Tutorial on Lymphoproliferative and Plasma Cell Disorders February 16-18, 2018 Lucknow, India EuroClonality Workshop: Clonality assessment in Pathology European Scientific foundation for Laboratory Hemato Oncology (ESLHO) Chairs: PJTA Groenen, F Fend, AW Langerak February 19-21, 2018 Nijmegen, The Netherlands

1st European Myeloma Network Meeting Società Italiana di Ematologia (SIE) Chair: M Boccadoro April 19-21, 2018 Torino, Italy EHA-TSH Hematology Tutorial on Acute Leukemias April 28-29, 2018 Istanbul, Turkey EHA Hematology Tutorial on Thalassemia May 10-11, 2018 Shiraz, Iran 23rd Congress of EHA June 14-17, 2018 Stockholm, Sweden EHA-SAH Hematology Tutorial on lymphoid Malignancies and Plasma Cell Dyscrasias September 14-15, 2018 Buenos Aires, Argentina EHA-SWG Scientific Meeting on Aging and Hematology Chair: D Bron October 12-14, 2018 Location TBC

ESH Clinical Updates on CLL and Indolent Lymphoma European School of Haematology (ESH) Chairs: C Buske, C Wu, PL Zinzani March 2-4, 2018 Paris, France EHA-SWG Scientific Meeting on New Molecular Insights and Innovative Management Approaches for Acute Lymphoblastic Leukemia Chair: N Gökbuget April 12-14, 2018 Barcelona, Spain

Calendar of Events updated on January 15, 2018

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation

Table of Contents Volume 103, Issue 2: February 2018 Cover Figure Image generated by www.somersault1824.com.

Editorials 191

Allogeneic stem cell transplantation for FLT3 mutated acute myeloid leukemia in first complete remission: does age really matter? Stefan. O. Ciurea

193

Large granular lymphocyte cells and immune dysregulation diseases – the chicken or the egg? Anton W. Langerak and Jorn L.J.C. Assmann

195

Hemophilia A: different phenotypes may be explained by multiple and variable effects of the causative mutation in the F8 gene Giancarlo Castaman

Review Article 197

From transplant to novel cellular therapies in multiple myeloma: European Myeloma Network guidelines and future perspectives Francesca Gay et al.

Articles Bone Marrow Failure

212

Nationwide survey on the use of eltrombopag in patients with severe aplastic anemia: a report on behalf of the French Reference Center for Aplastic Anemia Etienne Lengline et al.

221

Rational management approach to pure red cell aplasia Suresh Kumar Balasubramanian et al.

Phagocyte Biology and Its Disorders

231

The clinical and laboratory evaluation of familial hemophagocytic lymphohistiocytosis and the importance of hepatic and spinal cord involvement: a single center experience Burcin Beken et al.

Myelodysplastic Syndrome

237

Acute Myeloid Leukemia

246

Micro-ribonucleic acid-155 is a direct target of Meis1, but not a driver in acute myeloid leukemia Edith Schneider et al.

256

Haematologica 2018; vol. 103 no. 2 - February 2018 http://www.haematologica.org/

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation Acute Lymphoblastic Leukemia

266

WT1 loss attenuates the TP53-induced DNA damage response in T-cell acute lymphoblastic leukemia Fulvio Bordin et al.

Non-Hodgkin Lymphoma

278

Epstein-Barr virus-associated primary nodal T/NK-cell lymphoma shows a distinct molecular signature and copy number changes Siok-Bian Ng et al.

288

A bioclinical prognostic model using MYC and BCL2 predicts outcome in relapsed/refractory diffuse large B-cell lymphoma Mark Bosch et al.

297

Outcomes among North American patients with diffuse large B-cell lymphoma are independent of tumor Epstein-Barr virus positivity or immunosuppression Sean I. Tracy et al.

Lymphoproliferative Disorders

304

Somatic STAT3 mutations in Felty syndrome: an implication for a common pathogenesis with large granular lymphocyte leukemia Paula Savola et al.

Chronic Lymphocytic Leukemia

313

Expression of COBLL1 encoding novel ROR1 binding partner is robust predictor of survival in chronic lymphocytic leukemia Hana Plešingerová et al.

Plasma Cell Disorders

325

Maternal embryonic leucine zipper kinase is a novel target for proliferation-associated high-risk myeloma Arnold Bolomsky et al.

336

Prognostic significance of tumor burden assessed by whole-body magnetic resonance imaging in multiple myeloma patients treated with allogeneic stem cell transplantation Jennifer Mosebach et al.

Coagulation & Its DIsorders

344

Clustered F8 missense mutations cause hemophilia A by combined alteration of splicing and protein biosynthesis and activity Irving Donadon et al.

351

Complement C3 is a novel modulator of the anti-factor VIII immune response Julie Rayes et al.

Blood Transfusion

361

Hypoxia modulates the purine salvage pathway and decreases red blood cell and supernatant levels of hypoxanthine during refrigerated storage Travis Nemkov et al.

Obituary 373

Remembering Professor Felice Gavosto (February 16, 1921 – December 11, 2017) Federico Caligaris Cappio, Massimo Aglietta, Clara Camaschella, Giuseppe Saglio and Robin Foà

Haematologica 2018; vol. 103 no. 2 - February 2018 http://www.haematologica.org/

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation

Letters to the Editor Letters are available online only at www.haematologica.org/content/103/2.toc

e42

Alternatively spliced fibronectin extra domain A is required for hemangiogenic recovery upon bone marrow chemotherapy Alessandro Malara et al. http://www.haematologica.org/content/103/2/e42

e46

Short-term administration of JAK2 inhibitors reduces splenomegaly in mouse models of b-thalassemia intermedia and major Carla Casu et al. http://www.haematologica.org/content/103/2/e46

e50

Inflammatory molecule reduction with hydroxyurea therapy in children with sickle cell anemia Rhiannon R. Penkert et al. http://www.haematologica.org/content/103/2/e50

e55

MDS1 and EVI1 complex locus (MECOM): a novel candidate gene for hereditary hematological malignancies Tim Ripperger et al. http://www.haematologica.org/content/103/2/e55

e59

Knockdown of TP53 in ASXL1 negative background rescues apoptotic phenotype of human hematopoietic stem and progenitor cells but without overt malignant transformation Susan Hilgendorf and Edo Vellenga http://www.haematologica.org/content/103/2/e59

e63

JAK2, CALR, MPL and ASXL1 mutational status correlates with distinct histological features in Philadelphia chromosome-negative myeloproliferative neoplasms Waihay J. Wong et al. http://www.haematologica.org/content/103/2/e63

e69

Outcomes after allogeneic hematopoietic stem cell transplantation in patients with acute myeloid leukemia harboring t(7;11)(p15;p15) Kaito Harada et al. http://www.haematologica.org/content/103/2/e69

e73

High PDGFRA expression does not serve as an effective therapeutic target in ERG-deleted B-cell precursor acute lymphoblastic leukemia Isabel S. Jerchel et al. http://www.haematologica.org/content/103/2/e73

e78

GA101 P329GLALA, a variant of obinutuzumab with abolished ADCC, ADCP and CDC function but retained cell death induction, is as efficient as rituximab in B-cell depletion and antitumor activity Sylvia Herter et al. http://www.haematologica.org/content/103/2/e78

Case Reports Case Reports are available online only at www.haematologica.org/content/103/2.toc

e82

Worldwide study of hematopoietic allogeneic stem cell transplantation in pyruvate kinase deficiency Stephanie van Straaten et al. http://www.haematologica.org/content/103/2/e82

e87

A novel AGGF1-PDGFRb fusion in pediatric T-cell acute lymphoblastic leukemia Matthew S. Zabriskie et al. http://www.haematologica.org/content/103/2/e87

e92

Promising activity of selinexor in the treatment of a patient with refractory diffuse large B-cell lymphoma and central nervous system involvement Sabela Bobillo et al. http://www.haematologica.org/content/103/2/e92

Haematologica 2018; vol. 103 no. 2 - February 2018 http://www.haematologica.org/

EDITORIALS Allogeneic stem cell transplantation for FLT3 mutated acute myeloid leukemia in first complete remission: does age really matter? Stefan. O. Ciurea The University of Texas MD Anderson Cancer Center, Houston, TX, USA E-mail: sciurea@mdanderson.org doi:10.3324/haematol.2017.186346

ince its first discovery1 and initial report on the prognostic impact in patients with FMS-related tyrosine kinase 3 (FLT3)-mutated acute myeloid leukemia (AML),2 a large amount of data has been accumulated which have helped devise the best therapeutic approach for patients with this disease. Constitutive activation of FLT3 by internal tandem duplications (ITD) occurs in approximately 20-30% of patients with cytogenetically normal (diploid) AML (CNAML), the most frequent molecular aberration in patients with AML, while the less common mutations (7%) are those found in the tyrosine kinase domain (FLT3-TKD).3,4 The presence of FLT3-ITD mutations is widely accepted as a poor prognostic factor in CN-AML owing to its chemoresistance, high risk of relapse and short relapse-free survival (RFS), whereas the prognostic impact of FLT3-TKD mutation remains unclear.3,5-7 Although evidence from a large meta-analysis indicated that patients with either cytogenetic high- or intermediaterisk AML benefit from allogeneic hematopoietic stem cell transplantation (AHCT),8 until recently, the role of AHCT in patients with FLT3-ITD-mutated AML remained a matter of debate, as post-transplant outcomes were inconsistent between studies. In a study by Gale et al., the outcomes of adult AML patients treated according to the United Kingdom Medical Research Council (UK MRC) protocols were analyzed. Results from the donor-versusâ&#x20AC;&#x201C;no donor analysis of patients with FLT3-ITD-mutated AML showed a significantly lower relapse rate in patients with a donor, but overall survival (OS) was not significantly improved when compared with the no donor group. The authors concluded that the presence of an FLT3-ITD mutation should not influence the decision to proceed to transplantation. However, in total only a small number of patients received an allograft, and only 37 of the 68 FLT3-ITD-positive patients (54%) with donors actually received an allograft in first complete remission (CR1) in this study. Moreover, this analysis may be subject to selection bias as there was no direct comparison between FLT3-ITD patients receiving allografts and those receiving chemotherapy alone.9 On the contrary, several more recent studies indicate that AHCT is likely the best consolidation therapy for patients with FLT3-ITD-mutated AML, and should be performed as soon as possible in CR1.10-13 In a study by DeZern and colleagues, there was significantly better RFS of FLT3-ITDmutated AML patients treated with AHCT as compared to the non-transplant group (54 months vs. 8.6 months),12 while a study from the MD Anderson Cancer Center, which compared post-remission treatment with consolidation chemotherapy and AHCT in 227 FLT3-mutated AML patients who achieved CR1 after induction chemotherapy, showed that AHCT reduced the risk of relapse and haematologica | 2018; 103(2)

improved both RFS and OS regardless of NPM1 status and FLT3 allelic ratio.10 Moreover, our group analyzed the outcomes of 200 FLT3mutated AML patients (either ITD or TKD mutations) treated with AHCT with various donor types, including haploidentical donor transplants.11 This study showed a dramatic increase in the relapse rate and progressively worse progression-free survival (PFS) for patients transplanted beyond CR1, suggesting, again, that patients benefit the most from receiving ASCT in first remission, and that the lack of a human leukocyte antigen (HLA)-matched donor should not be a limitation to transplantation, as haploidentical transplants had similar survival with HLA-matched donor transplants.11 Albeit several advances have been made to improve outcome after AHCT, mortality related to the procedure is still a major concern. The initial hope was that the development of FLT3 inhibitors would have provided a dramatic effect on FLT3-mutated AML, and perhaps delay or reduce the need for transplantation, similar to the effect of tyrosine kinase inhibitors in patients with chronic myeloid leukemia; however, this was not realized. Thus far, several FLT3 inhibitors have been studied in FLT3-mutated AML as part of induction, consolidation as well as post-transplant maintenance therapy.14,15 Most recently, the results of a randomized, placebo-controlled trial of induction and consolidation chemotherapy with or without midostaurin for newly diagnosed FLT3-mutated AML patients (the RATIFY study) indicated a survival benefit for patients receiving midostaurin, and resulted in the FDA approval of this drug, in combination with chemotherapy, for induction and consolidation treatment of those with newly diagnosed FLT3-mutated AML. Although AHCT was not mandated in this protocol, more than half of patients received AHCT at some point during the disease course. Even though patients were not randomized to receive AHCT, results from the analysis, starting from time at transplant, showed a remarkable difference in the survival of midostaurin-treated patients who underwent AHCT in CR1 compared to those on the placebo arm, suggesting not only that midostaurin associated with induction chemotherapy might help provide deeper responses but may also improve transplant outcomes, especially for those who received transplantation in first remission.15 Nevertheless, as of yet there is no evidence to demonstrate that midostaurin, or any other FLT3 inhibitors, can provide survival benefit over AHCT in FLT3-ITD-mutated AML. Taken together, the available evidence suggests that AHCT ameliorates the prognostic impact of FLT3-ITD mutations, and is the preferred consolidation treatment for younger AML patients with FLT3-ITD mutations after achieving CR1. Transplantation for older patients is being increasingly performed worldwide. Data from the Center for International 191

Editorials

Blood and Marrow Transplant Research (CIBMTR) showed an increased number of transplants for older patients,16 including alternative donor transplants,17 with outcomes similar to those of HLA-matched donors.18 Multiple studies have demonstrated that AHCT can provide long-term survival benefit in elderly AML patients with high risk for disease relapse, and advanced age per se should not be used as a contraindication for AHCT. Questions remain, however, as to whether older patients with FLT3-ITD-mutated AML would benefit from AHCT in CR1, considering the fact that AML is a disease that is most often encountered in the older population and the majority of reports on AHCT outcomes of FLT3-ITDmutated AML were performed in patients younger than 60 years. In this issue of Haematologica, PoirĂŠ and colleagues examined the role of AHCT in a large cohort of FLT3ITD-mutated AML patients, aged 60 or over, reported on behalf of the Acute Leukemia Working Party (ALWP) of the European Group for Blood and Marrow Transplantation (EBMT).19 In addition to a very promising long-term survival posttransplant result of 56% at two years for patients in CR1, this study has brought up some compelling findings regarding FLT3-ITD-mutated AML in the elderly that are worth highlighting. First, increasing age as well as conditioning regimen intensity did not seem to influence nonrelapse mortality (NRM). In this cohort, the majority of patients received a reduced-intensity conditioning regimen (82%), with an acceptable NRM at 2 years of only 20% for all patients (18% for patients in CR1), which is particularly low for a registry-based study. This implies that AHCT is feasible for older patients with FLT3-ITDmutated AML, and adds to a growing body of evidence that AHCT should not be denied because of advanced age alone. Not surprisingly, disease status at transplant was strongly associated with a higher risk of relapse and worse survival, similar to the findings in younger patients,11 while the best outcomes were seen in patients in molecular remission before transplant and those transplanted within 42 days of diagnosis.19 In this cohort, approximately two-thirds of patients relapsed within two years when transplanted beyond first remission. These results confirm the importance of early AHCT, which should be performed without delay once morphologic remission is achieved. Nevertheless, the benefit of AHCT for those beyond first remission or with relapse/refractory diseases was less pronounced due to a very high relapse rate. The incorporation of FLT3 inhibitors before and/or after transplant might help improve the outcomes of these patients. Unfortunately, the RATIFY study on the efficacy of midostaurin only included patients younger than 60 years,15 whereas the use of sorafenib in combination with induction chemotherapy for an elderly group of patients seemed to be associated with a lower remission rate and unacceptable toxicity.20 We hope that newer FLT3 inhibitors will improve the safety profile and further enhance outcomes in these patients. In spite of these encouraging outcomes, several issues have not been addressed in the current study: the effect of FLT3 allelic ratio on transplant outcomes of elderly 192

patients with FLT3-ITD-mutated AML, since evidence indicates that it can influence post-transplant outcomes in younger patients,13 outcomes of patients with FLT3-TKD mutations, which were not included in this study, and the use of haploidentical donor transplants, as more patients are transplanted with a haploidentical donor when a HLA-matched donor is not available. In conclusion, although there are no prospective randomized studies to specifically compare the role of allogeneic transplantation with conventional chemotherapy for FLT3-ITD AML patients in CR1, sufficient evidence from several retrospective analyses clearly suggests that allogeneic stem cell transplantation can provide survival benefit in all age groups, which now includes older patients, and should be considered as a preferred consolidation strategy at least for patients with FLT3-ITDmutated AML in CR1. Future studies are needed to clarify the impact of hematopoietic cell transplantation comorbidity index (HCT-CI) on the survival of these patients, the role of FLT3 inhibitors with initial therapy as well as in post-transplant maintenance therapy in the older age group as well as the use of other therapeutic approaches like natural killer (NK) cell therapy, to further improve the outcomes of these patients, especially those with more advanced disease.

References 1. Nakao M, Yokota S, Iwai T, et al. Internal tandem duplication of the flt3 gene found in acute myeloid leukemia. Leukemia. 1996; 10(12): 1911-8. 2. Kiyoi H, Naoe T, Nakano Y, et al. Prognostic implication of FLT3 and N-RAS gene mutations in acute myeloid leukemia. Blood. 1999; 93(9): 3074-80. 3. Thiede C, Steudel C, Mohr B, et al. Analysis of FLT3-activating mutations in 979 patients with acute myelogenous leukemia: association with FAB subtypes and identification of subgroups with poor prognosis. Blood. 2002;99(12):4326-35. 4. Kottaridis PD, Gale RE, Frew ME, et al. The presence of a FLT3 internal tandem duplication in patients with acute myeloid leukemia (AML) adds important prognostic information to cytogenetic risk group and response to the first cycle of chemotherapy: analysis of 854 patients from the United Kingdom Medical Research Council AML 10 and 12 trials. Blood. 2001;98(6):1752-9. 5. Kottaridis PD, Gale RE, Langabeer SE, Frew ME, Bowen DT, Linch DC. Studies of FLT3 mutations in paired presentation and relapse samples from patients with acute myeloid leukemia: implications for the role of FLT3 mutations in leukemogenesis, minimal residual disease detection, and possible therapy with FLT3 inhibitors. Blood. 2002; 100(7): 2393-8. 6. Schnittger S, Schoch C, Dugas M, et al. Analysis of FLT3 length mutations in 1003 patients with acute myeloid leukemia: correlation to cytogenetics, FAB subtype, and prognosis in the AMLCG study and usefulness as a marker for the detection of minimal residual disease. Blood. 2002;100(1):59-66. 7. Dohner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017; 129(4): 424-47. 8. Koreth J, Schlenk R, Kopecky KJ, et al. Allogeneic stem cell transplantation for acute myeloid leukemia in first complete remission: systematic review and meta-analysis of prospective clinical trials. Jama. 2009; 301(22):2349-61. 9. Gale RE, Hills R, Kottaridis PD, et al. No evidence that FLT3 status should be considered as an indicator for transplantation in acute myeloid leukemia (AML): an analysis of 1135 patients, excluding acute promyelocytic leukemia, from the UK MRC AML10 and 12 trials. Blood. 2005;106(10):3658-65. 10. Oran B, Cortes J, Beitinjaneh A, et al. Allogeneic transplantation in first remission improves outcomes irrespective of FLT3-ITD allelic ratio in FLT3-ITD-positive acute myelogenous leukemia. Biol Blood Marrow Transplant. 2016;22(7): 1218-26.

haematologica | 2018; 103(2)

Editorials

11. Gaballa S, Saliba R, Oran B, et al. Relapse risk and survival in patients with FLT3 mutated acute myeloid leukemia undergoing stem cell transplantation. American journal of hematology 2017; 92(4): 331-7. 12. DeZern AE, Sung A, Kim S, et al. Role of allogeneic transplantation for FLT3/ITD acute myeloid leukemia: outcomes from 133 consecutive newly diagnosed patients from a single institution. Biol Blood Marrow Transplant. 2011; 17(9): 1404-9. 13. Schlenk RF, Kayser S, Bullinger L, et al. Differential impact of allelic ratio and insertion site in FLT3-ITD-positive AML with respect to allogeneic transplantation. Blood. 2014;124(23):3441-9. 14. Fiedler W, Serve H, Dohner H, et al. A phase 1 study of SU11248 in the treatment of patients with refractory or resistant acute myeloid leukemia (AML) or not amenable to conventional therapy for the disease. Blood. 2005;105(3):986-93. 15. 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):454-64. 16. Muffly L, Pasquini MC, Martens M, et al. Increasing use of allogeneic hematopoietic cell transplantation in patients aged 70 years and

older in the United States. Blood. 2017;130(9):1156-64. 17. Ciurea SO, Shah MV, Saliba RM, et al. Haploidentical Transplantation for Older Patients with Acute Myeloid Leukemia and Myelodysplastic Syndrome. Biol Blood Marrow Transplant. 2017 Sep 14. pii: S10838791(17)30713-9. doi: 10.1016/j.bbmt.2017.09.005. [Epub ahead of print] 18. Ciurea SO, Zhang MJ, Bacigalupo AA, et al. Haploidentical transplant with posttransplant cyclophosphamide vs matched unrelated donor transplant for acute myeloid leukemia. Blood. 2015; 126(8): 1033-40. 19. Poire X, Labopin M, Polge E, et al. Allogeneic stem cell transplantation benefits for patients ≥ 60 years with acute myeloid leukemia and FLT3-ITD; a study from the Acute Leukemia Working Party (ALWP) of the European Society of Blood and Marrow Transplantation (EBMT). Haematologica 2017;103(2):256-265. 20. Serve H, Krug U, Wagner R, et al. Sorafenib in combination with intensive chemotherapy in elderly patients with acute myeloid leukemia: results from a randomized, placebo-controlled trial. J Clin Oncol. 2013;31(25):3110-8.

Large granular lymphocyte cells and immune dysregulation diseases – the chicken or the egg? Anton W. Langerak and Jorn L.J.C. Assmann Department of Immunology, Laboratory Medical Immunology, Erasmus MC, Rotterdam, the Netherlands E-mail: a.langerak@erasmusmc.nl doi:10.3324/haematol.2017.186338

n less than 1% of patients suffering from rheumatoid arthritis (RA), a condition known as Felty syndrome (FS) can develop.1,2 FS, first described by the American physician Augustus Roi Felty in 1924, is characterized by the triad of RA, (unexplained) neutropenia, and splenomegaly, and is more common in people aged 50 years and over. In spite of the clear association with RA, the common underlying cause of all three characteristic symptoms remains largely elusive. It is known that FS shows an overlap with a rare type of T-cell leukemia, called T-cell large granular lymphocyte (T-LGL) leukemia, which also typically presents in elderly individuals with a median age of 60 years. In fact, T-LGL leukemia patients can show all the features that FS patients have, albeit at varying frequencies, thus making the differential diagnosis between LGL leukemia and FS problematic at times.3 An additional complication to that of the overlapping clinical features is the fact that both conditions share the presence of LGL cells. Based on the overlapping features, Savola and colleagues explored a potential common pathogenic mechanism between FS and LGL leukemia. In this issue of Haematologica they report on a cohort of 14 FS patients, which were evaluated by next-generation sequencing (NGS) technology for the occurrence of somatic mutations in the STAT3 and STAT5B genes.4 Both of these genes have been notably implicated in a subset of the CD8+ TCRαβ+ T-cell type of LGL leukemia, and the occurrence of STAT3 mutations in LGL leukemia is strongly associated with RA and neutropenia.5-7 Indeed, in >40% of FS cases somatic STAT3 hotspot mutations were found, which is at a rate comparable to LGL leukemia cases. The fact that the STAT3 variant allele frequencies haematologica | 2018; 103(2)

were lower in FS can be explained by the smaller T-cell clone sizes in FS patients. Nevertheless, LGL cell proportions were increased in FS patients, and in two cases the LGL cell numbers additionally fulfilled the criteria for LGL lymphocytosis. Taken together, these observations firmly support the idea that FS and LGL leukemia are part of a disease spectrum with a common pathogenesis, in

Figure 1. Overlap between LGL leukemia / proliferation and immune dysregulation diseases and conditions that show an increase of LGL cells and/or the presence of STAT3 mutated cells. AA: aplastic anemia; CD: celiac disease; FS: Felty syndrome; MDS: myelodysplastic syndrome; PNH: paroxysmal nocturnal hemoglobinuria; PRCA: pure red cell aplasia; LGL: large granular lymphocyte.

193

Editorials

which LGL leukemia ranges from poly/oligoclonal to monoclonal.8 The concept of a disease continuum is further strengthened by the observation that IL-15RA and CXCL10 levels are also increased in both conditions. Hitherto, the occurrence of LGL cells and clones were associated with bone marrow failure syndromes such as aplastic anemia (AA) and (hypoplastic) myelodysplastic syndrome (MDS), both of which are believed to reflect an autoimmune pathogenesis.9,10 However, in conditions such as pure red cell aplasia (PRCA), characterized by the killing of red cell precursors, and paroxysmal nocturnal hemoglobinuria (PNH), characterized by the complement-mediated destruction of red blood cells, LGL proliferations have also been reported.11,12 It is of interest that in many of these conditions STAT3 mutations have been documented. The current description of high LGL counts and STAT3 mutations in FS thus extends the spectrum of immune dysregulation disease conditions in which LGL cells are prominently associated (Figure 1). Notably, in their series, Savola et al. found four FS cases that also showed celiac disease (CD),4 which confirms earlier suggestions of links between LGL cells, STAT3 mutations, and CD.13,14 The presumed common pathogenesis of LGL leukemia and immune dysregulations such as FS, CD, and bone marrow failure and related conditions (AA, hypoplastic MDS, PNH) would pave the way for more sophisticated treatment strategies than the current general immune suppressive modalities which are frequently applied.15 Such targeted strategies would have to exploit common molecular aberrations, one of the most promising approaches being STAT3 inhibition in JAK/STAT hyperactive LGL cells. This is of special interest as constitutive STAT activation is observed in almost all LGL proliferations regardless of the presence of mutations in the STAT3 gene.16 In addition, specific inhibition of JAK family kinases with the AG-490 tyrosine kinase inhibitor has been demonstrated to induce apoptosis of LGL leukemia cells in vitro and antisense oligonucleotide therapy to inhibit STAT3 activation, which has been shown to restore FAS sensitivity in LGL leukemia cells.17 However, further investigation into other common molecular aberrations might reveal distinct molecular mechanisms or pathways in subgroups of these diseases, which could be targeted by yet different molecule-centered treatment strategies. This would require deep sequencing approaches in larger cohorts, and might have to go beyond genomics by studying the transcriptome, the epigenome, and perhaps also small non-coding RNAs. In all the associations between LGL leukemia / proliferations and immune dysregulations, however, one central question remains to be addressed more definitely; “the chicken or the egg” question. In other words, in scientific parlance, are LGL cells truly causing the immune dysregulation, or is their presence more likely to be an epiphenomenon as a consequence of chronic (auto)antigen stimulation in these conditions? Based on existing literature, it is tempting to speculate that LGL cells have an authentic causative role, nevertheless, before definite conclusions can be drawn further research is required. This should undoubtedly include studies that aim to eradicate the LGL clones in patients through targeted treatment, and 194

which could categorically disclose the presumed causative role of LGL cells.

References 1. Balint GP, Balint PV. Felty's syndrome. Best Practice Res Clin Rheumatol 2004;18(5):631-645. 2. Owlia MB, Newman K, Akhtari M. Felty’s syndrome, insights and updates. Open Rheumatol J 2014;8:129-136. 3. Liu X, Loughran TP Jr. The spectrum of LGL and Felty's Syndrome. Current Opin Hematol 2011;18(4):254-259. 4. Savola P, Brück O, Olson T, Kelkka T, Kauppi MJ, Kovanen PE, Kytölä S, Sokka-Isler T, Loughran TP, Leirisalo-Repo M, Mustjoki S. Somatic STAT3 mutations in the Felty syndrome: an implication for a common pathogenesis with large granular lymphocyte leukemia. Haematologica 2018;103(2):304-312. 5. Koskela HL, Eldfors S, Ellonen P, van Adrichem AJ, Kuusanmäki H, Andersson EI, Lagström S, Clemente MJ, Olson T, Jalkanen SE, Majumder MM, Almusa H, Edgren H, Lepistö M, Mattila P, Guinta K, Koistinen P, Kuittinen T, Penttinen K, Parsons A, Knowles J, Saarela J, Wennerberg K, Kallioniemi O, Porkka K, Loughran TP Jr, Heckman CA, Maciejewski JP, Mustjoki S. Somatic STAT3 mutations in large granular lymphocytic leukemia. New Engl J Med 2012;366(20):19051913. 6. Rajala HL, Eldfors S, Kuusanmäki H, van Adrichem AJ, Olson T, Lagström S, Andersson EI, Jerez A, Clemente MJ, Yan Y, Zhang D, Awwad A, Ellonen P, Kallioniemi O, Wennerberg K, Porkka K, Maciejewski JP, Loughran TP Jr, Heckman C, Mustjoki S. Discovery of somatic STAT5b mutations in large granular lymphocytic leukemia. Blood 2013;121(22):4541-4550. 7. Rajala HL, Olson T, Clemente MJ, Lagström S, Ellonen P, Lundan T, Hamm DE, Zaman SA, Lopez Marti JM, Andersson EI, Jerez A, Porkka K, Maciejewski JP, Loughran TP, Mustjoki S. The analysis of clonal diversity and therapy responses using STAT3 mutations as a molecular marker in large granular lymphocytic leukemia. Haematologica 2015;100(1):91-99. 8. Langerak AW, Sandberg Y, Van Dongen JJM. Spectrum of T‐large granular lymphocyte lymphoproliferations: ranging from expanded activated effector T cells to T‐cell leukaemia. Br J Haematol 2003;123(3):561-562. 9. Go RS, Tefferi A, Li CY, Lust JA, Phyliky RL. Lymphoproliferative disease of granular T lymphocytes presenting as aplastic anemia. Blood 2000;96(10):3644-3646. 10. Kochenderfer JN, Kobayashi S, Wieder ED, Su C, Molldrem JJ. Loss of T-lymphocyte clonal dominance in patients with myelodysplastic syndrome responsive to immunosuppression. Blood 2002;100(10): 3639-3645. 11. Kwong YL, Wong KF. Association of pure red cell aplasia with T large granular lymphocyte leukaemia. J Clin Pathol 1998; 51(9):672-675. 12. Risitano AM, Maciejewski JP, Muranski P, Wlodarski M, O'Keefe C, Sloand EM, Young NS. Large granular lymphocyte (LGL)-like clonal expansions in paroxysmal nocturnal hemoglobinuria (PNH) patients. Leukemia 2005;19(2):217-222. 13. Malamut G, Meresse B, Verkarre V, Kaltenbach S, Montcuquet N, Van Huyen JPD, Callens C, Lenglet J, Rahmi G, Samaha E, Ranque B, Macintyre E, Radford-Weiss I, Hermine O, Cerf-Bensussan N, Cellier C. Large granular lymphocytic leukemia: a treatable form of refractory celiac disease. Gastroenterology 2012;143(6):1470-1472. 14. Ettersperger J, Montcuquet N, Malamut G, Guegan N, Lopez-Lastra S, Gayraud S, Reimann C, Vidal E, Cagnard N, Villarese P, AndreSchmutz I, Gomes Domingues R, Godinho-Silva C, Veiga-Fernandes H, Lhermitte L, Asnafi V, Macintyre E, Cellier C, Beldjord K, Di Santo JP, Cerf-Bensussan N, Meresse B. Interleukin-15-dependent T-cell-like innate intraepithelial lymphocytes develop in the intestine and transform into lymphomas in celiac disease. Immunity 2016;45(3):610625. 15. Lamy T, Moignet A, Loughran TP Jr. LGL leukemia: from pathogenesis to treatment. Blood 2017;129(9):1082-1094. 16. Coppe A, Andersson EI, Binatti A, Gasparini VR, Bortoluzzi S, Clemente M, Herling M, Maciejewski J, Mustjoki S, Bortoluzzi S. Genomic landscape characterization of large granular lymphocyte leukemia with a systems genetics approach. Leukemia 2017;31(5):1243-1246. 17. Epling-Burnette PK, Liu JH, Catlett-Falcone R, Turkson J, Oshiro M, Kothapalli R, Li Y, Wang JM, Yang-Yen HF, Karras J, Jove R, Loughran TP Jr. Inhibition of STAT3 signaling leads to apoptosis of leukemic large granular lymphocytes and decreased Mcl-1 expression. J Clin Invest 2001;107(3):351-362.

haematologica | 2018; 103(2)

Editorials

Hemophilia A: different phenotypes may be explained by multiple and variable effects of the causative mutation in the F8 gene Giancarlo Castaman Center for Bleeding Disorders, Department of Oncology, Careggi University Hospital, Florence, Italy E-mail: giancarlo.castaman@unifi.it doi:10.3324/haematol.2017.186353

n this issue of Haematologica, Donadon et al.1 investigated at molecular level the effect of a mutation in the frequent F8 gene (p.R2016W) in determining the circulating Factor VIII (FVIII) level in patients with hemophilia A carrying this missense mutation. The p.R2016W change was expressed and found to impair both FVIII secretion and activity. Furthermore, the nucleotide change (c.6046C>T) associated with the mutation also decreased the correct splicing at mRNA level, contributing to a further lowering of the already reduced FVIII level expected on the basis of impaired secretion and activity. Interestingly, other mutations clustered in the same area of p.R2016W displayed a variable proportion of these mechanisms, providing a good explanation for the genotypephenotype relationships in patients with hemophilia A carrying these mutations. Mutations in coding regions are usually thought to produce an altered biosynthesis or dysfunction of proteins. But in hemophilia this is usually suggested on the basis of phenotype characterization: using immunological assays in plasma to measure the concentration of the protein, or to measure protein activity by coagulation or chromogenic assays. Thus, it is generally assumed that null mutations (e.g. intron 22 inversion in hemophilia A, large gene deletions, stop codons, etc.) reduce or abolish the synthesis and/or the release of the protein while mutations altering the amino acid composition (missense mutations) may affect in particular the activity of the protein in addition to reducing biosynthesis. However, the combination of different mechanisms potentially associated with a given mutation has only very rarely been investigated at molecular level in hemophilia. There has been previous evidence to suggest that, indeed, missense mutations may have pleiotropic effects impairing not only biosynthesis and activity, but also mRNA properties, at least for some coagulation factors.2,3 Donadon et al.1 were able to quantitatively evaluate the pleiotropic effects of a nucleotide change at the RNA and protein (codon) levels. To do so, they chose the p.R2016W F8 mutation which is very frequent in Italy4 and has been putatively suggested to induce also mRNA splicing impairment.5 Very few hemophilia A missense mutations have so far been characterized because of the very low secretion efficiency of recombinant full-length FVIII. Donadon et al.1 used a lentiviral-mediated delivery of expression cassette consisting of the codon-optimized FVIII cDNA lacking the B domain. By this means, they showed that the p.R2016W mutation impairs both FVIII secretion and function. Furthermore, they investigated the F8 mRNA splicing pattern by using mRNA from patientsâ&#x20AC;&#x2122; leukocytes. The mutated c.6046C>T associated haematologica | 2018; 103(2)

with the mutation was found to alter splicing and to decrease the proportion of correct transcript to approximately 75% of wild type by inducing a variable degree of skipping of exon 19. Surprisingly, they found that several other missense changes in the same exon 19 (p.G2013R, p.E2018G, and p.N2038S) are responsible for variable splicing alterations, and that only the combination of altered RNA processing and abnormal protein biology produced a clinically relevant defect (see Figure 4 in Donadon et al.1). This is probably a general feature of several other hematologic disorders frequently associated with mutations that are defined simply as â&#x20AC;&#x153;missenseâ&#x20AC;? without the interplay among several molecular mechanisms producing the disease being known. However, the results obtained by Donadon et al.1 suggest slightly milder hemophilia phenotypes than those observed in patients by clotting assays (see Table 1 in Donadon et al.1). It is tempting to speculate that other components could have a role in influencing the phenotype in addition to the mechanisms highlighted by Donadon et al.1 For example, studies evaluating the half life of different missense mutations are very rare and yet these are needed in order to assess the impact of clearance mechanisms on these mutant proteins.6 Thus, other studies should be carried out with other F8 mutations to confirm that the experimental conditions used in this study are reliably reproducible and to compare the effects of the different pleiotropic mechanisms with other mutants located in different domains of the FVIII protein. Unfortunately, demonstrating the combination of different mechanisms can be very complicated, because expression studies are needed both at the recombinant protein and at RNA levels. Very few missense mutations among the thousands causing hemophilia A have been characterized for FVIII protein expression,7-11 and this study makes a significant contribution to current knowledge in showing that it is possible to dissect the different molecular mechanisms influencing the phenotype, especially in mild and moderate hemophilia A. However, this is a demanding approach, particularly with FVIII mutants with very reduced synthesis and activity. Donadon et al.1 exploited the technology for B-domainless FVIII, used to magnify FVIII expression for substitutive therapy of hemophilia A, to obtain a reliable estimate of the residual level of altered FVIII protein and function, to be combined with the partial negative effects of the RNA splicing defect. Thus, the combination of differentially altered mRNA processing and FVIII biosynthesis and co-factor activity has been shown to make a substantial contribution to variable significant FVIII deficiency caused by clustered variants generated by missense mutations in exon 19 of the F8 gene. 195

Editorials

References 1. Donadon I, McVey JH, Garagiola I, et al. Clustered F8 missense mutations cause hemophilia A by combined alteration of splicing and protein biosynthesis/activity. Haematologica 2018;(2):344-350. 2. Balestra D, Barbon E, Scalet D, et al. Regulation of a strong F9 cryptic 5'ss by intrinsic elements and by combination of tailored U1snRNAs with antisense oligonucleotides. Hum Mol Genet. 2015;24(17):48094916. 3. Tajnik M, Rogalska ME, Bussani E, et al. Molecular Basis and Therapeutic Strategies to Rescue Factor IX Variants That Affect Splicing and Protein Function. PLoS Genet. 2016;12(5):e1006082. 4. Garagiola I, Seregni S, Mortarino M, et al. A recurrent F8 mutation (c.6046C>T) causing hemophilia A in 8% of northern Italian patients: evidence for a founder effect. Mol Genet Genomic Med. 2016;4(2):152-159. 5. Theophilus BD, Enayat MS, Williams MD, Hill FG. Site and type of mutations in the factor VIII gene in patients and carriers of haemophilia A. Haemophilia. 2001;7(4):381-391. 6. Lenting PJ, van Mourik JA, Mertens K. The life cycle of coagulation

196

10. 11.

factor VIII in view of its structure and function. Blood. 1998;92 (11):3983-3996. O'Brien DP, Pattinson JK, Tuddenham EG. Purification and characterization of factor VIII 372-Cys: a hypofunctional cofactor from a patient with moderately severe hemophilia A. Blood. 1990;75(8):1664-1672. Pipe SW, Eickhorst AN, McKinley SH, Saenko EL, Kaufman RJ. Mild hemophilia A caused by increased rate of factor VIII A2 subunit dissociation: evidence for nonproteolytic inactivation of factor VIIIa in vivo. Blood. 1999;93(1):176-183. Pipe SW, Saenko EL, Eickhorst AN, Kemball-Cook G, Kaufman RJ. Hemophilia A mutations associated with 1-stage/2-stage activity discrepancy disrupt protein-protein interactions within the triplicated A domains of thrombin-activated factor VIIIa. Blood. 2001;97(3):685-691. Nogami K, Zhou Q, Wakabayashi H, Fay PJ. Thrombin-catalyzed activation of factor VIII with His substituted for Arg372 at the P1 site. Blood. 2005;105(11):4362-4368. Jourdy Y, Nougier C, Roualdes O, et al. Characterization of five associations of F8 missense mutations containing FVIII B domain mutations. Haemophilia. 2016;22(4):583-589.

haematologica | 2018; 103(2)

REVIEW ARTICLE

From transplant to novel cellular therapies in multiple myeloma: European Myeloma Network guidelines and future perspectives

Francesca Gay,1 Monika Engelhardt,2 Evangelos Terpos,3 Ralph Wäsch,2 Luisa Giaccone,4 Holger W. Auner,5 Jo Caers,6 Martin Gramatzki,7 Niels van de Donk,8 Stefania Oliva,1 Elena Zamagni,9 Laurent Garderet,10 Christian Straka,11 Roman Hajek,12 Heinz Ludwig,13 Herman Einsele,14 Meletios Dimopoulos,3 Mario Boccadoro,1 Nicolaus Kröger,15 Michele Cavo,9 Hartmut Goldschmidt,16 Benedetto Bruno4 and Pieter Sonneveld,17

Myeloma Unit, Division of Hematology, University of Torino, Azienda-Ospedaliero Universitaria Città della Salute e della Scienza di Torino, Italy; 2Universitätsklinikum Freiburg, Medical Department, Hematology, Oncology & Stem Cell Transplantation, Freiburg, Germany; 3Department of Clinical Therapeutics, National and Kapodistrian University of Athens, School of Medicine, Greece; 4Department of Oncology, A.O.U Città della Salute e della Scienza di Torino, and Department of Molecular Biotechnology and Health Sciences, University of Torino, Italy; 5Centre for Haematology, Department of Medicine, Imperial College London, UK; 6Department of Clinical Hematology, Centre Hospitalier Universitaire de Liège, Domaine Universitaire du Sart Tilman, Liège, Belgium; 7Division of Stem Cell Transplantation and Immunotherapy, 2nd Medical Department, University Hospital Schleswig-Holstein, Campus Kiel, Kiel, Germany; 8 Department of Hematology, VU University Medical Center, Amsterdam, the Netherlands; 9Seragnoli Institute of Hematology, Bologna University School of Medicine, Italy; 10INSERM, UMR_S 938, Proliferation and Differentiation of Stem Cells, Paris, APHP, Hôpital Saint Antoine, Département d'Hématologie et de Thérapie Cellulaire; Sorbonne Universités, UPMC Univ Paris 06, France; 11Tumorzentrum München, Germany; 12Department of Hematooncology, University Hospital Ostrava, Czech Republic and Faculty of Medicine University of Ostrava, Czech Republic; 13Wilhelminen Cancer Research Institute, c/o Department of Medicine I, Center of Oncology, Hematology and Palliative Care, Vienna, Austria; 14Department of Internal Medicine II, University Hospital Würzburg, Germany; 15Department of Stem cell Transplantation, University Medical Center Hamburg-Eppendorf, Germany; 16Medizinische Klinik, Abteilung Innere Medizin V, Universitätsklinikum Heidelberg und National Centrum für Tumorerkrankungen (NCT), Heidelberg, Germany and 17Department of Hematology, Erasmus Medical Center, Rotterdam, the Netherlands

Ferrata Storti Foundation

Haematologica 2018 Volume 103(2):197-211

Correspondence: benedetto.bruno@unito.it

ABSTRACT

urvival of myeloma patients has greatly improved with the use of autologous stem cell transplantation and novel agents, such as proteasome inhibitors, immunomodulatory drugs and monoclonal antibodies. Compared to bortezomib- and lenalidomide-based regimens alone, the addition of high-dose melphalan followed by autologous transplantation significantly improves progression-free survival, although an overall survival benefit was not observed in all trials. Moreover, follow up of recent trials is still too short to show any difference in survival. In the light of these findings, novel agent-based induction followed by autologous transplantation is considered the standard upfront treatment for eligible patients (level of evidence: 1A). Post-transplant consolidation and maintenance treatment can further improve patient outcome (1A). The availability of several novel agents has led to the development of multiple combination regimens such as salvage treatment options. In this context, the role of salvage autologous transplantation and allotransplant has not been extensively evaluated. In the case of prolonged remission after upfront autologous transplantation, another autologous transplantation at relapse can be considered (2B). Patients who experience early relapse and/or have high-risk features have a poor prognosis and may be considered as candidates for clinical trials that, in young and fit patients, may also include an allograft in combination with novel agents (2B). Ongoing studies are evaluating the role of novel cellular therapies, such as incluhaematologica | 2018; 103(2)

Received: June 28, 2017. Accepted: December 5, 2017. Pre-published: December 7, 2017. doi:10.3324/haematol.2017.174573 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/197 ©2018 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.

197

Author et al.

sion of antibody-based triplets and quadruplets, and chimeric antigen receptor-T cells. Despite encouraging preliminary results, longer follow up and larger patient numbers are needed before the clinical use of these novel therapies can be widely recommended.

Introduction The treatment landscape and clinical outcome of multiple myeloma (MM) patients have changed in the last decades,1 with an improved median survival of 8-10 years. Multiple combinations of proteasome inhibitors (PIs) and immunomodulatory drugs (IMIDs) have been under evaluation in the transplant and non-transplant settings, and studies are still ongoing. Several pre-transplant inductions and post-transplant novel agent-based consolidation and maintenance regimens have been investigated, although direct comparisons between such strategies have rarely been performed. Autologous stem cell transplant (ASCT) is currently considered the standard of care for fit newly diagnosed MM (NDMM) patients, although remarkable results obtained in the non-transplant setting2,3 with novel agent-based treatment have raised questions as to the role of upfront versus delayed ASCT. The availability of 2nd-generation PIs and IMIDs, monoclonal antibodies, histone deacetylase inhibitors, and, more recently, check-point inhibitors and small molecules, has led to the development of multiple salvage options that include different combinations of these drugs. In this context, the role of salvage ASCT and allotransplant have not been extensively evaluated. These exciting advances require a critical review to delineate the merit of different induction, consolidation and maintenance approaches, as well as to define the role of upfront ASCT, salvage ASCT and allotransplant in the novel agent era. These important considerations prompted the European Myeloma Network (EMN) to provide guidelines to harmonize treatment selection. A brief overview of novel cellular therapies, which can be considered the new frontier for transplant, is also provided.

Methodology Clinical EMN experts on MM developed these recommendations based on published data through August 2017. Expert consensus was used to suggest recommendations in case of inconclusive data. Grades of recommendations were assigned using the GRADE criteria for grade of recommendation (Online Supplementary Table S1). The manuscript underwent revision in 3 rounds until the EMN experts reached mutual consent.

Upfront autologous transplant The current treatment paradigm for NDMM patients eligible for ASCT consists of 4 phases: pre-transplant induction, transplant, post-transplant consolidation and maintenance.

Pre-transplant induction Induction treatment generally consists of 3-6 cycles with the goal of achieving rapid disease control, improve symptoms, and allow for subsequent successful stem cell 198

collection. The current standard is a 3-drug bortezomibbased combination. Doxorubicin-bortezomib-dexamethasone (PAD) proved to be superior to standard chemotherapy in a randomized trial,4 and more recently, bortezomibcyclophosphamide-dexamethasone (VCD) was found to be non-inferior to PAD.5 Improved responses were observed with combinations including both PIs and IMIDs. Indeed, complete response (CR) rates were significantly higher with bortezomib-thalidomide-dexamethasone (VTD) compared with thalidomide-dexamethasone (TD) in 2 randomized trials (35% vs. 14%, P=0.0001; 31% vs. 11%, P<0.001).6,7 VTD versus VCD improved CR rates (13% vs. 9%, respectively).8 Higher CR rates were reported with bortezomib-dexamethasone plus the 2nd-generation IMID lenalidomide (VRD) (23-48%) (Table 1).9,10 No direct, randomized comparisons of PAD versus VTD have been made. Expected efficacy of a given regimen is one of the main factors to be considered in the treatment choice, the second factor being the expected toxicity. Infections are common events in NDMM, often to the underlying disease itself and to the treatment. The main issue with the use of bortezomib (in particular when combined with thalidomide) is the occurrence of peripheral neuropathy (PNP), which can be decreased substantially with subcutaneous and once-weekly administrations. The main concern with combinations including thalidomide or doxorubicin is the thromboembolic risk. Both PNP and thromboembolism rates seem to be lower when bortezomib is associated with cyclophosphamide (Table 2). Given that the benefit of bortezomib could be hampered by its neurological side effects, 2nd-generation PIs with minimal neurotoxicity demonstrated that induction treatment with ixazomib-lenalidomide-dexamethasone (IRD) was very well tolerated (no grade 3-4 PNP, cardiac, liver or renal toxicities) and led to a 12% CR rate.11 Carfilzomib-thalidomide-dexamethasone (KTd)12 or carfilzomib-lenalidomide-dexamethasone (KRd)13 lead to a 1824% CR rate, although cardiovascular toxicities (mainly hypertension) have been reported. The impact of depth of response on outcome14 provides the rationale for choosing the most effective induction regimen, provided the toxicity profile is acceptable. Nevertheless, only one randomized trial (Myeloma XI) investigated a response-adapted approach based on the sequential use of chemotherapeutic agents, with different modes of action in patients with a suboptimal response (minimal response/partial response) to thalidomide-based induction. Some 40% of patients upgraded their response with VCD before ASCT and significant improvement in PFS was observed (median 48 vs. 38 months; P<0.0001).15 However, the trial included suboptimal induction regimens (CTD and cyclophosphamide-lenalidomide-dexamethasone) not widely used outside the UK. The current standard of care is bortezomib plus IMIDs or chemotherapy, supported also by the results of two meta-analyses16,17 that showed the superiority of bortezomib- over nonhaematologica | 2018; 103(2)

Transplant and cellular therapy in myeloma

bortezomib-induction treatments. Thus, the impact of switching treatment, with currently much more effective induction regimens, still needs to be confirmed.

Autologous transplantation Several trials compared different chemotherapy regimens to standard high-dose melphalan (200 mg/m2, MEL200), showing a favorable risk-benefit profile with MEL200 over busulfan/melphalan, idarubicin/melphalan/cyclophosphamide, BCNU/etoposide/melphalan, melphalan 100/140 mg/m2. Conditioning regimens including novel agents have so far only been evaluated in single arm studies.18 Given the efficacy and favorable toxicity profile of MEL200, this regimen remains the standard. Efficacy of novel-agent treatments in the non-transplant setting, together with a manageable safety profile and the advantage of the administration in the outpatient setting, questioned the role of MEL200-ASCT. Four randomized trials compared MEL200-ASCT versus novel agent-based triplets. In two trials, patients received Rd induction and were randomized to tandem MEL200-ASCT or oral lenalidomide-based chemotherapy [melphalan-prednisonelenalidomide (MPR)/cyclophosphamide-lenalidomide-dexamethasone (CRD)]. Median PFS was significantly longer for patients randomized to tandem MEL200-ASCT than for those randomized to MPR (43 vs. 22 months; P<0.001) or CRD (43 vs. 28 months; P<0.001). Tandem ASCT also improved the 4-year OS rate versus MPR (82% vs. 65%; P=0.02) or CRD (86% vs. 73%; P=0.004).19,20

Two large studies compared MEL200-ASCT versus bortezomib-based regimens. MEL200-ASCT significantly prolonged PFS versus bortezomib-lenalidomide-dexamethasone (VRD)10 (median 50 vs. 36 months; P<0.001), and versus bortezomib-melphalan-prednisone (VMP)21 (3-year PFS 65% vs. 57%; P=0.001). Follow up of these two trials is still too short to show any differences in OS. Indeed, data confirmed that the toxicity profile was better and more manageable in the non-transplant arm, but no increase in toxic deaths was reported with MEL200ASCT.10,19,20,21 Before the introduction of novel agents, several studies showed a prolonged event-free survival (EFS) with double versus single ASCT.22 A subgroup analysis of one of those trials reported an improved OS only in patients achieving less than very good partial response (VGPR) after the first ASCT.23 A more recent integrated analysis of patient-level data from 4 European trials demonstrated that, in patients receiving bortezomib-based induction, the greatest benefit with double versus single ASCT in terms of extended PFS [Hazard Ratio (HR)=0.41] and OS (HR=0.22) was seen in patients with t(4;14) and/or del(17p) who failed CR to induction therapy.24 Preliminary results of the EMN02 trial confirmed that patients receiving double ASCT have a superior PFS in comparison with patients randomized to a single ASCT (3-year PFS 74% vs. 62%; P=0.05). The benefit was particularly evident in patients with high-risk cytogenetics (3-year PFS 65% vs. 41%; HR 0.49, P=0.046).25 On the contrary, the STAMINA trial showed no

Table 1. Efficacy of sequential approaches with autologous transplantation: improvement in response rates, progression-free survival and overall survival with sequential induction, transplant, and consolidation-maintenance regimens.

Regimen PAD MEL 200 V maintenance VTD MEL 200 VTD consolidation VTD MEL 200 T/INF/VT maintenance VRD MEL 200 VRD consolidation R maintenance VCD MEL 200 VRD/no consolidation R maintenance KRD MEL200 KRD consolidation R maintenance IRD MEL200 IRD/IR consolidation I maintenance

N of patients

Median FU (months)

413

236

130

350

1499

CR (%) 7 21 36 23 49 61 35 46 59Â° 32 24 41 61 12 17 29/44 -

PFS

Study ref

50% at 35 months

61% at 60 months

60% at 36 months*

90% at 36 months*

50% at 56 months

74% at 48 months

50% at 50 months

81% at 48 months

65% at 36 months

86% at 36 months

91% at 24 months

83% at 20 months

95% at 20 months

CR: complete response; MEL 200: melphalan 200 mg/m2; PFS: progression-free survival; OS: overall survival; Study ref: references in literature; FU: follow up; R: lenalidomide; RP: lenalidomide-prednisone; N: number; T: thalidomide; V: bortezomib; VTD: bortezomib-thalidomide-dexamethasone; INF: interferon; VT: bortezomib-thalidomide; IR: ixazomiblenalidomide; I: ixazomib; PAD: bortezomib-adriamycin-dexamethasone; VRD: bortezomib-lenalidomide-dexamethasone; VCD: bortezomib-cyclophosphamide-dexamethasone; KRD: carfilzomib-lenalidomide-dexamethasone; IRD: ixazomib-lenalidomide-dexamethasone; -: data not available. Â°Response to the overall treatment. *PFS/OS from the start of consolidation.

haematologica | 2018; 103(2)

199

F. Gay et al.

from this approach.30,31 In addition, grade 3-4 PNP (7-19%) and treatment discontinuation due to PNP limit the longterm use of thalidomide. Bortezomib maintenance seems to be a better option: a landmark analysis of the HOVON-65/GMMG-HD4 trial showed that bortezomib maintenance significantly reduced the risk of progression (P=0.04) and death (P=0.05) as compared with thalidomide, with a similar rate of grade 3-4 PNP (5% vs. 8%).4 Results of this trial also suggest that pre-transplant bortezomib induction followed by bortezomib maintenance significantly reduces the high-risk impact of del(17p) and renal impairment on survival.32 More recently, longer PFS was reported also with the bortezomib-thalidomide combination versus thalidomide alone.33 Lenalidomide is another valid strategy for long-term treatment, with limited neurotoxicity: 4 trials subsequently evaluated lenalidomide maintenance after ASCT,19,34–36 showing a consistent PFS benefit for lenalidomide versus no maintenance (HR range 0.46-0.50). A meta-analysis of the first three randomized trials reported a significant increase also in OS (7-year OS 62% vs. 50%; HR 0.75, P=0.001) across all subgroups analyzed with the exception of patients with high-risk cytogenetics. In the MRC trial, a significant PFS benefit was maintained also in patients with high-risk cytogenetics, but no data on OS are currently available. Main grade 3-4 toxicities were neutropenia (23-51%), and infections (6-13%).19,34,35 Although second primary malignancies (SPMs) were higher with lenalidomide maintenance versus control (hematologic SPM 6.1% vs. 2.8%; solid tumor SPM 7.3% vs. 4.2%),37 the OS benefit outweighed the SPM risk. Recommendations in NDMM patients eligible for high-dose therapy and ASCT, sequential treatment including novel agent-based induction, upfront transplant, post-transplant bortezomib plus IMIDs consolidation and maintenance is recommended (1A) (Figure 1). Treatment choice should be based on evidence supporting a specific treatment, and on a thorough evaluation of the patient’s characteristics, toxicity of the expected regimens, and availability of drugs in the specific countries (Table 3).

improvement in PFS in patients receiving double ASCT followed by maintenance versus single ASCT followed by VRD consolidation and lenalidomide maintenance. However, different induction regimen, more effective and prolonged therapy with better disease control before ASCT, as well as non-adherence to the double ASCT policy in 30% of patients can prove to be a limitation of this comparative trial.26

Consolidation regimens Consolidation is a commonly adopted approach after transplant to improve depth of response. In “naïve” patients, bortezomib consolidation prolonged PFS versus no consolidation (median 27 vs. 20 months, respectively; P=0.05), but no difference in OS was seen.27 In another trial, VTD consolidation increased the CR rate from 15% to 49% and the molecular remission rate from 3% to 18%.14 More recently, post-ASCT consolidation with the same induction regimens was assessed. VTD increased the CR/nCR rate from 63% to 73%.28 Similarly, CR plus stringent CR rate increased from 47% to 50% after VRD.9 Preliminary results of the EMN-02 trial suggest that posttransplant VRD consolidation also prolongs PFS versus no consolidation (3-year PFS 65% vs. 60%, respectively; P=0.045).29 The STAMINA trial did not find any improvement in PFS with single ASCT followed by VRD consolidation and lenalidomide maintenance versus single ASCT followed by lenalidomide maintenance. However, the rate of non-compliance to VRD was sizeable at 12%.26 Similarly to induction phase, combining 2nd-generation PIs and IMIDs is advantageous also in the consolidation phase, enhancing CR rates from 20% to 32% with IRD, from 31% to 64% with KTD, and from 45% to approximately 70% with KRD.11-13

Maintenance regimens The optimal maintenance regimen should aim at prolonging the remission duration without affecting patients' quality of life. Although meta-analyses showed a reduced risk of progression (HR=0.65) and death (HR=0.84) with thalidomide maintenance, in the IFM and MRC IX studies, patients with unfavorable cytogenetics did not benefit

Table 2. Safety (grade >3 adverse events) of selected pre-transplant induction and post-transplant consolidation/maintenance regimens.

Regimen Induction PAD VTD VCD KRD Consolidation VTD KRD Maintenance V T R TV

Neutropenia (%)

Thrombocytopenia (%)

Anemia (%)

Thromboembolism (%)

PNP (%)

Infection (%)

Study ref

3 10 35° 16

10 8 4 2

8 6 2

4 12 3# -

24 14 8# -

26 21 22# 15

4 6 5 13

5* 15

1 -

1 2

28 13

0 1-16 23-51 13

4 2 4-14 10

1 1 2-5 -

1 1 2-3 -

5 8-14 1 15

24 18 6-8 -

4 4,33 19,20,34,35 33

R: lenalidomide; T: thalidomide;V: bortezomib; VTD: bortezomib-thalidomide-dexamethasone; PAD: bortezomib-adryamicin-dexamethasone;VRD: bortezomib-lenalidomide-dexamethasone; VCD: bortezomib-cyclophosfamide-dexamethasone; KRD: carfilzomib-lenalidomide-dexamethasone; IRD: ixazomib-lenalidomide-dexamethasone; Study ref: references in literature; TV: thalidomide-bortezomib; PNP: peripheral neuropathy; -: data not available. °Including leukopenia. #: ≥grade 2. *All grade events.

200

haematologica | 2018; 103(2)

Transplant and cellular therapy in myeloma

Special considerations Currently, selection criteria for high-dose therapy include age and comorbidities. However, a definite age cut off, rather than assessment of patientâ&#x20AC;&#x2122;s biological age, comorbidities, fitness and frailty/comorbidity scores is suboptimal. Besides age, the performance status, and cardiac, pulmonary, hepatic and renal functions should be considered to better evaluate the risk-benefit ratio of transplant for each patient, and specific risk-assessment models, such as the Myeloma Comorbidity Index (MCI) and/or Hematopoietic Cell Transplantation Comorbidity Index (HCT-CI) should be used to better modulate the dose of chemotherapy.38-40 Specific considerations refer to patients with renal failure (RF) and elderly patients.

Renal failure (RF) Approximately 20% of patients have creatinine more than 2 mg/dL at diagnosis. Bortezomib-based regimens remain the cornerstone of management of renal failure (RF). Indeed, higher response rates were reported with PAD versus VAD induction in patients with RF (81% vs. 63%; P=0.31).41 In dialysis patients, bortezomib-based induction versus conventional chemotherapy significantly increased pre-transplant (83% vs. 36%; P=0.02) and post-

MM: multiple myeloma; ASCT: autologous stem cell transplantation; VTD: bortezomib-thalidomide-dexamethasone; VRD: bortezomib-lenalidomide-dexamethasone; VCD: bortezomibcyclophosphamide-dexamethasone; PAD: bortezomib-adriamycin-dexamethasone; MEL 200: melphalan 200 mg/m2; VGPR: very good partial response; PD: progressive disease; pts: patients.

Figure 1. Recommended sequential treatment.

Table 3. Recommendations for up-front treatment in transplant-eligible patients. Induction

Transplant

Regimens

Recommendation

Rationale for recommendation

VTD (1A) VRD (1B) PAD (1A) VCD (1B) -

Treatment choice: -Non-neurotoxic agents (doxorubicin, lenalidomide, cyclophosphamide) preferred in pts with PNP. Non-thrombotic agents (cyclophosphamide) to be considered in pts with thrombosis. - Lenalidomide use is supported by better toxicity profile than thalidomide, and the advantage of an oral use as compared with doxorubicin.

Treatment choice: - Randomized comparisons showing the superiority of one of these regimens over the others are lacking. - Treatment choice should consider patients' characteristics and expected toxicity of the proposed regimens. - VTD showed superiority vs. TD, chemotherapy without novel agents and VCD.6,7,8 - VRD showed promising phase II and III efficacy results, with a good safety profile, but randomized comparisons VRD vs. other induction regimens are lacking.9,10

Number of cycles: - Treatment should be continued for at least 3-4 cycles with all regimens. - Patients achieving >PR with VTD can continue for another 2 cycles.

Number of cycles: - Most of the trials evaluated 3-4 cycles of induction. - Phase III data on efficacy and toxicities of > 4 cycles are lacking, except for VTD.4,5,6,7,8 - Randomized comparison of prolonged induction until best response and ASCT vs. fixed duration of induction and ASCT are lacking.

Treatment choice: MEL200

Treatment choice: - Randomized trials showed a favorable efficacy and safety profile of MEL200 vs. other regimens (Bu/Mel, Ida/Mel/Cy, BCNU/Etoposide/Mel, Mel100, Mel140).18 - Novel agents in the conditioning regimens so far evaluated only in single arm studies. 18

MEL200 (1A)

Number of cycles: Number of cycles: 2 MEL200-ASCT are recommended in particular - Data from meta-analysis and 2 phase III trials suggest that the greatest benefit in patients with high-risk disease and <CR. with double vs. single ASCT is for patients with high-risk disease.4,24,25 Phase III data of STAMINA trial showed equal PFS between patients that, after a first ASCT, 1 MEL200-ASCT can be considered for standard were randomized to consolidation with a second ASCT plus lenalidomide risk patients achieving >VGPR. maintenance, or VRD consolidation followed by maintenance or maintenance only, but these results may be affected by non-adherence to the second transplant policy in 30% of patient maintenance.4,26 - Integrated patient level meta-analysis in the context of bortezomib induction showed the greatest benefit for double vs. single ASCT in patients who failed CR to induction therapy. Before novel agent treatment, the benefit of double ASCT was reported in patients achieving <VGPR after the first ASCT.23 continued on the next page

haematologica | 2018; 103(2)

201

F. Gay et al.

transplant (100% vs. 58%; P=0.01) overall response rate. Prolonged EFS and a trend towards less time on hemodialysis (6 vs. 17 months) was also reported.42 Combination of bortezomib with high cut-off hemodialysis led to prompt and remarkable responses.43 RF does not appear to affect the quality of stem cell collection.44 Persistent RF or dialysis are not contraindications to high-dose therapy and ASCT,45 since patients may improve renal function after ASCT. Nevertheless, the rate of treatment-related mortality (TRM) ranges from 0 to 29% in different reports and with different melphalan doses.42,44 Thus, due to the potentially higher toxicity of 200 mg/m2, dose reductions are mandatory, particularly in dialysis patients. Other suggested reductions in case of impaired organ function are reported in Online Supplementary Table S2. Of note, a recent large retrospective analysis showed no significant differences in the 5-year PFS and OS between transplant patients with normal, moderate [glomerular filtration rate (GFR) 30-60 mL/min/1.73 m2)] and severe RF (GFR<30). For patients with moderate RF, 5-year PFS was 18% with melphalan 140 mg/m2, and 46% with melphalan 200 mg/m2 (P=0.009); 5-year OS was 67% and 68%, respectively (P=0.52). In patients with severe RF (GFR<30), no

differences in 5-year PFS and OS were reported between groups. Relapse remained the primary cause of death in all patient subgroups.46 In this report, 85% patients achieved dialysis independence post ASCT even though, in previous case series, rate of dialysis independence varied from 6% to 25%.44 Of interest, around 10% of younger patients may achieve long-lasting responses, which makes them potential candidates for renal transplantation. However, many issues, including donor availability, the immunosuppression risks and the possible disease relapse on the xenograft, need to be considered. Thus, patients with lowrisk disease and with negative minimal residual disease (MRD) might be considered eligible for transplantation in the future but currently, due to limited data, no recommendations can be made.44

Transplant in the elderly Aging is associated with reduced organ function and drug metabolism, with potentially increased toxicity and TRM. The potential increase in toxicity has led to the evaluation of reduced doses of melphalan conditioning (100140 mg/m2). Many studies, mostly retrospective, observa-

continued from the previous page

Consolidation

Regimens

Recommendation

Rationale for recommendation

VTD (2A) VRD (2A)

Treatment choice: - Lenalidomide use is supported by a better toxicity profile than thalidomide. - Lenalidomide should be preferred in pts with PNP.

Treatment choice: - Randomized comparisons showing the superiority of one of the regimens over the other are lacking. - Treatment choice should consider patient characteristics and expected toxicity of the proposed regimen.

Duration of therapy: - 2 VTD cycles. - 2 VRD cycles.

Duration of therapy: - A randomized trial showed the superiority of 2 VTD vs. 2 TD consolidation in terms of response rate and PFS.7 - Preliminary data of a randomized trial showed the superiority of 2 VRD cycles vs. no consolidation in terms of PFS.29

Treatment choice: - Lenalidomide use is supported by a better toxicity profile than thalidomide, which favors the long-term administration. - Bortezomib use is supported by a better toxicity profile than thalidomide, and a potentially higher efficacy. - IMIDs alone could be suboptimal in high-risk patients and patients with renal failure, who may benefit from bortezomib.

Treatment choice: - Treatment choice should consider patients' characteristics and expected toxicity of the proposed regimen. - Thalidomide and lenalidomide maintenance have been evaluated in several trials.19,30,31,34,35 - One study showed the superiority of bortezomib over thalidomide maintenance, but results are limited by the fact that patients receiving bortezomib maintenance received bortezomib induction, while patients randomized to thalidomide received VAD.4 - Randomized comparisons showing the superiority of lenalidomide vs. thalidomide/bortezomib are lacking. - Subgroup analyses of randomized trials showed an uncertain benefit of IMIDs in patients with high-risk cytogenetics and renal failure, and a possible benefit with bortezomib.30,31,4,35,37

Duration of therapy: - Lenalidomide: at least 2 years or until tolerated. - Thalidomide: until tolerated - Bortezomib: 2 years.

Duration of therapy: - There are no randomized trials comparing 2 years of lenalidomide vs. lenalidomide until PD, but the median duration of maintenance is approx. 2 years in most of the trials. - Long-term thalidomide use is limited by the poor tolerance (PNP). - Bortezomib maintenance has been administered in clinical trials for up to 2 years.

Maintenance Lenalidomide (1A) Thalidomide (1A) Bortezomib (1B)

MEL200: melphalan 200 mg/m2; ASCT: autologous stem cell transplant; PFS: progression-free survival; OS: overall survival;VTD: bortezomib-thalidomide-dexamethasone; PAD: bortezomib-adriamycindexamethasone;VRD: bortezomib-lenalidomide-dexamethasone;VCD: bortezomib-cyclophosphamide-dexamethasone;VAD: vincristine-doxorubicine-dexamethasone; PNP: peripheral neuropathy; IMID: immunomodualtory drugs.TD: thalidomide-dexamethasone; PD: progressive disease; PR: partial response; VGPR: very good partial response; CR: complete response; pts: patients.

202

haematologica | 2018; 103(2)

Transplant and cellular therapy in myeloma

tional or registry-based, provided encouraging results with ASCT in patients over 65 years of age, with TRM less than 3-4%.47 No differences in TRM (1%) were reported with tandem melphalan 140 mg/m2 in patients aged 60-65 years versus 65-70 years in the large DSMM II trial.48 Interestingly, a recent study found that ASCT-TRM was 0% with either melphalan 140 mg/m2 or 200 mg/m2, which may partly be due to improvements in supportive therapy and better patient selection.49 A recent European Society for Blood and Marrow Transplantation (EBMT) study confirms increased utilization and safety of ASCT with improved post-transplant survival, particularly in elderly MM patients.50 Former analysis of non-ASCT treatment versus ASCT in the elderly (65-75 years) compared thalidomide-based chemotherapy (MPT) versus reduced-intensity (melphalan 100 mg/m2) ASCT in patients aged 65-75 years in the IFM9906 trial. MPT significantly reduced the risk of progression (HR 0.54, P=0.0002)51 and death (HR 0.69, P=0.027), but the lack of novel agents in the pre-ASCT induction and the low melphalan dosing could be a limitation to the study. The rate of toxic deaths was also higher (5%) during induction in the ASCT arm. Other prospective trials subsequently evaluated a sequential approach including novel agent based-induction, consolidation and maintenance. One study showed that PAD induction, followed by MEL100-ASCT, lenalidomide-prednisone consolidation and lenalidomide maintenance was highly efficacious (VGPR rate 82%, 5-year OS 63%) and feasible, in particular for patients under 70 years of age who reported a significantly lower rate of TRM in comparison with elderly patients (5% vs. 19%).52 A recent report suggests that bortezomib consolidation after ASCT may determine clinical outcomes in older patients, who may have been less heavily pre-treated, as in younger patients treated with standard doses of melphalan.53 The phase III DSMM XIII trial compared continuous Rd versus Rd induction followed by tandem melphalan 140 mg/m2-ASCT and lenalidomide maintenance. Results of the planned interim

analysis showed a 3-year-survival rate of 75% for all patients. A longer follow up is needed to evaluate the potential advantages and disadvantages of combining lenalidomide with high-dose melphalan-ASCT as compared with continuous RD.54 Recommendations: biological age rather than chronological age, PS, and organ function should be considered to better evaluate the risk-benefit ratio of transplant for each patient (1B) (Figure 2). Objective risk-assessment scores, such as the Revised-Myeloma Comorbidity Index (RMCI) and/or the Hematopoietic Cell Transplantation Comorbidity Index (HCT-CI) can be used to define the appropriate dose of chemotherapy38-40 (1B) (Table 4).

Transplant at relapse Upfront versus rescue transplant In the past, several randomized trials confirmed the PFS benefit with early ASCT as compared with chemotherapy. In 3 randomized studies, OS was similar whether ASCT was performed early or at first relapse. Despite similar OS, early ASCT improved the average time without symptoms and reduced treatment-related toxicities in 1 trial.55 However, at the time of these trials, most novel agents were not available. Based on the impressive results of novel agent-based treatments in the non-transplant setting, the option of delaying ASCT until first relapse was reconsidered.2,3,56 In all the recent randomized phase III trials comparing ASCT versus novel agent-based therapies, patients who did not receive ASCT upfront were recommended to receive it at first relapse. A pooled analysis including the GIMEMA and the EMN441 trials showed that only 53% of patients eligible for Mel200-ASCT at diagnosis actually received ASCT at first relapse. Upfront MEL200-ASCT significantly improved not only PFS1, but also PFS2 (4-year PFS2 71% vs. 54%; HR 0.53, P<0.001) and OS (4-year OS 84% vs. 70%; HR 0.51, P<0.001) as compared with oral chemotherapy plus lenalidomide.57

Figure 2. Factors to consider for transplantation. MM: multiple myeloma; ASCT: autologous stem cell transplant; R-MCI: Revised-Myeloma Comorbidity Index; HCT-CI: Hematopoietic Cell Transplantation - Specific Comorbidity Index.

haematologica | 2018; 103(2)

203

F. Gay et al.

Of note, in the IFM2009 trial, in which up to 79% of patients treated with lenalidomide plus bortezomib upfront were rescued with ASCT at relapse, no differences in OS were noticed.10

Transplant in patients relapsing after prior autograft Multiple retrospective analyses showed that chemo-sensitivity and remission duration after first ASCT are the most important prognostic factors for long-term disease control after salvage ASCT.58,59 Most reports also highlighted the impact of the number of prior therapies on outcome, suggesting that salvage ASCT should be part of the initial salvage strategies, rather than be offered to patients who have failed multiple therapy lines. A retrospective analysis on 1061 patients showed a significantly longer median survival for patients who received salvage ASCT (4 years) versus those who received salvage IMIDs/PIs and no ASCT (3.3 years), and those who received conventional chemotherapy (2.5 years).60 A limitation of this analysis is a possible selection bias as patients who were treated with ASCT may have been in better clinical condition compared with those who were not. Nevertheless, the phase III multicenter randomized Myeloma X trial showed a significant advantage in time to progression (19 vs. 11 months; P<0.001) and OS (67 vs. 52 months; P=0.022) in patients relapsing after a previous ASCT, and then randomized to receive either a second ASCT or oral cyclophosphamide.61

The limitation of these trials, however, is that, even though all patients were re-induced with PAD prior to randomization, the control arm with cyclophosphamide alone can now be considered suboptimal. A recent retrospective EBMT analysis showed that even a third ASCT at relapse may be feasible, with more than 80% of patients achieving at least a PR, although with increased non-relapse mortality. Particularly in severely cytopenic patients in whom hematologic toxicity of conventional treatment may be prohibitive, ASCT may be a rescue option. The option of a third ASCT mostly followed a previous upfront approach with tandem ASCT; some patients received a first ASCT followed by a second ASCT at second relapse and a third ASCT at subsequent relapse. The first scenario resulted in better results with a median OS of more than four years if the relapse occurred after more than three years after the upfront tandem ASCT.62 Recommendations: upfront ASCT remains the standard option for patients eligible for HDT (1A) (Figure 1). A second transplant at relapse should be considered after a minimal duration of remission of 18 months after a first ASCT (1B); this cut off could be extended to 24 months in the context of novel induction/maintenance.63 A second ASCT should be offered as a first salvage therapy rather than after failing multiple lines (2B). Novel-agent based induction and consolidation-maintenance should be adopted also in the elderly (1A).

Table 4. Recommendations for transplant in elderly patients and patients with co-morbidities. All recommendations are level 2C.

Factor to consider Age

Cut off for fulldose melphalan <65 years

Performance Status

Co-morbidities

Karnofsky>90%

HCT-CI = 0 R-MCI 0-3

Recommendation

Rationale for recommendation

- Age should be considered not as single factor but - Retrospective data showed in recent years no increase in TRM in elderly together with Performance Status and co-morbidities patients, probably due to better supportive measures and patient (HCT-CI/MCI). selection. These results have been achieved not only with reduced dose - Biological rather than chronological age should be of melphalan, but also with full dose.50â&#x20AC;&#x201C;53,55â&#x20AC;&#x201C;57 used in deciding eligibility to ASCT. - In patients between 65-70 years, with Karnofsky PS>90% and HCT-CI = 0 or R-MCI 0-3, it is reasonable to consider full dose melphalan (200 mg/m2). - Based on biological age, melphalan dose reductions (melphalan 100-140 mg/m2) can be appropriate. - In patients with Karnofsky PS <90% melphalan dose - Retrospective analysis of registry data showed an inferior OS in patients reductions (melphalan 100-140 mg/m2) should with Karnofsky PS<90%.41 be considered. - Poor PS can be related to MM (i.e. bone disease, and rib and vertebral - Full dose melphalan (200 mg/mq) could be fractures that affect respiratory function, suboptimal response of MM considered in patients with poor PS related to the MM, to previous therapy can lead to anemia and fatigue). more than to other co-morbidities. Achieving optimal disease control can improve patient PS. - In patients with HCT-CI >1 or R-MCI 4-6 melphalan - Retrospective analysis of registry data showed an inferior OS in dose reductions (melphalan 100-140 mg/m2) need patients with HCT-CI 1-2 or >2, even if TRM at 1 year was equivalent to be considered. in HCT-CI 0 or >2.41. Specifically, in case of impaired: - Retrospective data showed also inferior OS in patients with a) cardiac function (LVEF 40-50%; NYHA II) R-MCI >4 vs. 0-3.43 b) liver function (bilirubin >1.5 ULN, AST/ALT >2.5 ULN) c) pulmonary function (DLCO/FEV1 40-80%) - Reduced organ function can be related to MM, in particular in case d) renal function (GFR <60) of renal failure and reduced pulmonary function due to bone but, in particular, for c) and d) a careful evaluation fractures (thoracic cage). Achieving optimal disease control can improve of the cause of impaired organ function should be done, organ function, in particular in patients with renal failure, as shown and in case of impaired renal function related to MM, in retrospective studies.49 the risk benefit of full-dose melphalan should be considered.

Study ref: references in literature. HCT-CI/MCI: Hematopoietic Cell Transplantation - Specific Comorbidity Index/Myeloma Comorbidity Index; ASCT: autologous stem cell transplant; PS: Performance Score; R-MCI: Revised-Myeloma Comorbidity Index; MM: multiple myeloma; LVEF: left ventricular ejection fraction; ULN: upper limit normal; AST: aspartate transaminase; ALT: alanine transaminase; DLCO/FEVI: diffusion lung capacity for carbon monoxide/forced expiratory volume; GFR: glomerular filtration rate; TRM: treatment-related mortality; OS: overall survival.

204

haematologica | 2018; 103(2)

Transplant and cellular therapy in myeloma

Allotransplant When and in which patients A review of approximately 3000 ASCTs and allo-SCTs, performed in the USA between 2007 and 2009 showed that, overall, 47.1% of ASCTs and only 3.6% of allo-SCTs were performed in MM patients.64 However, the number of allo-SCTs for MM in Europe steadily increased from 1990 to 2012.65 Before new drugs became readily available almost 20 years ago, in a series of "biologically" randomized prospective studies, the concept of splitting myeloablation and graft-versus-myeloma (GvM) by a tandem approach with a standard ASCT followed by a non-myeloablative allo-SCT from a matched sibling or an unrelated donor was explored in NDMM (Table 5).66-76 Results were discordant, and this was likely due to differences in study design, target population and post-transplant immunosuppression (Table 3). Moreover, only at long-term follow up were differences in clinical outcomes between arms observed.73,74 Of note, at that time, most studies could not include new drugs either at induction or as post allo-SCT maintenance/consolidation. Partly due to the conflicting results and to the introduction of new drugs, in recent years used allo-SCT has tended to be used as a salvage strategy at relapse, often not in the context of clinical trials. Most reports were single institution or registry analyses. Only a few comparative studies have been conducted, and these are limited by their retrospective nature and/or small patient cohorts (Table 6). In a recent EBMT report65 on 7333 MM patients who underwent allo-SCT between 1990 and 2012, 3405 had received allo-SCT as a second line or beyond regimen; this report showed that 25% of the patient cohort who received allo-SCT more than eight months from the first ASCT survived at ten years, suggesting that cure may have been reached through a GvM mechanism in some patients. Another retrospective EBMT analysis identified

patient and donor cytomegalovirus (CMV) seronegativity as the key prognostic factor for better outcome after alloSCT in relapsed patients.77 One prospective study78 concluded that, with well-matched donors, the non-relapse mortality was 10%, and approximately 20% of patients achieved long-term disease-free survival. The high response rates seen after donor lymphocyte infusions (DLI) administration provide additional evidence for the GvM effect. Taken together, these studies have showed the feasibility of allo-SCT in relapsed MM; however, given the heterogeneous patient cohorts and differences in conditioning regimens and supportive care, its real role and curative potential has not been clearly established. Both reducedintensity and myeloablative conditionings have been successfully used and, so far, the choice should be based on center policy and patientsâ&#x20AC;&#x2122; comorbidities. Considering the lack of effective therapy for high-risk patients carrying del(17p), gain(1q), t(4;14) and t(14;16) abnormalities, new treatment modalities should be sought in this patient subset. The negative prognostic impact of high-risk cytogenetics appeared to be partly neutralized by GvM in two recent studies. KrĂśger et al. did not observe significant differences in PFS between patients harboring del17p13 and/or t(4;14) and those without these genetic abnormalities after a median follow up of six years (24% vs. 30%; P=0.70). Depth of remission had a remarkable impact on 5-year PFS: 17% for PR, 41% for CR, 57% for molecular CR, and 85% for sustained molecular CR.79 A French trial also showed no differences in clinical outcomes between t(4;14) and non-t(4;14) patients. Moreover, the 3-year progression rate did not exceed 45% in patients with del(17p).80 Taken together, these findings raise the question as to whether high-risk patients who usually experience poor outcomes and easily develop resistance to novel agents would benefit from allo-SCT earlier in the course of the disease.

Table 5. Allogeneic stem cell transplant upfront, donor versus no-donor prospective trials.

Study design High-risk patients BU-FLU-ATG allo-SCT Auto-SCT 2nd auto-SCT FLU-MEL allo-SCT Auto-SCT <CR 2nd auto-SCT 2Gy TBI allo-SCT Auto-SCT 2nd auto-SCT 2Gy TBI allo-SCT Auto-SCT Auto-SCT +/-maintenance 2Gy TBI allo-SCT Auto-SCT Maintenance T/IFN FLU-TBI allo-SCT Auto-SCT +/- 2nd auto-SCT

Patients

Median FU

65 4.8 years 219 25 5.2 years 85 80 7 years 82 185 3.3 years 397 122 6.4 years 138 91 8 years 249

PFS

Median 19% vs. 22% (vs.=0.58) Median not reached vs. 31 months (P=0.08) Median 2.8 years vs. 2.4 years (P=0.005) At 3 years 43% vs. 46% (P=0.67) At 6 years 28% vs. 22% (P=0.19) At 8 years 22% vs. 12% (P=0.027)

Median 34% vs. 48% (P=0.07) Median not reached vs. 58 months (P=0.9) Median not reached vs. 4.25 years (P=0.001) At 3 years 77% vs. 80% (P=0.191) At 6 years 55% vs. 55% (P=0.68) At 8 years 49% vs. 36% (P=0.030)

Study ref

67,68

66,73

72,74

FU: follow up; PFS: progression-free survival; OS: overall survival; Study ref: references in literature; SCT: stem cell transplant; BU: busulfan; FLU: fludarabine; ATG: anti-thymocyte globuline; MEL: melphalan; TBI: total body irradiation; Gy: Gray; T: thalidomide; IFN: interferon.

haematologica | 2018; 103(2)

205

F. Gay et al.

Evidence of graft-versus-myeloma effect Response to DLIs is often seen as proof of GvM effect. However, the prolonged post-relapse survival reported after tandem auto-allo-SCT upfront suggests an important synergy between novel agents and GvM.73,74 In several reports, DLIs have been used as salvage treatment. Beitinjaneh et al.81 reported on 23 of 162 patients with MM receiving DLI post allo-SCT for residual or relapsed disease: 22% achieved VGPR or better with a median duration of 21.8 months. Similarly, an analysis of EBMT registry data reported a response rate of 63% in 70 patients when DLI was given pre-emptively and 52% when given at relapse.82 Ladetto et al. reported a gradual reduction of residual disease with longer follow up. Minimal residual disease negativity, detected by molecular methods, remained low up to three months post alloSCT, then increased up to 44% at six and 47% at 12 months.83 Importantly, these patients did not receive any maintenance/consolidation treatment. These findings also compared favorably with the molecular analysis conducted by the same group in patients undergoing autografting and VTD consolidation.14 Finally, although not univocal, many trials reported a favorable association between development of chronic graft-versus-host disease (GvHD) and prolonged PFS and OS,84,85 again supporting a GvM effect.

Novel agents and graft-versus-myeloma effect Although the introduction of â&#x20AC;&#x2DC;new drugsâ&#x20AC;&#x2122; has made allografting a less attractive treatment option because of its toxicity, the mechanisms of action of new drugs and immune-mediated GvM effects are by no means mutually

exclusive.73,74 Given that one of the most important predictors of survival is the response at the time of transplant, and the major limitation remains disease recurrence as for all other treatments, the new anti-MM drugs may strongly improve outcomes of allo-SCT. Moreover, the concept of maintenance treatment was also recently introduced in the setting of allografting. Bortezomib has been used before allo-SCT and early after allo-SCT to eliminate residual disease and to decrease GvHD incidence and severity based on its presumed immunomodulatory potency in at least two prospective studies86,87 on 16 and 12 high-risk MM patients, respectively. Both trials proved feasible and safe and, based on these results, the expert panel agree that larger confirmatory studies should be designed. Lenalidomide is also of interest in the allo-SCT setting, although this should be considered with caution because of the risk of GvHD flares if given too soon after transplant. Three trials86,88,89 demonstrated that post allo-SCT lenalidomide maintenance was feasible and contributed to further reduce MM tumor burden with PFS rates of 52% at three years;88 63% at three years,89 and 60% at two years.90 GvHD flares were observed in 28- 47% of cases.

Update on current studies At the 2016 American Society of Hematology meeting (December 2016), reports on myeloma and allo-SCT mainly focused on interactions of new drugs and GvM effect, and three groups unanimously reported remarkable responses to new drugs used as post-allo-SCT salvage, clearly showing a synergism with GvM effect. A retro-

Table 6. Allogeneic stem cell transplant at relapse

Study design

Patients

Tandem auto-allo-SCT 23 at first relapse, retrospective Allo-SCT RIC and MAC, retrospective 149 (121 RIC) Donor vs. no doner, 75 donor (68 allo-SCT) retrospective 94 no donor First relapse post auto-SCT: allo-SCT vs. 2nd auto-SCT, retrospective First relapse: MAC + lenalidomide maintenance RIC in relapse post auto-SCT, retrospective First relapse post auto: allo-SCT vs. auto-SCT, retrospective Allo-SCT at relapse, retrospective Allo-SCT at first relapse post auto-SCT, retrospective

19 allo-SCT

Median FU

PFS

Study ref

27 months

Median 36.8 months

61% at 2 years

111

28.5 months

15% at 5 years Donor 51% at 2 years

21% at 5 years Donor 42% at 2 years

112

No donor 53% at 2 years P=0.32 Median 6 months

No donor 18% at 2 years P<0.0001 Median 19 months

Median 19 months P=0.56

Median 27 months P=0.255

19 months 57 months from diagnosis

27 auto-SCT

113

114

19 months

52% at 3 years

79% at 3 years

413

Median 9.6 months

Median 24 months

6% at 3 years 12% at 3 years P=0.038 25% at 5 years 33% at 5 years P<0.0001 28% at 5 years

20% at 3 years 46% at 3 years P<0.001 10% at 5 years 15% at 5 years P<0.0001 57% at 5 years

152 allo-SCT 30 months 137 auto-SCT 639 before 2004 36 months 2766 after 2004 89

48 months

115

65 116

FU: follow up; Study ref.: references in literature; auto-SCT: autologous stem cell transplant; allo-SCT: allogeneic stem cell transplant; BU: busulfan; FLU: fludarabine; ATG: anti-thymocyte globuline; MEL: melphalan; TBI: total body irradiation; T: thalidomide; IFN: interferon. SCT: stem cell transplant; RIC: reduced intensity conditioning; MAC: myeloablative conditioning; EBMT: European Group for Blood and Marrow Transplantation; CIBMTR: Center for International Blood and Marrow Transplant Research. *In 99 patients completing allo-SCT program there was a prolonged progression-free survival (PFS) compared to 155 completing the other arm (P=0.04).

206

haematologica | 2018; 103(2)

Transplant and cellular therapy in myeloma

spective study comparing OS after relapse from upfront auto-allo (n=178) versus double auto-SCT (n=404) was conducted through the registry of the Center for International Blood and Marrow Transplant Research (CIBMTR).91 Despite a higher risk population (46% of early relapse from 2nd SCT vs. 26%) in the allo-SCT group, long-term reduction in post-relapse mortality (HR for death in auto-auto-SCT=1.55; P=0.0052) was observed. This was clearly attributable to improved response to salvage therapy due to the donor-derived immunological milieu that potentiated the immune effects of new agents. Similarly, Giaccone et al. showed prolonged OS from 1st relapse post tandem auto-allo-SCT compared to double auto-SCT (89.8 months vs. 23.5 months; P=0.009).92 López-Corral et al. reported similar pre-transplant and post-transplant response rates and durability of response achieved with new drugs before and after allo-SCT; responses post allo-SCT were at least similar in proportion and durability to those observed in the pre-transplant setting, which is in contrast to the usual course of the disease outside the allo-SCT setting.93 Another study reported on 18 high-risk MM patients who received upfront auto-SCT followed by RIC allo-SCT and bortezomib as maintenance, which was overall well tolerated, although 4 of 18 had asymptomatic Epstein Barr virus (EBV) reactivation. Depth of response improved after bortezomib, with 67% of patients in CR or stringent CR.94 Daratumumab has also demonstrated encouraging efficacy in 10 heavily pre-treated relapsed/refractory patients after allo-SCT. The safety profile was good with, in the majority of cases, non-severe adverse effects (AEs) mostly after the first infusion; 5 of 9 evaluable patients responded and all responding patients maintained their responses 7, 14, 35, 54 and 84 days after the first administration.95 Cook et al. monitored immune biomarkers with the use of lenalidomide after T-cell-depleted reduced intensity conditioning (RIC)-alloSCT, showing that the agent allowed sustained quantitative and functional reconstitution of donor immune homeostasis.96 McKiernan et al.97 reported a long-term comparison in patients receiving allo-SCT as upfront consolidation (n=75) or as salvage therapy (n=43). The 10-year OS for patients who received allo-SCT as salvage was 36% versus 68% for the consolidation group (P=0.0007). Of note, having undergone 2 or more prior auto-SCTs predicted for a higher risk of mortality (P=0.05). Chronic GvHD was favorable, associated with a 36% improvement in OS (P=0.0008).

Recommendations: previous studies that did not include novel agents reported long-term molecular remissions, and possibly cure, in patient subsets. Well-designed prospective trials combining GvM and new drugs may become urgent in young high-risk/ultra high-risk patients whose treatment remains an unmet clinical need. However, there are no current data supporting an upfront allograft. A clinical indication or recommendation may also become “early relapse” after first-line treatment (including the new PI and IMIDs) which identifies patients at very poor prognosis independent of other prognostic factors (Table 7). Re-induction to obtain tumor shrinking using novel drugs as a bridge to transplant is highly recommended/mandatory in this setting.98 Novel agent-based combinations should be considered also in association with DLI in case of relapse after allogeneic transplant.

Future developments Treatment for MM has undergone a dramatic improvement in the past decade given the considerable advances in the understanding of the disease pathogenesis and the approval of numerous novel drugs and combinations for the disease. However, despite the development of novel agents which target not only MM cells but also the microenvironment,99 the prognosis of patients with early relapsed/refractory MM remains poor. Thus, new therapeutic modalities are urgently needed to overcome resistance to current therapies. Several immunotherapies have recently been proposed which, among others, include monoclonal antibodies, antibody-drug conjugates, chimeric antigen receptor T-cell therapy (CAR-T cells), tumor vaccines and immune checkpoint inhibitors.100 Preliminary results observed in patients with B-cell hematologic malignancies with infusion of T cells genetically modified to express synthetic CARs against the lineagespecific surface antigen CD19 were impressive. T cells engineered with an anti-CD19 CAR induced CR also in a patient with MM.101 Recently, a number of other CAR-T cells have been designed to target surface antigens expressed by MM cells and include CD38,102 CD138,103 CD269, the B-cell maturation antigen (BCMA),104 κ light chains,105 CS1 (CD319)106 and CD44v6.107 However, despite their efficacy, CAR-T cells have raised many concerns on their short- and long-term toxicities, in particular, the development of life-threatening cytokine release syn-

Table 7. Recommendations for allografting in transplant eligible patients.

Level of evidence At diagnosis

At relapse

Maintenance

Practical considerations

Rationale for considerations

Clinical trial in young ultra high-risk/high-risk patients.

Though results were discordant, prospective randomized studies (designed in the late ’90s – early 2000s) showed long-term disease control in subsets of patients who were not treated with new drugs; the combination of new drugs and graft-versus-myeloma may be of benefit in patients where prognosis remains currently very poor. Regardless of prognostic features, early relapse is overall associated with poor diagnosis. Retrospective studies support the existence of a potential benefit of graft-versus-myeloma in this setting. The inclusion in control trials would be recommended. Maintenance therapy is currently part of prospective trials open to accrual. post allografting. Results are eagerly awaited.

Young patients with early relapse (18 months) from first-line treatment with/without high-risk features. Clinical trial.

haematologica | 2018; 103(2)

207

F. Gay et al.

drome (CRS) and prolonged aplasia of the healthy counterparts.108 Genetic modifications of cells belonging to the innate immune system, such as natural killer (NK) cells, are also being explored, and modification of the human NK-cell lines NKL and NK-92 with a lentiviral vector encoding for CS1 and CD138 CARs has proven to be feasible.109 However, several steps to optimize and validate CAR-modified NK cells have to be undertaken before their wider clinical use can be considered.

Conclusions Over the last two decades, changes in the treatment paradigm for MM patients have dramatically improved survival. Clearly, results of the most recently published trials confirm the role of ASCT in the era of novel agents, with new drugs administered both in the pre-transplant and post-transplant phases. The expert panel emphasizes that current clinical research should maintain a balance between treatment efficacy and quality of life, identify the optimal sequencing of treatment, the appropriate tools for patient selection, evaluate costs of prolonged novel-agent application versus transplant remission efficacy, and treatment-free intervals, and it should identify how to best induce long-term remission.110 In the future, objective, prospective and proficiently performed fitness tools may prove to be of benefit before intensive treatment is start-

References 1. Kumar SK, Dispenzieri A, Lacy MQ, et al. Continued improvement in survival in multiple myeloma: changes in early mortality and outcomes in older patients. Leukemia. 2014;28(5):1122-1128. 2. San Miguel JF, Schlag R, Khuageva NK, et al. Bortezomib plus melphalan and prednisone for initial treatment of multiple myeloma. N Engl J Med. 2008;359(9):906-917. 3. Benboubker L, Dimopoulos MA, Dispenzieri A, et al. Lenalidomide and dexamethasone in transplant-ineligible patients with myeloma. N Engl J Med. 2014;371(10): 906-917. 4. Sonneveld P, Schmidt-Wolf IGH, van der Holt B, et al. Bortezomib induction and maintenance treatment in patients with newly diagnosed multiple myeloma: results of the randomized phase III HOVON-65/ GMMG-HD4 trial. J Clin Oncol. 2012;30(24): 2946-2955. 5. Mai EK, Bertsch U, Dürig J, et al. Phase III trial of bortezomib, cyclophosphamide and dexamethasone (VCD) versus bortezomib, doxorubicin and dexamethasone (PAd) in newly diagnosed myeloma. Leukemia. 2015;29(8):1721-1729. 6. Rosiñol L, Oriol A, Teruel AI, et al. Superiority of bortezomib, thalidomide, and dexamethasone (VTD) as induction pretransplantation therapy in multiple myeloma: a randomized phase 3 PETHEMA/GEM study. Blood. 2012;120(8):1589-1596. 7. Cavo M, Tacchetti P, Patriarca F, et al. Bortezomib with thalidomide plus dexamethasone compared with thalidomide plus dexamethasone as induction therapy before, and consolidation therapy after, dou-

208

10.

11.

12.

13.

ed, especially since fitness assessments made by patients and physicians are not as objective as fitness evaluations derived from well-defined tests and scores. Future randomized studies will also need to evaluate the role of ASCT as salvage treatment in the context of the novel combinations currently available as salvage options. The trend in survival improvement is likely to continue in the future with new classes of drugs [such as monoclonal antibodies (MoAbs)] and 2nd-generation PIs and IMIDs moving in the upfront setting. If most patients can now expect long-term disease control, the optimal definition of high-risk disease and the specific treatment for these patients remains a major challenge. Based on the available data, the opinion of the expert committee is that allotransplant in combination with novel agents might be considered in the context of clinical trials for high-risk patients who are willing to accept the TRM for a chance of a better long-term survival. Moreover, cellular therapies, that for the moment are still highly experimental, should be optimized and made more widely available and cost approved so they can be included in our treatment armamentarium. Acknowledgments This work is supported by the Deutsche Krebshilfe (grants 1095969 and 111424 to ME and RW). The expert panel thanks all the investigators of the EMN group in the different countries for their support.

ble autologous stem-cell transplantation in newly diagnosed multiple myeloma: a randomised phase 3. Lancet. 2010;376(9758): 2075-2085. Moreau P, Hulin C, Macro M, et al. VTD is superior to VCD prior to intensive therapy in multiple myeloma: results of the prospective IFM2013-04 trial. Blood. 2016;127(21): 2569-2574. Roussel M, Lauwers-Cances V, Robillard N, et al. Front-line transplantation program with lenalidomide, bortezomib, and dexamethasone combination as induction and consolidation followed by lenalidomide maintenance in patients with multiple myeloma: a phase II study by the Intergroupe Francophone du Myélo. J Clin Oncol. 2014;32(25):2712-2717. Attal M, Lauwers-Cances V, Hulin C, et al. Lenalidomide, Bortezomib, and Dexamethasone with Transplantation for Myeloma. N Engl J Med. 2017;376(14): 1311-1320. Moreau P, Hulin C, Caillot D, et al. Ixazomib-Lenalidomide-Dexamethasone (IRd) Combination before and after Autologous Stem Cell Transplantation (ASCT) Followed By Ixazomib Maintenance in Patients with Newly Diagnosed Multiple Myeloma (NDMM): A Phase 2 Study from the Intergroupe Francophone. Blood. 2016;128(22):674. Wester R, van der Holt B, Asselbergs E, et al. Phase 2 Study of Carfilzomib, Thalidomide, and Low-Dose Dexamethasone As Induction/Consolidation in Newly Diagnosed, Transplant Eligible Patients with Multiple Myeloma, the Carthadex Trial. Blood. 2016;128(22):1141. Roussel M, Lauwers-Cances V, Robillard N, et al. Frontline Therapy with Carfilzomib,

14.

15.

16.

17.

18.

19.

20.

Lenalidomide, and Dexamethasone (KRd) Induction Followed By Autologous Stem Cell Transplantation, Krd Consolidation and Lenalidomide Maintenance in Newly Diagnosed Multiple Myeloma (NDMM) Patients: Primary Results of the Intergroupe Francophone Du MyéLome (IFM) Krd Phase II Study. Blood. 2016;128(22):1142. Ferrero S, Ladetto M, Drandi D, et al. Longterm results of the GIMEMA VEL-03-096 trial in MM patients receiving VTD consolidation after ASCT: MRD kinetics’ impact on survival. Leukemia. 2015;29(3):689-695. Jackson GH, Davies FE, Pawlyn C, et al. Response Adapted Induction Treatment Improves Outcomes for Myeloma Patients; Results of the Phase III Myeloma XI Study. Blood. 2016;128(22):244. Sonneveld P, Goldschmidt H, Rosiñol L, et al. Bortezomib-based versus nonbortezomib-based induction treatment before autologous stem-cell transplantation in patients with previously untreated multiple myeloma: a meta-analysis of phase III randomized, controlled trials. J Clin Oncol. 2013;31(26):3279-3287. Nooka AK, Kaufman JL, Behera M, et al. Bortezomib-containing induction regimens in transplant-eligible myeloma patients. Cancer. 2013;119(23):4119-4128. Shah N, Callander N, Ganguly S, et al. Hematopoietic Stem Cell Transplantation for Multiple Myeloma: Guidelines from the American Society for Blood and Marrow Transplantation. Biol. Blood Marrow Transplant. 2015;21(7):1155-1166. Palumbo A, Cavallo F, Gay F, et al. Autologous transplantation and maintenance therapy in multiple myeloma. N Engl J Med. 2014;371(10):895-905. Gay F, Oliva S, Petrucci MT, et al.

haematologica | 2018; 103(2)

Transplant and cellular therapy in myeloma

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

Chemotherapy plus lenalidomide versus autologous transplantation, followed by lenalidomide plus prednisone versus lenalidomide maintenance, in patients with multiple myeloma: a randomised, multicentre, phase 3 trial. Lancet Oncol. 2015;16(16): 1617-1629. Cavo M, Beksac M, Dimopoulos MA, et al. Intensification Therapy with BortezomibMelphalan-Prednisone Versus Autologous Stem Cell Transplantation for Newly Diagnosed Multiple Myeloma: An Intergroup, Multicenter, Phase III Study of the European Myeloma Network (EMN02/HO95 MM Trial). Blood. 2016;128(22):673. Fermand JP, Alberti C, Marolleau JP. Single versus tandem high dose therapy (HDT) supported with autologous blood stem cell (ABSC) transplantation using unselected or CD34-enriched ABSC: Results of a two by two designed randomized trial in 230 young patients with multiple myeloma (MM). Hematol J. 2003;4(Suppl 1):S59. Attal M, Harousseau J-L, Facon T, et al. Single versus Double Autologous Stem-Cell Transplantation for Multiple Myeloma. N Engl J Med. 2003;349(26):2495-2502. Cavo M, Salwender H, Rosiñol L, et al. Double Vs Single Autologous Stem Cell Transplantation After Bortezomib-Based Induction Regimens For Multiple Myeloma: An Integrated Analysis Of Patient-Level Data From Phase European III Studies. Blood. 2013;122(21):767. Cavo M, Petrucci MT, Di Raimondo F, et al. Upfront Single Versus Double Autologous Stem Cell Transplantation for Newly Diagnosed Multiple Myeloma: An Intergroup, Multicenter, Phase III Study of the European Myeloma Network (EMN02/HO95 MM Trial). Blood. 2016;128 (22):991. Stadtmauer EA, Pasquini MC, Blackwell B, et al. Comparison of Autologous Hematopoietic Cell Transplant (autoHCT), Bortezomib, Lenalidomide (Len) and Dexamethasone (RVD) Consolidation with Len Maintenance (ACM), Tandem Autohct with Len Maintenance (TAM) and Autohct with Len Maintenance (AM) for up-Front. Blood. 2016;128(22):LBA-1. Mellqvist U-H, Gimsing P, Hjertner O, et al. Bortezomib consolidation after autologous stem cell transplantation in multiple myeloma: a Nordic Myeloma Study Group randomized phase 3 trial. Blood. 2013;121(23): 4647-4654. Cavo M, Pantani L, Petrucci MT, et al. Bortezomib-thalidomide-dexamethasone is superior to thalidomide-dexamethasone as consolidation therapy after autologous hematopoietic stem cell transplantation in patients with newly diagnosed multiple myeloma. Blood. 2012;120(1):9-19. Sonneveld P, Beksac M, van der Holt B, et al. Consolidation Followed By Maintenance Therapy Versus Maintenance Alone in Newly Diagnosed, Transplant Eligible Patients with Multiple Myeloma (MM): A Randomized Phase 3 Study of the European Myeloma Network (EMN02/HO95 MM Trial). ASH. 2016;128(22):242. Attal M, Harousseau J-L, Leyvraz S, et al. Maintenance therapy with thalidomide improves survival in patients with multiple myeloma. Blood. 2006;108(10):3289–3294. Morgan GJ, Gregory WM, Davies FE, et al. The role of maintenance thalidomide therapy in multiple myeloma: MRC Myeloma IX results and meta-analysis. Blood. 2012;119 (1):7-15.

haematologica | 2018; 103(2)

32. Goldschmidt H, Lokhorst HM, Mai EK, et al. Bortezomib before and after high-dose therapy in myeloma: long-term results from the phase III HOVON-65/GMMG-HD4 trial. Leukemia. 2017 Jul 4. [Epub ahead of print] 33. Rosiñol L, Oriol A, Teruel AI, et al. Bortezomib and thalidomide maintenance after stem cell transplantation for multiple myeloma: a PETHEMA/GEM trial. Leukemia. 2017;31(9):1922-1927. 34. Attal M, Lauwers-Cances V, Marit G, et al. Lenalidomide Maintenance after Stem-Cell Transplantation for Multiple Myeloma. N Engl J Med. 2012;366(19):1782-1791. 35. McCarthy PL, Owzar K, Hofmeister CC, et al. Lenalidomide after stem-cell transplantation for multiple myeloma. N Engl J Med. 2012;366(19):1770-1781. 36. Jackson GH, Davies FE, Pawlyn C, et al. Lenalidomide Is a Highly Effective Maintenance Therapy in Myeloma Patients of All Ages; Results of the Phase III Myeloma XI Study. Blood. 2016;128 (22):1143. 37. McCarthy PL, Holstein SA, Petrucci MT, et al. Lenalidomide Maintenance After Autologous Stem-Cell Transplantation in Newly Diagnosed Multiple Myeloma: A Meta-Analysis. J Clin Oncol. 2017;35(29): 3279-3289. 38. Saad A, Mahindra A, Zhang M-J, et al. Hematopoietic cell transplant comorbidity index is predictive of survival after autologous hematopoietic cell transplantation in multiple myeloma. Biol Blood Marrow Transplant. 2014;20(3):402-408.e1. 39. Labonté L, Iqbal T, Zaidi MA, et al. Utility of comorbidity assessment in predicting transplantation-related toxicity following autologous hematopoietic stem cell transplantation for multiple myeloma. Biol Blood Marrow Transplant. 2008;14(9):1039-1044. 40. Engelhardt M, Domm A-S, Dold SM, et al. A concise revised myeloma comorbidity Index as a valid prognostic instrument in a large cohort of 801 multiple myeloma patients. Haematologica. 2017;102(5):910-921. 41. Scheid C, Sonneveld P, Schmidt-Wolf IGH, et al. Bortezomib before and after autologous stem cell transplantation overcomes the negative prognostic impact of renal impairment in newly diagnosed multiple myeloma: a subgroup analysis from the HOVON-65/GMMG-HD4 trial. Haematologica. 2014;99(1):148-154. 42. Breitkreutz I, Heiss C, Perne A, et al. Bortezomib improves outcome after SCT in multiple myeloma patients with end-stage renal failure. Bone Marrow Transplant. 2014;49(11):1371-1375. 43. Zannetti BA, Zamagni E, Santostefano M, et al. Bortezomib-based therapy combined with high cut-off hemodialysis is highly effective in newly diagnosed multiple myeloma patients with severe renal impairment. Am J Hematol. 2015;90(7):647–652. 44. Gavriatopoulou M, Terpos E, Kastritis E, Dimopoulos MA. Current treatments for renal failure due to multiple myeloma. Expert Opin Pharmacother. 2016;17(16): 2165-2177. 45. Dimopoulos MA, Sonneveld P, Leung N, et al. International Myeloma Working Group Recommendations for the Diagnosis and Management of Myeloma-Related Renal Impairment. J Clin Oncol. 2016;34(13): 1544-1557. 46. Mahindra A, Hari P, Fraser R, et al. Autologous hematopoietic cell transplantation for multiple myeloma patients with renal insufficiency: a center for international blood and marrow transplant research

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

analysis. Bone Marrow Transplant. 2017 Sep 18. [Epub ahead of print] Ozaki S, Shimizu K. Autologous stem cell transplantation in elderly patients with multiple myeloma: past, present, and future. Biomed Res Int. 2014;2014:394792. Straka C, Liebisch P, Salwender H, et al. Autotransplant with and without induction chemotherapy in older multiple myeloma patients: long-term outcome of a randomized trial. Haematologica. 2016;101(11): 1398-1406. Garderet L, Beohou E, Caillot D, et al. Upfront autologous stem cell transplantation for newly diagnosed elderly multiple myeloma patients: a prospective multicenter study. Haematologica. 2016;101(11):13901397. Auner HW, Szydlo R, Hoek J, et al. Trends in autologous hematopoietic cell transplantation for multiple myeloma in Europe: increased use and improved outcomes in elderly patients in recent years. Bone Marrow Transplant. 2015;50(2):209-215. Facon T, Mary JY, Hulin C, et al. Melphalan and prednisone plus thalidomide versus melphalan and prednisone alone or reducedintensity autologous stem cell transplantation in elderly patients with multiple myeloma (IFM 99-06): a randomised trial. Lancet. 2007;370(9594):1209-1218. Gay F, Magarotto V, Crippa C, et al. Bortezomib induction, reduced-intensity transplantation, and lenalidomide consolidation-maintenance for myeloma: updated results. Blood. 2013;122(8):1376-1383. Straka C, Knop S, Vogel M, et al. Bortezomib Consolidation Following Autologous Transplant Equalizes the Outcome for Older Patients with Less Intensive Pretreatment Compared to Younger Patients with Newly Diagnosed Multiple Myeloma. Blood. 2016;128 (22):516. Straka C, Schaefer-Eckart K, Bassermann F, et al. Lenalidomide with Low-Dose Dexamethasone (Rd) Continuously Versus Rd Induction, Tandem MEL140 with Autologous Transplantation and Lenalidomide Maintenance: Planned Interim Analysis of a Prospective Randomized Trial in Patients 60-75 Years of Age with Mult. Blood. 2014;124(21):3969. Fermand JP, Ravaud P, Chevret S, et al. Highdose therapy and autologous peripheral blood stem cell transplantation in multiple myeloma: up-front or rescue treatment? Results of a multicenter sequential randomized clinical trial. Blood. 1998;92(9):31313136. Koreth J, Cutler CS, Djulbegovic B, et al. High-dose therapy with single autologous transplantation versus chemotherapy for newly diagnosed multiple myeloma: A systematic review and meta-analysis of randomized controlled trials. Biol Blood Marrow Transplant. 2007;13(2):183-196. Gay F, Oliva S, Petrucci MT, et al. Autologous transplant vs oral chemotherapy and lenalidomide in newly diagnosed young myeloma patients: a pooled analysis. Leukemia. 2017;31(8):1727-1734. Jimenez-Zepeda VH, Mikhael J, Winter A, et al. Second Autologous Stem Cell Transplantation as Salvage Therapy for Multiple Myeloma: Impact on ProgressionFree and Overall Survival. Biol Blood Marrow Transplant. 2012;18(5):773-779. Auner HW, Szydlo R, Rone A, et al. Salvage autologous stem cell transplantation for multiple myeloma relapsing or progressing after up-front autologous transplantation.

209

F. Gay et al. Leuk Lymphoma. 2013;54(10):2200-2204. 60. Grövdal M, Nahi H, Gahrton G, et al. Autologous stem cell transplantation versus novel drugs or conventional chemotherapy for patients with relapsed multiple myeloma after previous ASCT. Bone Marrow Transplant. 2015;50(6):808-812. 61. Cook G, Ashcroft AJ, Cairns DA, et al. The effect of salvage autologous stem-cell transplantation on overall survival in patients with relapsed multiple myeloma (final results from BSBMT/UKMF Myeloma X Relapse [Intensive]): a randomised, openlabel, phase 3 trial. Lancet Haematol. 2016;3(7):e340-e351. 62. Garderet L, Iacobelli S, van Biezen A, et al. Outcome of Third Salvage Autologous Stem Cell Transplantation in Multiple Myeloma. Blood. 2016;128(22):993. 63. Moreau P, San Miguel J, Sonneveld P, et al. Multiple myeloma: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol 2017;28(suppl_4): iv52iv61. 64. Kumar S, Zhang M-J, Li P, et al. Trends in allogeneic stem cell transplantation for multiple myeloma: a CIBMTR analysis. Blood. 2011;118(7):1979-1988. 65. Sobh M, Michallet M, Gahrton G, et al. Allogeneic hematopoietic cell transplantation for multiple myeloma in Europe: trends and outcomes over 25 years. A study by the EBMT Chronic Malignancies Working Party. Leukemia. 2016;30(10):2047-2054. 66. Bruno B, Rotta M, Patriarca F, et al. A Comparison of Allografting with Autografting for Newly Diagnosed Myeloma. N Engl J Med. 2007;356(11): 1110-1120. 67. Garban F, Attal M, Michallet M, et al. Prospective comparison of autologous stem cell transplantation followed by dosereduced allograft (IFM99-03 trial) with tandem autologous stem cell transplantation (IFM99-04 trial) in high-risk de novo multiple myeloma. Blood. 2006;107(9):34743480. 68. Moreau P, Garban F, Attal M, et al. Longterm follow-up results of IFM99-03 and IFM99-04 trials comparing nonmyeloablative allotransplantation with autologous transplantation in high-risk de novo multiple myeloma. Blood. 2008;112(9):3914-3915. 69. Rosinol L, Perez-Simon JA, Sureda A, et al. A prospective PETHEMA study of tandem autologous transplantation versus autograft followed by reduced-intensity conditioning allogeneic transplantation in newly diagnosed multiple myeloma. Blood. 2008;112 (9):3591-3593. 70. Krishnan A, Pasquini MC, Logan B, et al. Autologous haemopoietic stem-cell transplantation followed by allogeneic or autologous haemopoietic stem-cell transplantation in patients with multiple myeloma (BMT CTN 0102): a phase 3 biological assignment trial. Lancet Oncol. 2011;12(13):1195-1203. 71. Lokhorst HM, van der Holt B, Cornelissen JJ, et al. Donor versus no-donor comparison of newly diagnosed myeloma patients included in the HOVON-50 multiple myeloma study. Blood. 2012;119(26):6219-6225. 72. Björkstrand B, Iacobelli S, Hegenbart U, et al. Tandem Autologous/Reduced-Intensity Conditioning Allogeneic Stem-Cell Transplantation Versus Autologous Transplantation in Myeloma: Long-Term Follow-Up. J Clin Oncol. 2011;29(22):30163022. 73. Giaccone L, Storer B, Patriarca F, et al. Longterm follow-up of a comparison of nonmyeloablative allografting with autografting for

210

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

86.

newly diagnosed myeloma. Blood. 2011;117 (24):6721-6727. Gahrton G, Iacobelli S, Björkstrand B, et al. Autologous/reduced-intensity allogeneic stem cell transplantation vs autologous transplantation in multiple myeloma: longterm results of the EBMT-NMAM2000 study. Blood. 2013;121(25):5055-5063. Bruno B, Sorasio R, Patriarca F, et al. Unrelated donor haematopoietic cell transplantation after non-myeloablative conditioning for patients with high-risk multiple myeloma. Eur J Haematol. 2007;78(4):330337. Georges GE, Maris MB, Maloney DG, et al. Nonmyeloablative Unrelated Donor Hematopoietic Cell Transplantation to Treat Patients with Poor-Risk, Relapsed, or Refractory Multiple Myeloma. Biol Blood Marrow Transplant. 2007;13(4):423-432. Auner HW, Szydlo R, van Biezen A, et al. Reduced intensity-conditioned allogeneic stem cell transplantation for multiple myeloma relapsing or progressing after autologous transplantation: a study by the European Group for Blood and Marrow Transplantation. Bone Marrow Transplant. 2013;48(11):1395-1400. Kröger N, Shimoni A, Schilling G, et al. Unrelated stem cell transplantation after reduced intensity conditioning for patients with multiple myeloma relapsing after autologous transplantation. Br J Haematol. 2010;148(2):323-331. Kröger N, Badbaran A, Zabelina T, et al. Impact of High-Risk Cytogenetics and Achievement of Molecular Remission on Long-Term Freedom from Disease after Autologous–Allogeneic Tandem Transplantation in Patients with Multiple Myeloma. Biol Blood Marrow Transplant. 2013;19(3):398-404. Roos-Weil D, Moreau P, Avet-Loiseau H, et al. Impact of genetic abnormalities after allogeneic stem cell transplantation in multiple myeloma: a report of the Societe Francaise de Greffe de Moelle et de Therapie Cellulaire. Haematologica. 2011;96(10): 1504-1511. Beitinjaneh AM, Saliba R, Bashir Q, et al. Durable responses after donor lymphocyte infusion for patients with residual multiple myeloma following non-myeloablative allogeneic stem cell transplant. Leuk. Lymphoma. 2012;53(8):1525-1529. Crawley C, Lalancette M, Szydlo R, et al. Outcomes for reduced-intensity allogeneic transplantation for multiple myeloma: an analysis of prognostic factors from the Chronic Leukaemia Working Party of the EBMT. Blood. 2005;105(11):4532-4539. Ladetto M, Ferrero S, Drandi D, et al. Prospective molecular monitoring of minimal residual disease after non-myeloablative allografting in newly diagnosed multiple myeloma. Leukemia. 2016;30(5):1211-1214. Passera R, Pollichieni S, Brunello L, et al. Allogeneic Hematopoietic Cell Transplantation from Unrelated Donors in Multiple Myeloma: Study from the Italian Bone Marrow Donor Registry. Biol Blood Marrow Transplant. 2013;19(6):940-948. Donato ML, Siegel DS, Vesole DH, et al. The Graft-Versus-Myeloma Effect: Chronic GraftVersus-Host Disease but Not Acute GraftVersus-Host Disease Prolongs Survival in Patients with Multiple Myeloma Receiving Allogeneic Transplantation. Biol Blood Marrow Transplant. 2014;20(8):1211-1216. Michallet M, Sobh M, El-Cheikh J, et al. Evolving strategies with immunomodulating drugs and tandem autologous/allogeneic

87.

88.

89.

90.

91.

92.

93.

94.

95.

96.

97.

98.

99.

hematopoietic stem cell transplantation in first line high risk multiple myeloma patients. Exp Hematol. 2013;41(12):10081015. Caballero-Velázquez T, López-Corral L, Encinas C, et al. Phase II clinical trial for the evaluation of bortezomib within the reduced intensity conditioning regimen (RIC) and post-allogeneic transplantation for high-risk myeloma patients. Br J Haematol. 2013;162(4):474-482. Kröger N, Zabelina T, Klyuchnikov E, et al. Toxicity-reduced, myeloablative allograft followed by lenalidomide maintenance as salvage therapy for refractory/relapsed myeloma patients. Bone Marrow Transplant. 2013;48(3):403-407. Kneppers E, van der Holt B, Kersten M-J, et al. Lenalidomide maintenance after nonmyeloablative allogeneic stem cell transplantation in multiple myeloma is not feasible: results of the HOVON 76 Trial. Blood. 2011;118(9):2413-2419. Alsina M, Becker PS, Zhong X, et al. Lenalidomide maintenance for high-risk multiple myeloma after allogeneic hematopoietic cell transplantation. Biol Blood Marrow Transplant. 2014;20(8):11831189. Htut M, D’Souza A, Bruno B, et al. Survival after Relapse Following Tandem Allogeneic Vs. Tandem Autologous Hematopoietic Cell Transplantation (HCT) for Myeloma (MM). Blood. 2016;128(22):833. Giaccone L, Evangelista A, Patriarca F, et al. Prolonged Follow-up Confirmed a Role for Upfront Tandem Auto-Allo Transplant in Multiple Myeloma Also in the Era of New Drugs. Blood. 2016;128(22):3469. López Corral L, Caballero Velázquez T, López Godino O, et al. Response to Proteosome Inhibitors and Immunomodulatory Drugs before and after Allogeneic Transplantation in Patients with Multiple Myeloma: A Long Term Follow up Study. Blood. 2016;128(22):3436. LeBlanc R, Ahmad I, Terra R, et al. Bortezomib Consolidation after Nonmyeloablative Allogeneic Stem Cell Transplantation Leads to a High Incidence of Immunophenotypic Complete Response in Young and/or High-Risk Multiple Myeloma Patients. Blood. 2016;128(22):2306. Klyuchnikov E, von Pein U-M, Ayuk FA, et al. Daratumumab Is an Effective and Safe Salvage Therapy in Relapsed/Refractory Patients with Multiple Myeloma after Allogeneic Stem Cell Transplantation. Blood. 2016;128(22):3437. Cook G, Carter CR, Brock K, et al. Immune Biomarkers Identify Sustained Quantitative and Functional Immune Reconstitution in the Setting of Adjunctive Lenalidomide Following T-Depleted RIC-Allo SCT for Multiple Myeloma. Blood. 2016;128 (22):4585. McKiernan P, Siegel DS, Vesole DH, et al. Long-Term Survival Is Demonstrated in Patients with Multiple Myeloma Treated with Allogeneic Hematopoietic Stem Cell Transplantation in Both the Consolidation and Salvage Settings. Blood. 2016;128 (22):2302. Patriarca F, Giaccone L, Onida F, et al. New drugs and allogeneic hematopoietic stem cell transplantation for hematological malignancies: do they have a role in bridging, consolidating or conditioning transplantation treatment? Expert Opin Biol Ther. 2017;17(7):821-836. Ludwig H, Weisel K, Petrucci MT, et al. Olaptesed pegol, an anti-CXCL12/SDF-1

haematologica | 2018; 103(2)

Transplant and cellular therapy in myeloma Spiegelmer, alone and with bortezomibdexamethasone in relapsed/refractory multiple myeloma: a Phase IIa Study. Leukemia. 2017;31(4):997-1000. 100. Rodríguez-Otero P, Paiva B, Engelhardt M, Prósper F, San Miguel JF. Is immunotherapy here to stay in multiple myeloma? Haematologica. 2017;102(3):423-432. 101. Garfall AL, Maus M V, Hwang W-T, et al. Chimeric Antigen Receptor T Cells against CD19 for Multiple Myeloma. N Engl J Med. 2015;373(11):1040-1047. 102. Suck G, Odendahl M, Nowakowska P, et al. NK-92: an “off-the-shelf therapeutic” for adoptive natural killer cell-based cancer immunotherapy. Cancer Immunol Immunother. 2016;65(4):485-492. 103. Jiang H, Zhang W, Shang P, et al. Transfection of chimeric anti-CD138 gene enhances natural killer cell activation and killing of multiple myeloma cells. Mol Oncol. 2014;8(2):297-310. 104. Tai Y-T, Anderson KC. Targeting B-cell maturation antigen in multiple myeloma. Immunotherapy. 2015;7(11):1187-1199. 105. Yang J, Zhu H, Tan Z, et al. Comparison of two functional kappa light-chain transcripts amplified from a hybridoma. Biotechnol Appl Biochem. 2013;60(3):289-297. 106. Chu J, Deng Y, Benson DM, et al. CS1-spe-

haematologica | 2018; 103(2)

cific chimeric antigen receptor (CAR)-engineered natural killer cells enhance in vitro and in vivo antitumor activity against human multiple myeloma. Leukemia. 2014;28(4):917-927. 107. Casucci M, Nicolis di Robilant B, Falcone L, et al. CD44v6-targeted T cells mediate potent antitumor effects against acute myeloid leukemia and multiple myeloma. Blood. 2013;122(20):3461-3472. 108. Brudno JN, Kochenderfer JN. Toxicities of chimeric antigen receptor T cells: recognition and management. Blood. 2016;127(26): 3321-3330. 109. Chu J, Deng Y, Benson DM, et al. CS1-specific chimeric antigen receptor (CAR)-engineered natural killer cells enhance in vitro and in vivo antitumor activity against human multiple myeloma. Leukemia. 2014;28(4):917-927. 110. Engelhardt M, Ihorst G, Caers J, Gunther A, Wasch R. Autotransplants in older multiple myeloma patients: hype or hope in the era of novel agents? Haematologica. 2016;101(11):1276-1278. 111. Karlin L, Arnulf B, Chevret S, et al. Tandem autologous non-myeloablative allogeneic transplantation in patients with multiple myeloma relapsing after a first high dose therapy. Bone Marrow Transplant. 2011;46

(2):250-256. 112. Bashir Q, Khan H, Orlowski RZ, et al. Predictors of prolonged survival after allogeneic hematopoietic stem cell transplantation for multiple myeloma. Am J Hematol. 2012;87(3):272-276. 113. Patriarca F, Einsele H, Spina F, et al. Allogeneic Stem Cell Transplantation in Multiple Myeloma Relapsed after Autograft: A Multicenter Retrospective Study Based on Donor Availability. Biol Blood Marrow Transplant. 2012;18(4):617-626. 114. Wirk B, Byrne M, Dai Y, Moreb JS. Outcomes of Salvage Autologous Versus Allogeneic Hematopoietic Cell Transplantation for Relapsed Multiple Myeloma After Initial Autologous Hematopoietic Cell Transplantation. J Clin Med Res. 2013;5(3):174-184. 115. Freytes CO, Vesole DH, LeRademacher J, et al. Second transplants for multiple myeloma relapsing after a previous autotransplantreduced-intensity allogeneic vs autologous transplantation. Bone Marrow Transplant. 2014;49(3):416-421. 116. Klyuchnikov E, Wolschke C, Badbaran A, et al. Allogeneic Stem Cell Transplantation As Salvage Therapy for First Relapse after Autografting in Multiple Myeloma Patients. Blood. 2016;128(22):4619.

211

ARTICLE

Bone Marrow Failure

Ferrata Storti Foundation

Haematologica 2018 Volume 103(2):212-220

Nationwide survey on the use of eltrombopag in patients with severe aplastic anemia: a report on behalf of the French Reference Center for Aplastic Anemia Etienne Lengline,1 Bernard Drenou,2 Pierre Peterlin,3 Olivier Tournilhac,4 Julie Abraham,5 Ana Berceanu,6 Brigitte Dupriez,7 Gaelle Guillerm,8 Emmanuel Raffoux,1 Flore Sicre de Fontbrune,1 Lionel Ades,1 Marie Balsat,9 Driss Chaoui,10 Paul Coppo,11 Selim Corm,12 Thierry Leblanc,1,13 Natacha Maillard,14 Louis Terriou,15 Gerard Socié1,16* and Regis Peffault de Latour1*

Department of Hematology, CRNMR Aplasie Médullaire, Saint-Louis University Hospital – AP-HP, Paris; 2Department of Hematology, Hôpital Emile Muller - CH de Mulhouse; 3 Department of Hematology, Nantes University Hospital; 4Service d'Hematologie Clinique et de Therapie Cellulaire, CHU, Universite d'Auvergne, Clermont-Ferrand; 5 Service d'Hématologie Clinique et Thérapie Cellulaire, CHU de Limoges; 6Department of Hematology, Besançon University Hospital; 7Department of Hematology, Centre hospitalier de Lens; 8Department of Hematology and Oncology, CH Augustin Morvan, Brest; 9 Department of Hematology 1G, Centre Hospitalier Lyon Sud, Pierre Benite; 10 Department of Hematology, CH Victor Dupouy, Argenteuil; 11Department of Hematology, French Reference Center for Thrombotic Microangiopathies, Saint Antoine University Hospital, Paris; 12Department of Hematology, Hôpital Privé Médipole de Savoie, Challes les Eaux; 13Department of Pediatric Hematology, Robert-Debré University Hospital, Paris; 14Bone Marrow Transplant Unit Clinical Hematology, Hopital La Miletrie, Poitiers University Hospital; 15Department of Internal Medicine, Clinical Immunology, Hôpital Huriez Lille University Hospital and 16University Paris Denis Diderot & INSERM UMR 1160, France 1

*GS and RPL contributed equally to this work.

ABSTRACT

Correspondence: etienne.lengline@aphp.fr

Received: July 12, 2017. Accepted: November 15, 2017. Pre-published: November 23, 2017. doi:10.3324/haematol.2017.176339 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/212 ©2018 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.

212

ew therapeutic options are available for patients with aplastic anemia who are ineligible for transplantation or refractory to immunosuppressive therapy. Eltrombopag was recently shown to produce trilineage responses in refractory patients. However, the effects of real-life use of this drug remain unknown. This retrospective study (2012-2016) was conducted by the French Reference Center for Aplastic Anemia on patients with relapsed/refractory aplastic anemia, and patients ineligible for antithymocyte globulin or transplantation, who received eltrombopag for at least 2 months. Forty-six patients with aplastic anemia were given eltrombopag without prior antithymocyte globulin treatment (n=11) or after antithymocyte globulin administration (n=35) in a relapsed/refractory setting. Eltrombopag (median daily dose 150 mg) was introduced 17 months (range, 8-50) after the diagnosis of aplastic anemia. At last followup, 49% were still receiving treatment, 9% had stopped due to a robust response, 2% due to toxicity and 40% due to eltrombopag failure. Before eltrombopag treatment, all patients received regular transfusions. The overall rates of red blood cell and platelet transfusion independence were 7%, 33%, 46% and 46% at 1, 3, 6 months and last follow-up. Responses were slower to develop in antithymocyte treatment-naïve patients. In patients achieving transfusion independence, hemoglobin concentration and platelet counts improved by 3 g/dL (interquartile range, 1.4-4.5) and 42x109/L (interquartile range, 11-100), respectively. Response in at least one lineage (according to National Institutes of Health criteria) was observed in 64% of antithymocyte treatment-naïve and 74% of relapsed/refractory patients, while trilineage improvement was observed in 27% and 34%, respectively. We found high rates of hematologic improvement and transfusion independence in refractory aplastic anemia patients but also in patients ineligible for antithymocyte globulin receiving first-line treatment. In conclusion, elderly patients unfit for antithymocyte globulin therapy may benefit from eltrombopag. haematologica | 2018; 103(2)

Eltrombopag in aplastic anemia

Introduction Aplastic anemia (AA) is caused by the destruction of hematopoietic stem cells, leading to pancytopenia. Rapid front-line bone marrow transplantation with an HLAidentical sibling donor can lead to excellent outcomes.1â&#x20AC;&#x201C;4 However, most patients cannot undergo such a procedure because of the absence of a sibling donor or because of their age and/or co-morbidities. Immunosuppressive therapy with horse antithymocyte globulin (ATG) plus cyclosporine A (CsA) is considered to be the standard treatment in this situation, producing an overall hematologic response rate of 60-70%.5â&#x20AC;&#x201C;7 Nevertheless, few therapeutic options are currently open to patients with AA who fail to achieve a hematologic response or those who relapse after this therapy and are ineligible for allogeneic stem cell transplantation. Moreover, in elderly patients who cannot receive ATG, the chance of obtaining a hematologic response with CsA alone is low.8,9 In these latter cases, complications including infections,10 bleeding and anemia may occur and lead to significantly poorer quality of life, recurrent transfusions, hospital admissions, secondary hemochromatosis, and death. It has recently been reported that eltrombopag, a nonpeptide thrombopoietin mimetic oral drug which binds to the transmembrane domain of the MPL receptor, can induce trilineage response in patients with refractory AA.11 In a single center phase 2 trial, hematologic improvement was obtained in 17 out of 43 patients who had previously failed to benefit from one or several courses of ATG.12 However, the effects of real-life use of this drug remain largely unknown, as the risks and benefits have not yet been independently assessed, while data for five patients achieving a robust response to eltrombopag may yet be updated.12 In France, physicians have access to eltrombopag through a compassionate use program. We used this program to assess the indications for and the safety and efficacy of eltrombopag in a large number of AA patients in France who received eltrombopag in the case of refractory AA, but also as a first-line treatment for patients considered unfit to receive ATG.

Methods Identifying cases This retrospective study (2012-2016) was conducted in 15 centers. A survey to identify patients receiving eltrombopag in France in the setting of AA was created on behalf of the French Reference Center for Aplastic Anemia. For the purposes of this study, we screened files for all patients referred to the center, and sent three waves of e-mails to more than 100 specialized physicians (Online Supplementary Material). The study was conducted in accordance with the Declaration of Helsinki. The institutional review board of the national AA center approved the study, and anonymous data collection was declared to the appropriate authorities. In accordance with French law, written informed consent was not required for this retrospective, non-interventional study, as patients had provided a non-opposition statement.

Population Patients with a diagnosis of AA confirmed by a bone marrow biopsy, irrespective of their age or the primary etiology, were eligible for inclusion if they received at least 2 months of treatment with eltrombopag, regardless of their indication for treatment. haematologica | 2018; 103(2)

Patients who received eltrombopag in a relapsed or refractory setting (defined by occurring at least 6 months after initial ATG treatment) were considered to be relapsed/refractory, irrespective of the number of previous ATG courses. Unfit patients who had not received at least one course of ATG before eltrombopag were considered to be receiving first-line treatment, even if they had previously been administered CsA as a stand-alone therapy. Patients with moderate AA requiring blood transfusion less frequently than every 4 weeks were not included. Furthermore, prior diagnoses of myelodysplastic syndrome, acute leukemia or immune thrombocytopenia were considered to be exclusion criteria. Patients who received eltrombopag in the setting of thrombocytopenia after stem cell transplantation were also excluded.

Procedures and definitions The minimal initial diagnostic panel for AA required for this study consisted of a complete blood count, a bone marrow biopsy, bone marrow karyotype analysis and assessment of a paroxysmal nocturnal hemoglobinuria (PNH) clone (Online Supplementary Methods) Given the retrospective nature of this study, dose adaptations for adverse events or insufficient responses were not formally defined. However, the recommendations of the reference center were shared with all centers once eltrombopag was approved for treatment in France (July 1, 2012). Centers were therefore advised to start treatment at the dose of 75 mg/day for 2 weeks and then increase the dose to 150 mg/day if no clinical or biological adverse events occurred.5,11 In the absence of any hematologic response or adverse event, some clinicians used higher doses, of up to 225 mg/day, after 3 months. When eltrombopag was used with rabbit ATG, the eltrombopag was started on day 15 after the ATG treatment. Hematologic improvements were assessed using the National Institutes of Health (NIH) response criteria. Thus, a platelet response was defined by a platelet count increase of 20x109/L above baseline, or stable platelet counts with transfusion independence for at least 8 weeks; an erythroid lineage response was defined by a hemoglobin increase of >1.5 g/dL or a reduction in >4 units of transfused red blood cells (RBC) for 8 consecutive weeks; and a leukocyte response was defined as an absolute neutrophil increase of 100% or an absolute neutrophil count increase >0.5x109/L.12 A robust response was defined as platelets >50x109/L, hemoglobin >10 g/dL, and neutrophils >1x109/L for longer than 8 weeks without transfusion support.12 Not achieving at least one NIH criterion during eltrombopag treatment was considered to be treatment failure.

Statistical analysis The data were recorded as percentages for discrete variables and medians with interquartile ranges (IQR) for continuous variables. A Fisher exact test and Wilcoxon non-parametric rank sum test were used, respectively, to compare these two types of variables. Follow-up and survival were reported since the start of eltrombopag treatment. Kaplan-Meier plots were constructed to estimate overall survival. All statistical tests were two-sided, with P values â&#x2030;¤0.05 indicating statistical significance. Statistical analyses were performed using R 2.14.0 (http://www.R-project.org/) packages.

Results Patients Forty-six patients were identified for this study. We separated the subjects into two different cohorts according to 213

E. Lengline et al.

the rationale for having initiated eltrombopag treatment and principal clinical situations. Eleven ATG-naĂŻve patients (cohort A) were given eltrombopag because they were considered ineligible for ATG treatment. For nine of these patients, the reason for ATG ineligibility was their age of over 65 years. One patient had dyskeratosis congenita with severe portal hypertension and hepatic dysfunction, while another, a 48-year old woman, had severe ischemic myocardial disease, complicated by a post-anoxia encephalopathy. Of these 11 patients, five had previously failed to benefit from treatment with CsA alone, four had been treated with androgens, while two further ATG-naĂŻve patients had received eltrombopag as first-line therapy. Thirty-five patients (cohort B) received eltrombopag in the setting of refractory or relapsed disease, defined by the persistence or reappearance of transfusion dependency or a neutrophil count <0.5x109/L at least 6 months after treatment with ATG. The first ATG treatment used was horse ATG in 66% of cases and rabbit ATG in the other 33%. Thirty percent of these patients had initially responded before relapsing, and 70% were primary refractory to this treatment. Not all patients were candidates for allogeneic stem cell transplantation at the time of eltrombopag initiation, either because of the lack of a suitable donor and/or age or comorbidities. The median age was 53 years old (IQR, 26-63) and 48% of patients had already received one cycle of ATG. Thirty-seven percent had previously received two cycles, including eight patients with primary refractory disease who were given eltrombopag in combination with the second course of ATG (eltrombopag was started between day 15 and day 45 after the rabbit ATG). The remaining 11% of patients in this cohort received three cycles of ATG prior to eltrombopag.

Table 1 shows the main characteristics of the patients divided into the two cohorts described above. At the time of AA diagnosis, a PNH clone was detected in 37% of all patients. One patient in each cohort was found to have germinal mutations consistent with dyskeratosis congenita, including a patient with no extra-hematologic phenotype who received one ATG course prior to genetic testing.

Eltrombopag treatment Cohort A. The median time from the diagnosis of AA to the start of eltrombopag treatment was 7 months (IQR, 333). The patients received eltrombopag for a median of 5 months (range, 3-20). The median eltrombopag dose prescribed was 150 mg once a day, and CsA was associated with eltrombopag in two patients (18%) from this cohort. Cohort B. The median time from the diagnosis of AA to starting eltrombopag treatment was 23 months (IQR, 950). Overall, these patients received eltrombopag for a median of 6 months (range, 2-39). Those who had a hematologic response received treatment for a median of 8 months (IQR, 5-18), compared with 5 months (IQR, 4-6) in cases of treatment failure. The median eltrombopag dose prescribed was 150 mg once a day, and CsA was associated with eltrombopag in 20 patients (57%) from this cohort.

Hematologic evolution and transfusion dependency Before starting eltrombopag treatment, all patients in both cohorts were transfusion-dependent. In cohort A (n=11), all patients were RBC transfusion-dependent, requiring a median number of three packed RBC units (IQR, 2-4) per month, and ten of the 11 patients were dependent on platelet transfusions, being given a median

Table 1. Main baseline characteristics of the patients divided by cohort (A and B).

Cohort A Number Demographic characteristics Age at diagnosis, years [IQR] Age at eltrombopag initiation, years [IQR] Male, n. (%) Aplastic anemia characteristics Idiopathic, no PHN clone, n. (%) Idiopathic, with PHN clone, n. (%) Dyskeratosis congenita, n. (%) Biological characteristics Elevated transaminases, n. (%) Hemoglobin, g/dL [IQR] Reticulocytes, absolute count, x109/L [IQR] Mean corpuscular volume, fL [IQR] Neutrophils, absolute count, x109/L [IQR] Lymphocytes, absolute count, x109/L [IQR] Platelets, x109/L [IQR] Bone marrow blast cells before eltrombopag, % [IQR] Eltrombopag treatment Time from diagnosis to eltrombopag, months [IQR] Eltrombopag treatment duration, months [IQR] Median dose, mg/day [IQR] Association with CsA, n. (%)

P-value

Cohort B 35

73.7 [60.9, 77.5] 74.1 [67.4, 78.0] 4 (36.4)

53.4 [26.3, 67.3] 55.3 [35.9, 68.5] 21 (60.0)

0.003 0.003 0.298 0.152

4 (36.4) 6 (54.5) 1 (9.1)

23 (65.7) 11 (31.4) 1 (2.9)

0 (0.0) 8.0 [6.4, 8.4] 31.5 [27.0, 36.2] 101 [89, 105] 960 [700, 1205] 1100 [810, 1320] 11 [8, 15] 0.0 [0.0, 0.50]

4 (12.9) 7.2 [4.9, 9.0] 25.0 [14.0, 58.0] 105 [98, 109] 630 [362, 969] 1615 [1267, 1722] 11 [6, 24] 0.0 [0.0, 1.0]

0.558 0.719 0.873 0.478 0.082 0.100 0.820 0.682

6.8 [3.1, 33.2] 5.3 [3.6, 10.4] 150 [150, 175] 2 (18.2)

22.9 [9.2, 49.7] 6.1 [4.4, 11.5] 150 [100, 150] 20 (57.1)

0.089 0.403 0.262 0.038

IQR: interquartile range; PNH: paroxysmal nocturnal hemoglobinuria; CsA: cyclosporine.

214

haematologica | 2018; 103(2)

Eltrombopag in aplastic anemia

number of two units (IQR, 1.5-3.5) of platelet concentrates (0.5x1011x kg) per month. In cohort B, 34/35 patients were RBC transfusion-dependent, requiring a median number of four packed RBC units (IQR, 2-4) per month, and 33/35 patients were dependent on platelet transfusions, being given a median number of three units (IQR, 2-4) of platelet concentrates (0.5x1011x kg) per month. Figure 1 shows the kinetics and proportions of patients achieving transfusion independence (RBC and platelets) in both cohorts. At last follow up, we confirmed transfusion independence (both RBC and platelets) in 36% and 49% of cohort A and B patients, respectively. It is worth noting that patients treated with eltrombopag first-line seemed to respond more slowly, with no responders during the first 3 months of treatment, compared with 44% in cohort B (P=0.02). In patients achieving transfusion independence, the increased hemoglobin level was 5 g/dL (IQR, 3.3-6) and 2.75 g/dL (IQR, 1.15-4.03) in cohorts A and B, respectively (P=0.3). Neutrophil and platelet counts also improved significantly in both cohorts (Table 2). Hematologic improvements were also assessed for each lineage using NIH criteria (Figure 2). According to these criteria, a response was observed in at least one lineage in 64% and 74% of cohorts A and B, respectively, including trilineage responses in 27% and 34% of cohorts A and B, respectively. Furthermore, a robust hematologic response12 was observed in three patients from cohort A and seven patients from cohort B.

Among the patients with refractory AA, the median eltrombopag dose in responders (i.e., patients who had a response in at least one lineage according to NIH criteria) was 150 mg/day (IQR, 100-150), which was not different from that in non-responders. In cohort A patients, the median eltrombopag dose in responders was 150 mg/day (IQR, 131-162), which was not different from that in nonresponders. Hematologic responses were observed in two out of four and three out of three patients whose daily eltrombopag dose was increased above 150 mg, to a maximum of 300 mg/day, in refractory patients and patients treated with eltrombopag first-line, respectively. At last follow-up, 22 patients (49%) were still receiving eltrombopag treatment, four (9%) eventually stopped after gradual tapering due to a robust hematologic response, one (2%) due to limited toxicity, and 18 (40%) due to eltrombopag failure. There were no significant differences in these proportions between the two cohorts. The four patients who were weaned off eltrombopag all remained in hematologic response 27, 24, 12 and 7 months after eltrombopag withdrawal. We also conducted a separate analysis on the eight patients with primary refractory AA who received eltrombopag in combination with a second course of ATG and CsA. These patients were 49.2 years old (IQR, 33.8-55.5) and received eltrombopag at a median of 21 days after rabbit ATG therapy (having all previously received horse ATG). Remarkably, all eight patients achieved transfusion

Figure 1. Rate (percentage) of transfusion independence before and after eltrombopag treatment by cohort and blood product type.

haematologica | 2018; 103(2)

215

E. Lengline et al.

Table 2. Trilineage hematologic improvement after eltrombopag therapy. Number Hemoglobin, g/dL [IQR] Before eltrombopag Month 1 eltrombopag Month 3 eltrombopag Month 6 eltrombopag Last follow-up receiving eltrombopag Last follow-up Median variation Neutrophils, 109/L [IQR] Before eltrombopag Month 1 eltrombopag Month 3 eltrombopag Month 6 eltrombopag Last follow-up receiving eltrombopag Last follow-up Median variation Platelets, 109/L [IQR] Before eltrombopag Month 1 eltrombopag Month 3 eltrombopag Month 6 eltrombopag Last follow-up receiving eltrombopag Last follow-up Median variation

P-value

Cohort A

Cohort B

8.0 [7.3-8.3] 8.3 [6.8-9.7] 8.0 [7.8-9.6] 9.3 [7.9-11.0] 9.8 [8.0-11.3] 9.8 [8.2-11.5] +5.0 [3.3-6.0]

8.0 [7.0-9.0] 9.0 [8.2-9.4] 8.8 [8.0-10.0] 9.5 [8.0-11.0] 9.1 [8.0-11.0] 9.4 [8.0-12.0] +2.75 [1.1-4.3]

0.52 0.62 0.42 0.69 0.95 0.83 0.31

0.7 [0.4-0.9] 0.6 [0.6-0.7] 0.9 [0,5-1.4] 0.8 [0.5-1.6] 1.1 [0.6-2.8] 1.0 [0.6-2.9] +0.4 [0.1-1.3]

0.9 [0.5-1.3] 1.0 [0.5-1.8] 1.1 [0.6-1.8] 1.3 [0.7-1.9] 1.3 [0.8-2.2] 1.3 [0.8-2.2] +0.4 [-0.1-1]

0.58 0.33 0.62 0.5 0.91 0.98 0.46

12 [5-18] 10 [7-25] 14 [7-28] 17 [4-83] 18 [8-48] 18 [9-57] +52 [2-160]

13 [7-19] 16 [11-22] 20 [10-37] 21 [12-58] 20 [11-60] 27 [13-73] +42 [24-86]

0.43 0.57 0.28 0.55 0.53 0.47 0.75

IQR: interquartile range.

Figure 2. Hematologic responses for each cell lineage in accordance with National Institutes of Health criteria.

216

haematologica | 2018; 103(2)

Eltrombopag in aplastic anemia

independence (7/8 at 3 months), despite the fact that they had received a median of three packed RBC units/month and three platelet units/month before treatment. At 6 months, the median hemoglobin level was 11 g/dL, the median platelet count 85x109/L, and the median neutrophil count 1.6x109/L. In cohort B, four patients with a median age of 25 years (IQR, 19-37) underwent hematopoietic stem cell transplantation with an alternative donor (2 HLA mismatched donors and 2 cord blood transplants) at a median time of 8 months (IQR, 6-11) after eltrombopag failure. One patient eventually died of Epstein-Barr virus-associated post-transplant lymphoproliferative disease 4 months after cord blood transplantation. The other patients were alive at last follow-up, more than 2 years after transplantation. No patient in the first-line cohort was considered for hematopoietic stem cell transplantation. Of particular note, neither of the two patients with dyskeratosis congenita experienced a hematologic response. Finally, we failed to identify any baseline factor associated with the occurrence of transfusion independence or hematologic improvement in this cohort.

Safety analysis and clonal evolution We retrospectively recorded potential toxicities of treatment among patients who received eltrombopag. The median follow-up after starting eltrombopag was 9 and 13 months for cohorts A and B, respectively. Table 3 reports the adverse events recorded in both cohorts. Most of the events were related to bone marrow dysfunction, with infections (mainly febrile neutropenia) and hemorrhages reported in a total of 13 and six patients, respectively. Thirteen patients developed elevated transaminase levels (grade 1, n=9; grade 2, n=2; grade 3, n=2) between 1.5 and 8 times the upper limit of normal without hepatic dysfunction, and one patient developed grade 2 hyperbilirubinemia. One patient had grade 2 insomnia, and one patient developed a localized lung cancer that required surgery. No thrombotic events or thrombocytosis (maximum platelet count: 250x109/L) were observed in either cohort. Among patients who had their PNH clone size evaluated after eltrombopag treatment, we found a nonstatistically significant increase in size in 45%, which is consistent with the natural history of refractory AA. There was no statistically significant difference between responders and non-responders (43% versus 50%; P=NS). The median follow-up was 9 months (IQR, 5-15) in

cohort A and 13 months (IQR, 7.5-26) in cohort B. We recorded six deaths, all of which occurred in non-responding patients. Deaths were caused by cerebral hemorrhage in two thrombocytopenic patients and acute myeloid leukemia in one patient: another patient had a sudden death, probably caused by a pulmonary embolism occurring 6 months after eltrombopag had been discontinued, one patient died of septic shock, and one died following a cord blood transplant. While 42 (91%) of patients had a bone marrow karyotype analysis at a median of 95 days before eltrombopag was introduced, only 12 patients (26% of the whole study population) had a subsequent karyotype analysis after eltrombopag had been started, to evaluate the risk of clonal evolution, despite an overall exposure to eltrombopag of 428 patient-months. The post-treatment karyotype analysis was conducted at a median of 14 months (IQR, 222) after eltrombopag had been started. Trisomy 8 was identified in one patient during eltrombopag treatment, with no myelodysplastic marrow morphology, but this abnormality had already been present at the time of diagnosis of the AA. The two patients with monosomy 7 before starting eltrombopag treatment did not undergo new karyotype analysis after treatment initiation (1 had a complete response with no evolution after 2 years of follow-up, while the other developed acute myeloid leukemia and eventually died). Online Supplementary Table S1 provides details of all karyotype analyses performed on the entire study population.

Discussion Recently, eltrombopag has been reported to induce clinically significant increases in blood counts and/or decreases in transfusion requirements in 40% of patients with refractory AA, and some of these patients achieved multilineage responses.11,12 However, to date, little use has been made of this drug in refractory AA patients, with available prospective data for 43 patients from the NIH12 and data from another single center study that included ten patients from China.13 On behalf of the French Reference Center for Aplastic Anemia, we retrospectively collected data on all patients who received eltrombopag between 2012 and 2016. Forty-six patients were identified who had been given eltrombopag for relapsed/refractory AA after ATG treatment (n=35) or as first-line therapy (n=11). We confirmed the overall efficacy of eltrombopag which pro-

Table 3. Adverse events and outcomes by treatment cohort during the follow-up observation time.

Factor Number Infection, % Hemorrhage > grade I, n. (%) Elevated transaminase during eltrombopag, n. (%) Deceased at last follow up, n. (%) PNH clone size before eltrombopag, % [IQR] PNH clone size after eltrombopag, % [IQR] Bone marrow blast cells after eltrombopag, % [IQR] Follow up, months [IQR]

Cohort A

Cohort B

P-value

11 3 (27.3) 2 (18.2) 2 (18.2) 2 (18.2) 2.00 [0.50, 3.50] 9.00 [5.50, 9.00] 1.00 [1.00, 1.00] 9.18 [5.36, 15.34]

35 10 (28.6) 4 (11.8) 10 (28.6) 3 (8.6) 0.00 [0.00, 1.50] 1.00 [0.00, 8.50] 0.00 [0.00, 1.00] 13.25 [7.51, 26.36]

1.000 0.624 0.775 0.580 0.035 0.178 0.309 0.212

IQR: interquartile range; PNH: paroxysmal nocturnal hemoglobinuria.

haematologica | 2018; 103(2)

217

E. Lengline et al. Table 4. Summary of bone marrow karyotype analysis before and after eltrombopag treatment. Karyotype analysis performed, n. (%) Time from eltrombopag treatment [IQR] Normal Failure -Y -7 +8

AA diagnosis

Before starting eltrombopag

After starting eltrombopag

37 (80)

42 (91) 95 days before [40-322] 29 3 1 2 0

12 (26) 14 months after [2-22] 10 1 0 0 1

26 6 2 0 1

IQR: interquartile range; AA: aplastic anemia.

duced a 46% rate of red blood cell and platelet transfusion independence. Hematologic improvement in more than one lineage, according to NIH criteria, was observed in 74% of relapsed/refractory patients but also in 64% of ATG-naĂŻve patients, while trilineage improvement was observed in 27% and 34%, respectively. The standard treatment for patients who do not have an HLAâ&#x20AC;&#x201C;identical sibling donor is the association of horse ATG plus CsA, although 30% show primary refractoriness to this treatment,4,7,14 and 30% of initial responders relapse after this first therapy. In this situation, transplantation from a well-matched unrelated donor may be considered in younger patients (under 30 years of age) in the first year after the diagnosis of AA,15 while results with mismatched unrelated donors are currently not as favorable.4 A significant number of patients overall will therefore be considered to be refractory and possibly exposed to infectious complications and intensive transfusion support, mainly complicated by hemosiderosis and alloimmunization.1 Until recently, the standard second-line treatment for patients without a histocompatible donor was a second course of immunosuppression using rabbit, horse or anti-CD52 antibodies, which produces an overall response rate of 30%.7,16 This latter treatment requires extensive hospitalization and careful, close follow-up because of initial worsening of transfusion requirements and induced immune deficiency arising from T-cell depletion.6 Eltrombopag is an oral thrombopoietin mimetic that is easy to use in an outpatient context. In this study, we confirmed an overall hematologic response of 71%, with almost half of these responders achieving trilineage improvement. The only alternative, in this situation, would be androgens such as oxymetholone, which were used to treat AA before the emergence of ATG plus CsA, and are, in fact, still used to treat certain patients with constitutional AA in a relapsed/refractory setting. However, data available regarding idiopathic AA are very scarce, especially in relapsed/refractory patients.17 Furthermore, side effects are commonplace after prolonged exposure to androgens.8 In our experience, eltrombopag was very well tolerated, with mild hepatic dysfunction in only one patient who stopped taking the drug because of a rise in transaminase levels. On the whole, our experience has led us to prefer the use of eltrombopag in this situation. As already mentioned, 11 patients were given eltrombopag as first-line treatment alone (n=9) or in association with CsA (n=2). These patients were not eligible for conventional ATG plus CsA treatment because of their age (median 74 years, versus 53 years for patients with refrac218

tory) or comorbidities (notably kidney dysfunction, which precludes the use of CsA). In this subgroup, we observed a 40% rate of hematologic improvement and transfusion independence at 6 months, with no excess of mortality compared to that of younger patients. In this particular population, treatment options are limited most of the time and include growth factors, transfusion support and antibiotics. While requiring confirmation through prospective control trials, eltrombopag alone might be a reasonable option in these specific patients. Of note, in this group, responses seemed to be slower, as no response was observed during the first 3 months of treatment. Eltrombopag should not, therefore, be discontinued too early in this specific setting, and we propose a treatment algorithm based on the findings of this study in Figure 3. Regarding refractory patients, our results are in line with those published previously.11,12 Initial studies defined the optimal dose of eltrombopag as 150 mg/day for hematologic response at 3 to 4 months.11,12 In this compassionate use program, we recommended that French centers using eltrombopag in 35 relapsed/refractory patients took into account the published NIH studies.11,12 Following this treatment plan, 43% of patients responded at 3 months and 50% at 6 months, which illustrates that the optimal time to appreciate the efficacy of this treatment is about 6 months overall. Of note, only three out of 21 evaluated patients responded to the higher dose of eltrombopag of 225 mg/day, illustrating the minimal benefit of increasing dosage in Caucasian patients. Importantly, four patients fulfilled the criteria for robust response;12 in these patients eltrombopag was then tapered and discontinued after a median of 14 months. All four patients then maintained stable blood counts, with a median medication-free follow-up of 18 months. Of particular note, two patients were also included in our study because of genetically proven dyskeratosis congenita with no response. The mechanism by which eltrombopag acts in the setting of bone marrow failure remains largely unclear. Among the nine patients in our study who received eltrombopag alone as a first-line treatment because they were ineligible for standard treatment, five responded of whom three had a trilineage response. This suggests that abrogation of immune attack may not be necessary for a response to eltrombopag in patients with idiopathic AA, which is in line with reported observed responses to eltrombopag in patients with moderate AA not previously treated with immunosuppression.12 It is thus likely that eltrombopag acts directly to stimulate the proliferation of small numbers of residual stem-progenitor cells in patients haematologica | 2018; 103(2)

Eltrombopag in aplastic anemia

Figure 3. French guidelines for the use of eltrombopag in patients with aplastic anemia (2017). (a) All patients should be screened at diagnosis for (i) an inherited bone marrow failure regardless of their family history and clinical findings, (ii) clonal evolution. (b) Data for children and in the literature are insufficient at present to do anything more than generate hypotheses, and should not be applied in patients < 18 years old. (c) Patients with aplastic anemia are considered to be eligible for transplantation as a second-line treatment in case of refractory status after first-line immunosuppressive therapy, excellent health status and (i) if a matched sibling donor is available (for patients who have been offered immunosuppressive therapy first line because of age > 40 years) or (ii) if a matched unrelated donor is available for patients aged 30 years and under. Regarding the age limits, stated cutoff ages are recommendations and are therefore open to debate in accordance with institution and patient specificities. (d) For refractory patients, a careful reassessment of the diagnosis – to exclude a clonal evolution such as myelodysplastic syndrome or constitutional bone marrow failure – is mandatory. (d) In patients over 65-70 years old or patients with severe comorbidities (cardiac and/or renal failure), the use of ATG may be responsible for inadequate toxicity. (e) Rabbit antithymoglobulin: 3.75 mg/kg continuous intravenous administration over 12 h from day 1 to day 5, Cyclosporine A: 5 mg/kg/d from day 1 in order to achieve residual dosage of between 200 and 400 ng/mL. Start the treatment orally. Cyclosporine should not be withdrawn prematurely before 6 months unless toxicity grade >2 occurs. Eltrombopag 75 mg per day for 2 weeks, and thereafter increased to 150 mg per day from day 14 and as soon as transaminases < 2 times upper limit of normal (ULN) and bilirubin < 1,5 ULN for a minimum of 3 months. Eltrombopag should be interrupted in case of transaminases > 3N. (f) Eltrombopag should be initiated at 75 mg per day for 2 weeks in a fasting patient and thereafter increased to 150 mg per day if no clinical or biological toxicity is identified. The dosage should be halved for subjects of fully (both parents) East Asian heritage (i.e. Japanese, Chinese, Taiwanese and Korean) because plasma eltrombopag AUC(0-τ) concentrations have been found to be approximately 80% higher in healthy Japanese subjects than in nonJapanese healthy subjects (predominantly Caucasian). (g) The starting date for evaluation is the first day at 150 mg. In patients not responding at 150 mg per day, the dosage may be carefully increased, up to 225 mg per day. These patients should be monitored closely for adverse events (abdominal pain, diarrhea, cataract). (h) A robust response is considered for patients with Hemoglobin>10g/dL, neutrophils 1x109/L and platelets more than 50x109/L. HSCT: hematopoietic stem cell transplantation; ATG: anti-thymocyte globulins; hATG: horse anti-thymocyte globulins; rATG: rabbit anti-thymocyte globulins; CSA: cyclosporin A; ELT: eltrombopag; resp: responders; BM: bone marrow; BMF: bone marrow failure.

with AA. However, outstanding results published very recently regarding responses in treatment-naïve AA patients in a phase 2 trial investigating the association of ATG plus CsA plus eltrombopag18 highlighted the benefit of decreasing the intensity of the immune attack to improve responses to eltrombopag. Two ongoing prospective phase 3 trials have randomized the addition of eltrombopag to standard therapy (Figure 3). The RACE study is comparing horse ATG plus CsA with or without eltrombopag as front-line therapy for patients with severe AA (EudraCT: 2014-000363-40), while the EMAA study is comparing the use of CsA with or without eltrombopag as front-line therapy for patients with moderate AA (EudraCT: 2014-000147-19). We did not identify any predictors of response in our study. However, higher reticulocyte counts in refractory patients,12 as well as longer telomere length and younger age in treatment-naïve patients18 have been reported to be associated with better responses, although these fachaematologica | 2018; 103(2)

tors are also a reflection of the stem cell pool. Collectively, this suggests that a critical mass of stem cells is required for bone marrow recovery, and that this is better if the immune response has been abated. While the exact role of eltrombopag in AA is still being investigated, enough evidence exists to suggest that it directly stimulates hematopoietic stem and progenitor cells, which could theoretically affect the emergence of abnormal clones, as has already been retrospectively reported with the use of hematopoietic growth factor in AA.19,20 In two prospective studies examining both refractory12 and treatment-naïve patients,18 no evidence of a higher rate of clonal evolution was identified after comparison with historical controls treated in the same institution. Unfortunately, patients in our study were not examined serially, so any potential clonal evolution of karyotypic aberrations could not be extensively and systematically defined. Regarding the potential risk of clonal evolution associated with the use of eltrombopag alone, random219

E. Lengline et al.

ized studies with careful serial evaluation of clonal evolution, including karyotype, fluorescence in situ hybridization analysis and molecular studies, will enable the risks associated with eltrombopag in this setting to be assessed more accurately. We detected a non-significant increase in the size of the PNH clone after eltrombopag treatment, especially in patients who had already received immunosuppressive therapy. There are no data in the literature suggestimg a similar finding with eltrombopag; indeed, no firm conclusions can be drawn from this observation. Our work has both strengths and limitations. The latter are mostly due to its retrospective nature and limited numbers of patients. The strengths include the systematic enrollment of all AA patients who received eltrombopag in France between 2012 and 2016 through the French Reference Center for Aplastic Anemia. This provides a clear picture of our current practice for refractory patients but also on first-line monotherapy for patients who are not eligible for standard immunosuppression. In conclusion, eltrombopag has shown convincing efficacy in the majority of refractory AA patients both in this

References 9. 1. Marsh JCW, Kulasekararaj AG. Management of the refractory aplastic anemia patient: what are the options? Blood. 2013;122(22):3561–3567. 2. Bacigalupo A, Hows J, Gluckman E, et al. Bone marrow transplantation (BMT) versus immunosuppression for the treatment of severe aplastic anaemia (SAA): a report of the EBMT* SAA Working Party. Br J Haematol. 1988;70(2):177–182. 3. Yoshida N, Kobayashi R, Yabe H, et al. First-line treatment for severe aplastic anemia in children: bone marrow transplantation from a matched family donor versus immunosuppressive therapy. Haematologica. 2014;99(12):1784–1791. 4. Peffault de Latour R. Transplantation for bone marrow failure: current issues. Hematology Am Soc Hematol Educ Program. 2016;2016(1):90–98. 5. Scheinberg P, Nunez O, Weinstein B, et al. Horse versus rabbit antithymocyte globulin in acquired aplastic anemia. N Engl J Med. 2011;365(5):430–438. 6. Young NS. Pathophysiologic mechanisms in acquired aplastic anemia. Hematology Am Soc Hematol Educ Program. 2006;72–77. 7. Scheinberg P. Aplastic anemia: therapeutic updates in immunosuppression and transplantation. ASH Education Program Book 2012;2012(1):292–300. 8. Scheinberg P, Young NS. How I treat

220

10.

11.

12.

13.

14.

analysis and in previously published studies.11,12 We also confirmed that eltrombopag was able to restore trilineage hematopoiesis in half of the responders. Responses were identified up to 6 months after treatment with an overall acceptable toxicity profile. This report, based on real-life clinical practice, also provides three novel findings that require further investigation: (i) eltrombopag monotherapy may benefit older patients considered to be unfit for ATG; (ii) in patients with a first relapse or who are refractory after one cycle of ATG, high response rates may be achieved with eltrombopag when combined with a second course of ATG plus CsA treatment; and (iii) the optimal dose of eltrombopag merits further investigation, as it would appear that some patients might respond to a dose higher than 150 mg/day. In brief, the encouraging overall results now need to be confirmed through prospective controlled trials. Acknowledgments The authors would like to thank Dr Antonio Risitano for constructive intellectual input and advice and Mr David Williams for editorial assistance.

acquired aplastic anemia. Blood. 2012;120 (6):1185–1196. Marsh J, Schrezenmeier H, Marin P, et al. Prospective randomized multicenter study comparing cyclosporin alone versus the combination of antithymocyte globulin and cyclosporin for treatment of patients with nonsevere aplastic anemia: a report from the European Blood and Marrow Transplant (EBMT) Severe Aplastic Anaemia Working Party. Blood. 1999;93(7):2191–2195. Torres HA, Bodey GP, Rolston KVI, Kantarjian HM, Raad II, Kontoyiannis DP. Infections in patients with aplastic anemia: experience at a tertiary care cancer center. Cancer. 2003;98(1):86–93. Olnes MJ, Scheinberg P, Calvo KR, et al. Eltrombopag and improved hematopoiesis in refractory aplastic anemia. N Engl J Med. 2012;367(1):11–19. Desmond R, Townsley DM, Dumitriu B, et al. Eltrombopag restores trilineage hematopoiesis in refractory severe aplastic anemia that can be sustained on discontinuation of drug. Blood. 2014;123(12):1818– 1825. Gill H, Leung GMK, Lopes D, Kwong Y-L. The thrombopoietin mimetics eltrombopag and romiplostim in the treatment of refractory aplastic anaemia. Br J Haematol. 2017;176(6):991–994. Young NS, Calado RT, Scheinberg P. Current concepts in the pathophysiology and treatment of aplastic anemia. Blood.

2006;108(8):2509–2519. 15. Devillier R, Dalle J-H, Kulasekararaj A, et al. Unrelated alternative donor transplantation for severe acquired aplastic anemia: a study from the French Society of Bone Marrow Transplantation and Cell Therapies and the EBMT Severe Aplastic Anemia Working Party. Haematologica. 2016;101(7):884– 890. 16. Scheinberg P, Nunez O, Weinstein B, Scheinberg P, Wu CO, Young NS. Activity of alemtuzumab monotherapy in treatment-naive, relapsed, and refractory severe acquired aplastic anemia. Blood. 2012;119 (2):345–354. 17. Chuhjo T, Yamazaki H, Omine M, Nakao S. Danazol therapy for aplastic anemia refractory to immunosuppressive therapy. Am J Hematol. 2008;83(5):387–389. 18. Townsley DM, Scheinberg P, Winkler T, et al. Eltrombopag added to standard immunosuppression for aplastic anemia. N Engl J Med. 2017;376(16):1540–1550. 19. Kojima S, Ohara A, Tsuchida M, et al. Risk factors for evolution of acquired aplastic anemia into myelodysplastic syndrome and acute myeloid leukemia after immunosuppressive therapy in children. Blood. 2002;100(3):786–790. 20. Socie G, Mary J-Y, Schrezenmeier H, et al. Granulocyte-stimulating factor and severe aplastic anemia: a survey by the European Group for Blood and Marrow Transplantation (EBMT). Blood. 2007;109 (7):2794–2796.

haematologica | 2018; 103(2)

ARTICLE

Bone Marrow Failure

Rational management approach to pure red cell aplasia

Ferrata Storti Foundation

Suresh Kumar Balasubramanian,1 Meena Sadaps,2 Swapna Thota,3 Mai Aly,1 Bartlomiej P. Przychodzen,1 Cassandra M. Hirsch,1 Valeria Visconte,1 Tomas Radivoyevitch1 and Jaroslaw P. Maciejewski1* Department of Translational Hematology and Oncology Research, 2Department of Internal Medicine and 3Department of Hematology and Medical Oncology, Taussig Cancer Institute, Cleveland Clinic, OH, USA

Haematologica 2018 Volume 103(2):221-230

ABSTRACT

ure red cell aplasia is an orphan disease, and as such lacks rationally established standard therapies. Most cases are idiopathic; a subset is antibody-mediated. There is overlap between idiopathic cases and those with T-cell large granular lymphocytic leukemia, hypogammaglobulinemia, and low-grade lymphomas. In each of the aforementioned, the pathogenetic mechanisms may involve autoreactive cytotoxic responses. We selected 62 uniformly diagnosed pure red cell aplasia patients and analyzed their pathophysiologic features and responsiveness to rationally applied first-line and salvage therapies in order to propose diagnostic and therapeutic algorithms that may be helpful in guiding the management of prospective patients, 52% of whom were idiopathic, while the others involved large granular lymphocytic leukemia, thymoma, and B-cell dyscrasia. T-cell-mediated responses ranged between a continuum from polyclonal to monoclonal (as seen in large granular lymphocytic leukemia). During a median observation period of 40 months, patients received a median of two different therapies to achieve remission. Frequently used therapy included calcineurininhibitors with a steroid taper yielding a first-line overall response rate of 76% (53/70). Oral cyclophosphamide showed activity, albeit lower than that produced by cyclosporine. Intravenous immunoglobulins were effective both in parvovirus patients and in hypogammaglobulinemia cases. In salvage settings, alemtuzumab is active, particularly in large granular lymphocytic leukemia-associated cases. Other potentially useful salvage options include rituximab, anti-thymocyte globulin and bortezomib. The workup of acquired pure red cell aplasia should include investigations of common pathological associations. Most effective therapies are directed against T-cell-mediated immunity, and therapeutic choices need to account for associated conditions that may help in choosing alternative salvage agents, such as intravenous immunoglobulin, alemtuzumab and bortezomib.

Introduction Pure red cell aplasia (PRCA) can be inherited (Diamond Blackfan Anemia, [DBA]) or acquired (aPRCA). The latter is further subclassified as B19 parvovirus-associated (transient) aplastic crisis (TAC),1 drug-associated cases2 (e.g., allopurinol, azathioprine, diphenylhydantoin, rifampicin, valproic acid etc.), primary idiopathic aPRCA with various (immune) etiologies, and secondary aPRCA associated with other conditions, including B-cell dyscrasias (chronic lymphocytic leukemia [CLL]),3 WaldenstrĂśm macroglobulinemia,4 monoclonal gammopathy of undetermined significance (MGUS)5 and multiple myeloma,6 T-cell lymphoproliferative disorders (large granular lymphocytic (LGL) leukemia)7,8 or solid organ malignancies (most commonly thymoma),9 and collagen vascular/autoimmune processes with aPRCA overlap.10-13 Myelodysplastic syndromes (MDS) can mimic PRCA morphologically.14 aPRCA may also be a forerunner of acquired aplastic anemia (AA). The hallmark of PRCA is reticulocytopenic, malproductive anemia (absolute reticulocyte count < 10,000/ml [reticulocyte percentage, <1%]),15 which is diagnosed after the exclusion of obvious causes of anemia. The absence or profound haematologica | 2018; 103(2)

Correspondence: maciejj@ccf.org

Received: July 3, 2017. Accepted: December 6, 2017. Pre-published: December 7, 2017. doi:10.3324/haematol.2017.175810 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/221 ÂŠ2018 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.

221

S.K. Balasubramanian et al. Table 1. Patient characteristics (including immunological/molecular characteristics) in different acquired immune-mediated PRCA groups.

Characteristics Sex (male) Age (median years, range) Hb g/dl (mean±SD) EPO mIU/ml (mean±SD) Reticulocyte % (mean±SD) M:E ratio (mean±SD) MGUS+ (n) Hypogammaglobulinemia (n) TCR (+/tested) CD4/CD8 (median) CD16/56 (median) vβ expansion (+/tested) STAT3 (+/tested)

LGL/PRCA (14)

Idiopathic PRCA with thymoma (9)

True idiopathic PRCA without thymoma (32)

9 (64%) 68 (32-82) 6.8±2.1 1318±1062 0.6±0.5 13±10 3/14 3/14 14/14 0.2 1.5 13/14 5/11

4 (44%) 54 (25-89) 6.8±2.5 3393±1959 0.3±0.5 9±2 none 2/9 2/6 0.6 18.0 1/9 0/5

18 (56%) 60 (28-87) 7.9±2.3 1368±1199 0.4±0.3 13±9 4/32 6/32 6/30 1.5 7.0 5/32 0/12

EPO: erythropoietin; Hb: hemoglobin; LGL/PRCA: large granular lymphocytic leukemia related/pure red cell aplasia; M:E: myeloid-erythroid ratio; MGUS: monoclonal gammopathy of undetermined significance; TCR: T-cell rearrangement.

depletion of erythroid precursors is sine qua non for PRCA. Secondary PRCA is often dominated by the underlying disease. B19-associated PRCA is due to the lytic activity of the virus on pronormoblasts,16 which is seen as large proerythroblasts with vacuolated cytoplasm and pseudopodia (“giant pronormoblasts”). After the exclusion of viral etiologies, congenital diseases and drug reactions, idiopathic PRCA would be the most common cause, with the majority of cases thought to be mediated by autoreactive T-cells. This is likely via selective T- or natural killer (NK)-cell-mediated killing of erythroid colony (CFU-E) and burst (BFU-E) forming units, thereby inhibiting red cell precursor progression to mature erythrocytes.7,17-20 By analogy with acquired neutropenia, “idiopathic” PRCA is typically T-cell-mediated. In contrast, “autoimmune PRCA” mediated by antibodies is less common,10,13 and most anti-erythroid antibodies would typically result in immune hemolytic anemia. Anti-erythropoietin (EPO) antibody-mediated PRCA induced by recombinant EPO can be considered as a specific form of autoimmune-PRCA.21,22 Thymoma-associated PRCA may be either considered as secondary or primary with immune etiology. While some patients respond to T-celldirected immunosuppression, other patients have underlying diseases that involve humoral immunity (e.g., relation to myasthenia gravis),9 and those are cases which are less responsive to such therapies. A similar association is suggested by the occurrence of PRCA in the context of Good’s syndrome (thymoma, combined variable immunodeficiency and PRCA).23 Although PRCA is rare, it is very diverse. This diversity seems to be related to overlapping pathologic and clinical associations for which systematic studies of these correlations have not been performed. Clearly, the rarity of the disease precludes the accumulation of clinical experience based on clinical trial evidence. Instead, current practices are mostly driven by retrospective analyses of case reports and series. Using data from a large number of PRCA patients seen in our center, we analyzed clinical presentations, pathophysiologic features, and clinical responses to different therapeutic options. 222

Methods Patient populations Based on uniformly defined criteria (Online Supplementary Table S1), 62 cases of PRCA, confirmed by bone marrow biopsy and treated at the Cleveland Clinic between the years 2000 and 2016, were included in the retrospective analysis (Online Supplementary Table S2 and Online Supplementary Figure S1). Informed consent was obtained per protocols approved by the Institutional Review Board in accordance with the Declaration of Helsinki. Clinical parameters of the patients, including demographics, blood counts, treatment specifics, and survival times, were obtained from medical records. The primary end-point was a hematological response in terms of an appropriate rise in reticulocytes/hemoglobin and becoming transfusion-independent, as determined by two separate measurements after the first eight weeks of initiating the treatment. This is the earliest time for assessment of response. By contrast, borrowed from more systematic studies in aplastic anemia and systemic immunosuppression, response was again assessed at the 3-month milestone, and in less common circumstances monitored for six months until defining it as a non-responder to a particular treatment. Normal blood counts were determined using hospital standards.

Response criteria Complete response (CR) was defined by an appropriate rise in the reticulocyte count, commensurate to the level of hemoglobin rise, in addition to becoming transfusion-independent followed by subsequent normalization of hemoglobin levels after eight weeks of initiating the treatment (Online Supplementary Table S3). Partial response (PR) was defined when there was no appropriate rise in reticulocyte count and the patient was still anemic, but their transfusion requirement became less frequent than it was prior to initiating the drug. No response (NR) was defined when none of the above criteria were met at the end of eight weeks. Complete response rate (CRR) was calculated as the percentage of patients with CR over the total number of patients who received the drug. Overall response rate (ORR) was calculated as the percentage of patients with both CR and PR over the total number of patients who received the drug. We have distinguished primary and a coassociated PRCA cases. In those secondary cases with T-cell large haematologica | 2018; 103(2)

Biology and management of PRCA

granular lymphocytic (T-LGL) leukemia, the corresponding LGL leukemia treatment would not differ from that of “idiopathic” PRCA, and response in LGL would be determined according to the improvement of anemia. However, in cases associated with B-cell dyscrasia, B-cell modalities would include rituximab or bortezomib. In the latter cases, the response was monitored by monoclonal protein levels, but due to the low number of cases, a correlation between the level of decrease in M-protein and the overall response was not statistically established.

DNA Targeted Sequencing Targeted sequencing was completed using Nextera Custom Enrichment library (Illumina). The custom targeted panel consisted of capture probes for genes known to be relevant in MDS, acute myeloid leukemia (AML) and other hematologic malignancies (both somatic and germline targets, N=169). The list of genes targeted can be found in Online Supplementary Table S4. The enriched targets were subjected to massive sequencing using Illumina MiSeq sequencer, with an average target depth of 260x.

CTL: cytotoxic lymbhocytes; E:M: erythroid-myeloid ratio; LGL: large granular lymphocytes; PRCA: pure red cell aplasia; aPRCA: acquired PRCA.

haematologica | 2018; 103(2)

Figure 1. Pathogenesis of pure red cell aplasia (PRCA) in our cohort. (A) Different causes of PRCA affect red cell production in different ways. Parvovirus causes direct cytolysis of erythroid precursor cells. Several drugs are implicated in causing PRCA, probably through a direct toxic effect. EPO antibodies can recognize antigens at the erythroid burst forming unit (BFU-E) level and cause PRCA. There is a wide range of antigen recognition by CTL exhibiting a polyclonal response, seen as negative TCR (T-cell gene rearrangement) by testing, which, however, still causes PRCA, probably by direct cytotoxicity. At the other end of the spectrum are LGLs that are monoclonal with STAT3 mutations and can recognize antigens at the erythroid precursor level, and thus cause PRCA. (B) Flow cytometry Vβ analysis of the peripheral blood of individual representative patients from the cohort shows clonal skewing. Illustration shows a spectrum of polyclonal to oligoclonal to monoclonal response of CTL in individual patients with PRCA.

223

S.K. Balasubramanian et al.

The generation of BAM (.bam) files (alignment with BurrowsWheeler Aligner [BWA] 0.7.9a) with its preprocessing and detection of somatic point mutations / insertions and deletions was carried out in accordance with the Genome Analysis Toolkit (GATK) best practices. Only variants with high-quality reads and a minimum depth of at least 20 and six positives were considered for further analysis. Sequencing results were annotated using ANNOVAR.24 Variants were considered somatic if they were not reported within the Exome Aggregation Consortium (ExAC) germline database (ExAC, Cambridge, MA, USA). Novel

splicing, stop-gain and insertions-deletions variants were prioritized and studied further.

Statistical analysis The R package forest plot was used to plot odds ratios (OR) computed using Fisherâ&#x20AC;&#x2122;s exact test and response rates, the 95% confidence intervals of which were computed using logistic regression, i.e., the R function generalized linear model (glm) was used with y ~ 1 as the model and the statistical distribution family set to binomial.

CsA: cyclosporine; cs: corticosteroids; CP: cyclophosphamide; M: maintenance treatment; MTX: methotrexate; MGUS: monoclonal gammopathy of undetermined significance; RA: rheumatoid arthritis; ATG: anti-thymocyte globulin. Figure 2. Treatment algorithm for immune-mediated PRCA in our patient cohort. Cyclosporine or cyclophosphamide with a steroid taper is the first-line choice of treatment in both idiopathic and LGL- related PRCA. Maintenance treatment with immunosuppressive therapy is usually needed, and varies based on the sustainability of the response obtained (see text for details). Methotrexate is used in a salvage setting in LGL/PRCA, but has no role in idiopathic PRCA. Alemtuzumab is one of the commonly used salvage options in refractory PRCA.

224

haematologica | 2018; 103(2)

Biology and management of PRCA

Results Clinical features of PRCA From January 2000 to December 2016, we saw and treated 62 patients with PRCA who fulfilled the diagnostic criteria of one of the clinical subentities. The clinical clas-

sification of PRCA patients (n=62) in our cohort is demonstrated in Online Supplementary Figure S1 and Online Supplementary Table S2. The pathogenesis of PRCA seen in our cohort of patients is illustrated in Figure 1. Excluding the DBA patients (Online Supplementary Table S5), the median age at presentation was 62 years (25-87 y) with a

Table 2. Treatment response rates in all aPRCA patients.

Treatment #

Cyclosporine/tacrolimus CP MTX IVIG@ Rituximab$ Alemtuzumab (Campath) ATG Others*

First-line CRR

Overall CRR

ORR

40% (10/25) 15% (1/7) 0/4

49% (34/70) 27% (8/30) 20% (2/10)

76% (53/70) 45% (14/31) 30% (3/10)

NA NA NA NA

30% (4/13) 38% (5/13) 10% (1/10) 40% (6/15)

62% (8/13) 54% (7/13) 30% (3/10) 60% (9/15)

#Tacrolimus used only in salvage setting when initial response to CsA was suboptimal or when intolerant to CsA in the first-line (7/11 patients responded in salvage treatment when response to CsA was suboptimal). @IVIG responses are elaborated in detail in Table 3. $Rituximab used only in salvage setting when multiple prior IST failed. *Others include danazol, mycophenolate mofetil (Cellcept), bortezomib (Velcade) Âą steroids, and erythropoietin (Procrit). ATG: anti-thymocyte globulin; CP: cyclophosphamide (Cytoxan, CTX); CRR: complete response rate (CR/number treated); IVIG: intravenous immunoglobulin; ORR: overall response rate (CR+PR/number treated); MTX: methotrexate.

ATG: anti-thymocyte globulin; IVIG@: intravenous gammaglobulin (see also Table 3); other* includes danazol, mycophenolate mofetil, bortezomib, erythropoietin, abatacept (Orencia), tofacitinib (Xeljanz).

Figure 3. Response probabilities across various treatment options in aPRCA. (A) Cyclosporine and tacrolimus are the most effective drugs for PRCA in our cohort. The details of IVIG response rates are elaborated in Table 3. Rituximab is used only in the salvage setting when the front-line options have failed, including commonly used salvage treatments such as alemtuzumab. ATG is used purely as a salvage option, but only shows mediocre, but acceptable, response rates. The other salvage treatment options include danazol, mycophenolate mofetil, bortezomib, erythropoietin, abatacept (Orencia), and tofacitinib (Xeljanz) etc., all of which show good response rates when used in the right setting (see text for details). (B) Response proportion for different treatment options: comparison between LGL vs. non-LGL PRCA groups. Cyclosporine and tacrolimus show better RR in the non-LGL group, whereas campath (alemtuzumab), though not statistically significant, works better in LGL-PRCA. Patients with parvovirus PRCA, and those presenting with low immunoglobulin levels in PRCA show excellent RR with IVIG (see also Table 3).

haematologica | 2018; 103(2)

225

S.K. Balasubramanian et al. Table 3. Treatment response rates in LGL and non-LGL aPRCA patients.

Treatment #

Cyclosporine/tacrolimus CP MTX IVIG@ Rituximab$ Alemtuzumab ATG Others*

IVIG Overall RR

First-line CRR

LGL-associated PRCA Overall CRR

ORR

First line CRR

non-LGL PRCA Overall CRR

ORR

25% (2/8) (0/2) (0/4)

20% (4/20) 20% (3/15) 20% (2/10)

55% (11/20) 47% (7/15) 30% (3/10)

50% (9/18) 20% (1/5) NA

60% (30/50) 25% (4/16) NA

84% (42/50) 44% (7/16) NA

NA NA NA NA Parvovirus B19 PRCA 100 % (4/4)

(0/2) 63% (5/8) 25% (1/4) (0/7)

50% (1/2) 75% (6/8) 50% (2/4) 29% (2/7) PRCA with hypogammaglobulinemia 100% (11/11)

NA NA NA NA

36% (4/11) 64% (7/11) (0/5) 20% (1/5) (0/6) 17% (1/6) 75% (6/8) 88% (7/8) Idiopathic PRCA refractory to IST£ with LGL without LGL 66% (2/3) 13% (2/15)

Tacrolimus used only in salvage setting when initial response to CsA was suboptimal or when intolerant to CsA in the first-line (7/11 patients responded in salvage treatment when response to CsA was suboptimal). $Rituximab used only in salvage setting when multiple prior IST failed. Others include danazol, mycophenolate mofetil (Cellcept), bortezomib (Velcade) ±steroids, and erythropoietin (Procrit). £Normal levels of gammaglobulins. ATG: anti-thymocyte globulin; CP: cyclophosphamide (Cytoxan, CTX); CRR: complete response rate (CR/number treated); IVIG: intravenous immunoglobulin; MTX: methotrexate; ORR: overall response rate (CR+PR/number treated). #

slight male predominance (n=34, 55%; P=0.6). Among the aPRCA patients (Online Supplementary Table S1), most were idiopathic (n=32, 52%) or followed by LGL leukemia (n=14, 22%) or thymoma (n=9, 15%); less common were B19-related PRCA patients (n=4, 6%; P=0.00004). Intriguingly, one patient initially treated for LGL-associated PRCA was later found to be positive for parvovirus on repeated DNA polymerase chain reaction (PCR) testing. The clinical characteristics of the acquired immune-mediated PRCA groups are listed in Table 1. The median follow-up of aPRCA patients in our cohort was 40 months (range: 1-133 months). The total number of aPRCA patients who were in remission during the last follow up was 46 (including partial and complete). The median number of different therapeutic approaches used to achieve remission was two (range: 1-8). Maintenance treatment with immunosuppressive agents was usually continued for a year or two, at the least, and gradually tapered off if the disease was stable. The OS and diseasefree probability at 5/10 years in our cohort were 0.835 (95% confidence interval [CI] 0.695, 1) /0.674 (95% CI 0.472, 0.963) and 0.675 (95% CI 0.529, 0.861) /0.496 (95% CI 0.321, 0.764), respectively (Online Supplementary Table S2).

Immunological characterization of acquired immune-mediated PRCA patients The distributions of MGUS, hypogammaglobulinemia, presence of T-cell receptor (TCR) rearrangement, median CD4/CD8 ratio, median CD16/CD56 ratio, and Vβ expansion are listed in Table 1. There were eight cases in the idiopathic aPRCA group who had clonal TCR rearrangement but did not fulfill the criteria for LGL (Table 1 and Online Supplementary Table S6). The CD4/CD8 ratio was significantly lower among the LGL/PRCA group (0.2) compared to the idiopathic group with and without thymoma (0.9 and 1.5; P=0.0064) (Online Supplementary Figure S2). Some cases (n=6) of idiopathic aPRCA also demonstrated Vβ expansion. However, it appeared that there was a significant clinical overlap between LGL-associated and idiopathic cases of aPRCA. 226

Mutational status of acquired PRCA patients The STAT3 mutation in the LGL/PRCA patients were positive in 5/11 patients tested, whereas 17 patients tested negative in the idiopathic group. Mutational analysis using a targeted myeloid next-generation sequencing (NGS) panel (Nextera, Illumina; see Methods) was performed in a selected subgroup of patients during diagnostic workup of their anemia. No somatic clonal mutations typical of MDS were found in 12 patients tested with idiopathic, otherwise typical aPRCA. Alterations found in genes affected in MDS were present in five typical idiopathic aPRCA patients and were considered to be of germline origin (Online Supplementary Table S7). None of these patients fulfilled the diagnostic criteria for MDS.

Immunosuppressive therapy A rational treatment algorithm applied to our patients is illustrated in Figure 2. The choice of our first-line treatment differed based on the underlying cause of PRCA. The first-line CRR, overall CRR and ORR are illustrated in Figure 3 (see also Table 2 and Table 3). The most common first-line treatment used in idiopathic aPRCA was cyclosporine (CsA) combined with a steroid taper (prednisone [P]). This yielded an ORR of 76% (53/70) and first-line CRR and overall CRR of 40% (10/25) and 49% (34/70), respectively. Slight alterations in creatinine after initiation of CsA triggered periodic monitoring of kidney function. Further drug administration was either stopped or adjusted if creatinine trends worsened. ORR was better in non-LGL- vs. LGL-related PRCA (84% vs. 55%, P=0.01) (Figure 3B and Table 3). CsA was avoided as a first-line treatment, especially in patients with renal failure. Tacrolimus was used in salvage mode or in lieu of CsA when it was not tolerated by patients. The second most common first-line choice of treatment was cyclophosphamide (CP, [Cytoxan, (CTX)]) with a steroid taper. ORR was 47% (14/30), first-line CRR 15% (1/7) and overall CRR 27% (8/30), i.e., lower than with CsA (P=0.6 and P=0.2, respectively). In subgroup analysis, there was no discernible difference in overall RR among the LGL and non-LGL group (Figure 3B and Table 3). haematologica | 2018; 103(2)

Biology and management of PRCA

Table 4. Outcomes of PRCA reported from different studies.

Studies 39

Clark et al. [n=37]; 1984

Lacy et al.20 [n=47]; 1996

Kwong et al.32 [n=16]; 1996 Mamiya et al.33 [n=150]; 1997

Casadeval et al.22 [n=13]; 2002 Sawada et al.35 [n=73]; 2007 Hirokawa et al.31 [n=41]; 2008 Dâ&#x20AC;&#x2122;Arena et al.50 [n=4]; 2009 Risitano et al.43 [n=12]; 2010

Crabol et al.45 [n=10]; 2013 Wu et al.36 [n=34]; 2016

Korde et al.52 [n=51]; 2016

Present study [n=62] 4 congenital

Etiology

Drugs

Response rate (RR)

27 idiopathic 10 secondary

Cytotoxic drugs* + cs IST (CsA & CTX)

56% (18/32) remission with cytotoxic drugs* + cs 60% ORR (LGL/PRCA vs. idiopathic PRCA; 88% vs. 56%)

CsA & cs

23% (3/13) with cs, 66% (4/6) with CsA

CsA & cs

82% (31/38) with CsA & 54% (50/92) with cs

25 idiopathic 9 LGL leukemia 4 thymoma 4 CLL- and 2 NHL-related PRCA 9 idiopathic 7 secondary (3 LGL) 35 acute PRCA 51 idiopathic 64 secondary CKD and EPO antibodies Idiopathic Thymoma CLL 3 Thymoma (1 thymoma with MGUS), remaining idiopathic Parvovirus B19+ 15 idiopathic 5 thymoma 7 LGL 7 others# 12 MGUS 32 idiopathic 14 LGL 9 thymoma 4 parvovirus B19+

IST 46% (6/13) CsA & cs 74% (23/31) with CsA, 60% (12/20) with cs Surgical removal 95% (19/20) with CsA, 46% (6/13) with cs of thymus + IST (CsA & CTX) Rituximab 100% complete remission at 16wks. post treatment Alemtuzumab CRR of 67% (8/12); partial RR of 17% (2/12) IVIG CsA vs. cs vs. cs + CSA

90% (9/10) after first course CRR of 27.8% vs. 6.67% vs. 66.7%

Only 7 were treated with bortezomib & lenalidomide Etiology based therapeutic approach; CsA/tacrolimus, CTX, MTX, IVIG, alemtuzumab, rituximab, ATG and others^

43% (3/7) CRR

As in Figure 3 and Tables 2 & 3

*6 Mercaptopurine, azathioprine and cyclophosphamide. ^Others include danazol, mycophenolate mofetil, bortezomib, erythropoietin, abatacept (Orencia), and tofacitinib (Xeljanz). #Others include viral cause and mismatched allogeneic stem cell transplant. LGL: large granular lymphocytic; CLL: chronic lymphocytic leukemia; MGUS: monoclonal gammopathy of undetermined significance; CsA: cyclosporine; cs: corticosteroids; CTX: cytoxan (cyclophosphamide); IST: immunosuppressive therapy; IVIG: intravenous gammaglobulin; CKD: chronic kidney disease; EPO: erythropoietin; ORR: overall response rate; CRR: complete response rate; MTX: methotrexate; ATG: anti-thymocyte globulin; LGL/PRCA: large granular lymphocytic leukemia related/pure red cell aplasia; NHL: non-hodgkin lymphoma.

Occasionally, methotrexate (MTX) was used as a first-line choice of treatment with a steroid taper, especially in the context of LGL-associated PRCA: the overall CRR remained low (20%, 2/10) along with the ORR (30%, 3/10) (Figure 3A and Table 2). In our cohort, 13 refractory patients received Campath (alemtuzumab) therapy with an ORR of 54% (7/13) (Figure 3A and Table 2). Campath, used only as a salvage option on a compassionate basis, showed promising results in the LGL-associated PRCA group, with an ORR of 75% and overall CRR of 63% (Figure 3B and Table 3). Campath was given at a dose of 10 mg/week for 4-6 weeks following a test dose of 3 mg in the first week, and assessed for response with careful monitoring of the blood counts. Epstein-Barr virus (EBV) and cytomegalovirus (CMV) testing were carried out prior to starting campath haematologica | 2018; 103(2)

so that appropriate antiviral prophylaxis could be given with the drug. Patients with good responses stayed on a maintenance dose of 10 mg subcutaneously every four to eight weeks based on their response; in most cases, clinical response was associated with the improvement of relative - or when present - absolute lymphocytosis and a decrease in the percentages of LGL in the blood. None of our patients had any adverse events following the use of campath at this dose. When anti-thymocyte globulin (ATG) was used in otherwise refractory aPRCA, ORR was low at 30% with only 1/10 patients showing a CR (Figure 3, Tables 2 and Table 3). Rituximab, used only in the salvage setting (mainly deployed in situations where the patients failed multiple T-cell-directed immunosuppressive therapy), showed similarly low RR (overall CRR 36% and ORR 64%; Table 2 and Table 3). 227

S.K. Balasubramanian et al.

Special considerations Intravenous immunoglobulin (IVIG), primarily used in parvovirus-related PRCA, showed a 100% ORR in four selected patients (Table 3), and 3/4 patients remained in complete remission during their last follow up (median follow-up of 23.4 months), except for one patient whose response only lasted for about three months, and who was still not in CR after several rounds of different immunosuppressive therapies (IST). IVIG was also found to be effective in patients with low immunoglobulins (including Good’s syndrome) with an excellent ORR (11/11, 100%), however, the response was short-lived, lasting less than three months for most, with the exception of three patients who remained in CR, requiring only monthly maintenance of IVIG. The response to IVIG was only mediocre when used in idiopathic aPRCA patients who were refractory to prior rounds of multiple IST (22%, 4/18; Table 3). In two patients with MGUS and PRCA, bortezomib (Velcade) was used and a CR and PR were achieved. Similarly, a response was seen in patients with PRCA and immunoplasmacytic lymphoma/von Waldenstrom’s macroglobulinemia. Out of nine patients with thymoma, one achieved a CR after thymectomy without additional treatment; 6/7 remained in CR after requiring maintenance IST. With respect to DBA patients in our cohort, 3/4 were responsive to prednisone and/or anabolic steroids until the end of last follow up. Other salvage options used include anabolic steroids (danazol), mycophenolate mofetil (cellcept), abatacept (orencia), tofacitinib (xeljanz) and epoetin alfa (procrit). Orencia and cellcept were used primarily in LGL PRCA patients who failed to respond to campath, our first-line salvage option for them. In general, second-line salvage options, used in the right setting, showed a moderate ORR of 60%, but an overall RR of 88% in the non-LGL group (Figure 3, Table 2 and Table 3).

Discussion Progress in the management of rare diseases relies on the accumulation of empiric experiences, integrative analyses of several cases, and rarely, clinical trials. Guidelines for the management of PRCA are incomplete and, outside of a few referral centers, patients rely on the inquisitive minds of their hematologists. Given the rarity of this condition, we were inspired to further investigate the large number of cases of PRCA seen in our institution. A summary of our diagnostic approach and treatment results, and study outcomes previously published, are shown in Table 4. We also studied patient etiology, clinical phenotypes, immunological characteristics and genomic data. Our diagnostic approach was based on rational steps to exclude congenital disease (rare in adults), and malproductive anemia, which includes nutritional deficiencies, systemic diseases, and iatrogenic/drug-related diseases. Thereafter, infectious etiologies, in particular B19-mediated TAC, were distinguished in order to arrive at idiopathic- and pathophysiologically-related T-LGL cases. The diagnostic fractions obtained by this approach likely reflect those that would be seen in the community. Idiopathic aPRCA is present in a high percentage of cases and is likely T-cell-mediated. There are also reports of PRCA mediated by antibodies specific for erythroid progenitors.10,13 Mature red cells would have to lack such a tar228

get antigen as they would have been annihilated otherwise, as is the case in hemolytic anemia. While anti-EPO antibodies have been invoked in some cases treated with recombinant EPO, in our cohort not a single case suggested the presence of such antibodies. Nevertheless, rare responses to rituximab may imply the presence of erythroid antibodies in an occasional patient, but routine laboratory tests are not available to identify such patients. Lastly, the identification of clonal mutations, including subclonal hierarchy, with an assigned role in MDS may seem to suggest some are cases of “missed” MDS, or early developing MDS, but our results do not support this notion. Conversely, the presence of clonal events in older patients may simply be consistent with the diagnosis of clonal hematopoiesis of indeterminate potential (CHIP) or underscore, as in aplastic anemia, the contraction of normal progenitor and stem cell pools with a greater likelihood of detecting clonal events.25,26 The most common form of idiopathic aPRCA is mediated by a spectrum of T-cell responses, ranging from polyclonal to oligo and monoclonal, as seen in T-LGL where it constitutes a maximally skewed version of an originally polyclonal response. The presence of STAT3 mutations may support this notion,27 as STAT3 mutant clonal cells are usually a subfraction of clonally TCR-rearranged cytotoxic lymphocytes (CTL) as determined by quantitative analyses of TCR Vβ-chain sequencing (data not shown). Similarly, Vβ flow cytometry performed herein showed increased clonal skewing in some cases of aPRCA short of fully defined monoclonal expansion (Figure 1B). It is possible that a similar process is present in patients with B-cell dyscrasias. Abnormal B-cell or plasma cells can trigger a cross-reactive T-cell response28 that can evolve from polyclonal to clonal and be further fixed by STAT3 mutations. Such a scenario would explain the need to treat underlying B-cell/plasma cell lymphoproliferative processes, even if it is otherwise asymptomatic. Indeed, the presence of severe anemia may upstage the morphologically weighted severity of these conditions. Similar processes may also be operative in cases with thymoma, wherein a clearly ISTsensitive process is maintained by the presence of thymoma and interrupted by thymectomy. An alternative mechanism may be operative in hypogammaglobulinemia, wherein excessive, possibly cross-reactive responses are triggered by antigens that cannot be cleared due to the underlying immunodeficiency, as in cases of hypogammaglobulinemia in Good’s syndrome.29 Significantly, some cases of aPRCA have also been described in the context of NK-LGL,30 but while occasional cases did show nonspecific elevation of NK cells, we did not encounter such situations in our patients. The proper distinction of pathogenetic mechanisms is essential for selecting the most effective therapeutic modality. Historically, CsA has by far been the most successfully used drug in the treatment of idiopathic aPRCA, with responses ranging from 66-95%.31-36 However, most of these studies did not examine a rational selection of treatments in various clinical contexts, nor did they compare efficacies to those of other available drugs. Herein, the systematic administration of CsA showed a similar response rate, however, a subgroup analysis revealed better responsiveness of non-LGL vs. T-LGL-associated PRCA. Tacrolimus showed activity as a salvage option in a few reports,37,38 and we were able to substantiate those findings in our study. While no head-to-head comparisons haematologica | 2018; 103(2)

Biology and management of PRCA

were conducted, previous reports suggested lower response rates to CTX vs. CsA, but prior studies utilized various combinations of CTX and involved relatively small numbers of patients.20,31,39 Nevertheless, given results comparable to CsA, CTX remains a front-line option in aPRCA, particularly in older patients and taking into consideration the oncogenic risks associated with chronic exposure to CTX. Despite the widespread application of MTX in T-LGL, its utility is mostly in the treatment of idiopathic or LGL-associated neutropenia, and results with MTX in aPRCA appear to be inferior to those of CsA or CTX. Conceptually, the lack of response to IST may be due to insufficient intensity or alternate pathophysiology. Assuming that the latter forms are rare, salvage IST agents include campath and ATG. Campath has been effective in autoimmune disorders and in T-LGL,40-42 thus, we and others treat aPRCA with this drug,43,44 using low-dose subcutaneous regimens to limit infectious complications. Our experience with campath in aPRCA has shown that it yields the highest number of meaningful responses in patients, including those refractory to CTX or CsA and those who could not tolerate calcineurin inhibitors. Despite the small number of patients, our results indicate that the best response can be expected in T-LGL-associated PRCA. The concern of IST-related infectious complications with campath dictates antimicrobial-antiviral prophylaxis and careful monitoring. With these measures and low-dose subcutaneous treatment, the risks appear to be manageable. ATG has been reported to be used in a few patients with aPRCA. It was used in our series as a salvage option that had an acceptable success rate. The identification of B19-related chronic PRCA or TAC has obvious therapeutic implications.45,46 Indeed, the results herein are similar to those of a published series:45 3/4 patients receiving IVIG for parvovirus PRCA responded with durable remissions. It is possible that patients with immunodeficiency, even in the absence of diagnostic signs of B19, may benefit from IVIG that could clear a hypothetically unrecognized viral agent. We recorded responses in hypogammaglobulinemic patients (including Goods syndrome), who received IVIG as preparation for

References 1. Brown K, Young N. Parvovirus B19 infection and hematopoiesis. Blood Rev. 1995; 9(3):176-182. 2. Thompson DF, Gales MA. Drug‐induced pure red cell aplasia. Pharmacotherapy. 1996;16(6):1002-1008. 3. Stohlman F, Quesenberry P, Howard D, Miller M, Schur P. Erythroid aplasia: an autoimmune complication of chronic lymphocytic leukemia. Clin Res. 1971;19:566. 4. Masauzi N, Tanaka J, Watanabe M, et al. Primary Waldenström's macroglobulinemia associated with pure red cell aplasia in which Ts/c lymphocytes inhibiting erythroid precursors were detected. [Rinsho ketsueki] Jpn J Clin Hematol. 1993; 34(3):355-361. 5. Kobayashi T, Hanada T, Sato Y, et al. A case of pure red cell aplasia with monoclonal gammopathy: immune-mediated

haematologica | 2018; 103(2)

10.

IST. Responses to IVIG in auto-antibody mediated PRCA have been reported previously.47 Contrasting results have been reported with rituximab in aPRCA patients refractory to multiple IST.48,49 Ex juvantibus responsiveness to rituximab indicates a potentially B-cell mediated disease,50 or implies that PRCA may be a paraneoplastic syndrome of a coexisting B-cell dyscrasia.51 Consequently, one would not expect responses to rituximab in idiopathic or T-LGL-associated PRCA. Similarly, coexisting MGUS or plasma cell disorders have been successfully treated with bortezomib,52 and we were also able to make a similar observation in otherwise refractory aPRCA with co-associated monoclonal Immunoglobulin G (IgG) or Immunoglobulin M (IgM) gammopathy. In summary, the succession of treatments was rationally decided upon in order to include the most likely target and least intense treatment, i.e., CsA or CTX, as a first-line choice, while salvage therapeutic options could include either B-cell-directed modality, anabolic steroids or higher intensity of immunosuppression. While the resulting salvage RR is likely due to the regimen applied, we cannot rule out the cumulative effects of such increased immunosuppression with sequentially used immunosuppressive agents. Allogenic bone marrow transplant has been tried in the past for DBA patients who are refractory to steroids,53 however, the timing of transplant for acquired immune-mediated PRCA can still be debated as, to date, there are no randomized studies exploring the potential benefit of early bone marrow transplant. Arguably, it could be a potential option for medically refractory disease failing multiple different ISTs.54 In conclusion, the proper management of PRCA requires consideration of the clinical context, which can provide clues to the therapeutic modality, especially in the refractory setting. Funding This work was supported by the EPE1509JM - Edward P. Evans Foundation, R01HL123904 – NHLBI, R01HL118281 – NHLBI, R01HL128425 – NHLBI

inhibition of erythropoiesis. [Rinsho ketsueki] Jpn J Clin Hematol. 1987; 28(11): 2029-2033. Orchard J, Myint H, Hamblin TJ. A patient with myeloma who still has pure red cell aplasia despite the most intensive immune modulation. Leuk Res. 1997;21(4):353-354. Abkowitz JL, Kadin ME, Powell JS, Adamson JW. Pure red cell aplasia: lymphocyte inhibition of erythropoiesis. Br J Haematol. 1986;63(1):59-67. Levitt LJ, Reyes GR, Moonka DK, et al. Human T cell leukemia virus-I-associated T-suppressor cell inhibition of erythropoiesis in a patient with pure red cell aplasia and chronic T gamma-lymphoproliferative disease. J Clin Invest. 1988;81(2):538. Bailey RO, Dunn HG, Rubin AM, Ritaccio AL. Myasthenia gravis with thymoma and pure red blood cell aplasia. Am J Clin Pathol. 1988;89(5):687-693. Casadevall N, Dupuy E, Molho-Sabatier P,

11.

12.

13.

14. 15.

et al. Autoantibodies against erythropoietin in a patient with pure red-cell aplasia. N Engl J Med. 1996;334(10):630-633. Dessypris EN, Krantz SB, Roloff JS, Lukens JN. Mode of action of the IgG inhibitor of erythropoiesis in transient erythroblastopenia of children. Blood. 1982;59(1):114-123. Duchmann R, Schwarting A, Poralla T, zum Büschenfelde K-HM, Hermann E. Thymoma and pure red cell aplasia in a patient with systemic lupus erythematosus. Scand J Rheumatol. 1995;24(4):251-254. Peschle C, Marmont AM, Marone G, et al. Pure red cell aplasia: studies on an IgG serum inhibitor neutralizing erythropoietin. Br J Haematol. 1975;30(4):411-417. Williamson P, Oscier D, Bell A, Hamblin T. Red cell aplasia in myelodysplastic syndrome. J Clin Pathol. 1991;44(5):431-432. Kaznelson P. Zur entstehung der blutplättchen. Verh Dtsch Ges Inn Med. 1922;34:557-559.

229

S.K. Balasubramanian et al. 16. Young N. Hematologic and hematopoietic consequences of B19 parvovirus infection. Semin Hematol. 1988;25(2):159-172. 17. Browman G, Freedman M, Blajchman M, McBride J. A complement independent erythropoietic inhibitor acting on the progenitor cell in refractory anemia. Am J Med. 1976;61(4):572-578. 18. Charles RJ, Sabo KM, Kidd PG, Abkowitz JL. The pathophysiology of pure red cell aplasia: implications for therapy. Blood. 1996;87(11):4831-4838. 19. Hirayama Y, Nagai T, Ohta H, et al. A case of pure red cell aplasia accompanied with granular lymphocytic leukemia the tumor cells of which suppressed colony formation of BFU-E, and which was successfully treated by cyclophosphamide and cyclosporin A. [Rinsho ketsueki] Jpn J Clin Hematol. 1997;38(11):1206-1211. 20. Lacy MQ, Kurtin PJ, Tefferi A. Pure red cell aplasia: association with large granular lymphocyte leukemia and the prognostic value of cytogenetic abnormalities [see comments]. Blood. 1996;87(7):3000-3006. 21. Bergrem H DB, Eckardt KU, Kurtz A, Strisberg M. A case of antierythropoietin antibodies following recombinant human erythropoietin treatment. In: Erythropoietin: molecular physiology and clinical application. New York: Marcel Dekker. 1993;266(75). 22. Casadevall N, Nataf J, Viron B, et al. Pure red-cell aplasia and antierythropoietin antibodies in patients treated with recombinant erythropoietin. N Engl J Med. 2002; 346(7):469-475. 23. Murray W, Webb J. Thymoma associated with hypogammaglobulinaemia and pure red cell aplasia. Am J Med. 1966;41(6):974980. 24. Wang K, Li M, Hakonarson H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 2010;38(16):e164-e. 25. Maciejewski J, Balasubramanian SK. Clinical implications of somatic mutations in aplastic anemia and myelodysplastic syndrome in genomic age (Article in press). Hematology Am Soc Hematol Educ Program. 2017. 26. Yoshizato T, Dumitriu B, Hosokawa K, et al. Somatic mutations and clonal hematopoiesis in aplastic anemia. N Engl J Med. 2015;373(1):35-47. 27. 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. 28. Bassan R, Pronesti M, Buzzetti M, et al. Autoimmunity and B cell dysfunction in chronic proliferative disorders of large granular lymphocytes/natural killer cells. Cancer. 1989;63(1):90-95. 29. Good RA. Agammaglobulinaemia—a

230

30.

31.

32.

33. 34.

35.

36.

37.

38.

39.

40.

41.

42.

provocative experiment of nature. Bulletin of the University of Minnesota. 1954; 26:119. Handgretinger R, Geiselhart A, Moris A, et al. Pure red-cell aplasia associated with clonal expansion of granular lymphocytes expressing killer-cell inhibitory receptors. N Engl J Med. 1999;340(4):278-284. Hirokawa M, Sawada K-i, Fujishima N, et al. Long-term response and outcome following immunosuppressive therapy in thymoma-associated pure red cell aplasia: a nationwide cohort study in Japan by the PRCA collaborative study group. Haematologica. 2008;93(1):27-33. Kwong Y, Wong K, Liang R, et al. Pure red cell aplasia: clinical features and treatment results in 16 cases. Ann Hematol. 1996; 72(3):137-140. Mamiya S, Itoh T, Miura AB. Acquired pure red cell aplasia in Japan. Eur J Haematol. 1997;59(4):199-205. Sawada K, Fujishima N, Hirokawa M. Acquired pure red cell aplasia: updated review of treatment. Br J Haematol. 2008; 142(4):505-514. Sawada K-i, Hirokawa M, Fujishima N, et al. Long-term outcome of patients with acquired primary idiopathic pure red cell aplasia receiving cyclosporine A. A nationwide cohort study in Japan for the PRCA Collaborative Study Group. Haematologica. 2007;92(8):1021-1028. Wu X, Wang S, Shen W, et al. Adult patients with acquired pure red cell aplasia: treated by cyclosporine a and/or corticosteroids-single center experience. Blood. 2016;128(22):4818. Hashimoto K, Harada M, Kamijo Y. Pure red cell aplasia induced by anti-erythropoietin antibodies, well-controlled with tacrolimus. Int J Hematol. 2016;104(4):502505. Yoshida S, Konishi T, Nishizawa T, Yoshida Y. Effect of tacrolimus in a patient with pure red cell aplasia. Int J Lab Hematol. 2005;27(1):67-69. Clark DA, Dessypris EN, Krantz SB. Studies on pure red cell aplasia. XI. Results of immunosuppressive treatment of 37 patients. Blood. 1984;63(2):277-286. Au W-Y, Lam CC, Chim C-S, Pang AW, Kwong Y-L. Alemtuzumab induced complete remission of therapy-resistant pure red cell aplasia. Leuk Res. 2005; 29(10):1213-1215. Rodon P, Breton P, Courouble G. Treatment of pure red cell aplasia and autoimmune haemolytic anaemia in chronic lymphocytic leukaemia with Campath 1H. Eur J Haematol. 2003;70(5):319-321. Schutzinger C, Gaiger A, Thalhammer R, et al. Remission of pure red cell aplasia in Tcell receptor [gamma][delta]-large granular lymphocyte leukemia after therapy with

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

low-dose alemtuzumab. Leukemia. 2005; 19(11):2005-2008. Risitano AM, Selleri C, Serio B, et al. Alemtuzumab is safe and effective as immunosuppressive treatment for aplastic anaemia and single lineage marrow failure: a pilot study and a survey from the EBMT WPSAA. Br J Haematol. 2010;148(5):791796. Thota S, Patel BJ, Sadaps M, et al. Therapeutic outcomes using subcutaneous low dose alemtuzumab for acquired bone marrow failure conditions. Br J Haematol. 2017 Sep 14. [Epub ahead of print] Crabol Y, Terrier B, Rozenberg F, et al. Intravenous immunoglobulin therapy for pure red cell aplasia related to human parvovirus B19 infection: a retrospective study of 10 patients and review of the literature. Clin Infect Dis. 2013;56(7):968-977. Kurtzman G, Frickhofen N, Kimball J, et al. Pure red-cell aplasia of 10 years' duration due to persistent parvovirus B19 infection and its cure with immunoglobulin therapy. N Engl J Med. 1989;321(8):519-523. Linardaki GD, Boki KA, Fertakis A, Tzioufas AG. Pure red cell aplasia as presentation of systemic lupus erythematosus: antibodies to erythropoietin. Scand J Rheumatol. 1999;28(3):189-191. Auner HW, Wölfler A, Beham Schmid C, et al. Restoration of erythropoiesis by rituximab in an adult patient with primary acquired pure red cell aplasia refractory to conventional treatment. Br J Haematol. 2002;116(3):727-728. Dungarwalla M, Marsh J, Tooze J, et al. Lack of clinical efficacy of rituximab in the treatment of autoimmune neutropenia and pure red cell aplasia: implications for their pathophysiology. Ann Hematol. 2007; 86(3):191-197. D’Arena G, Vigliotti ML, Dell’Olio M, et al. Rituximab to treat chronic lymphoproliferative disorder associated pure red cell aplasia. Eur J Haematol. 2009;82(3):235-239. Alter R, Joshi SS, Verdirame JD, Weisenburger DD. Pure red cell aplasia associated with B cell lymphoma: demonstration of bone marrow colony inhibition by serum immunoglobulin. Leuk Res. 1990; 14(3):279-286. Korde N, Zhang Y, Loeliger K, et al. Monoclonal gammopathy associated pure red cell aplasia. Br J Haematol. 2016; 173(6):876-883. Roy V, Pérez WS, Eapen M, et al. Bone marrow transplantation for diamond-blackfan anemia. Biol Blood Marrow Transplant. 2005; z11(8):600-608. Tseng SB, Lin SF, Chang CS., et al. Successful treatment of acquired pure red cell aplasia (PRCA) by allogeneic peripheral blood stem cell transplantation. Am J Hematol. 2003;74(4):273-275.

haematologica | 2018; 103(2)

ARTICLE

Phagocyte Biology and its disorders

The clinical and laboratory evaluation of familial hemophagocytic lymphohistiocytosis and the importance of hepatic and spinal cord involvement: a single center experience

Ferrata Storti Foundation

Burcin Beken,1 Selin Aytac,2 Gunay Balta,2 Baris Kuskonmaz,2 Duygu Uckan,2 Sule Unal,2 Mualla Cetin2 and Fatma Gumruk2

1 Hacettepe University Department of Pediatrics, and 2Hacettepe University Department of Pediatric Hematology, Ankara, Turkey

Haematologica 2018 Volume 103(2):231-236

ABSTRACT

amilial hemophagocytic lymphohistiocytosis is an autosomal recessive, life-threatening condition characterized by defective immune response. A retrospective analysis was performed on 57 patients diagnosed with familial hemophagocytic lymphohistiocytosis at Hacettepe University Pediatric Hematology Department, Ankara, Turkey. Mutation analysis was performed on 37 patients, and of these: 11 had UNC13D, 10 had PRF1 and 3 had STX11 gene mutation. Of these patients, 44% were found to have central nervous system involvement on admission and spinal cord involvement was also seen in 5 patients. Remission was achieved in 24 patients with the treatment, in a median time of 76 days (min-max: 15-705 days). Time to remission was prolonged 3.1 times in patients with a ferritin level 1500 mg/dL or more. When patients were grouped according to age [Group 1 (â&#x2030;¤ 2 years), Group 2 (>2 years)]; patients in Group 1 had higher ferritin and aspartate aminotransferase levels but lower fibrinogen levels. The 5-year survival rate was also lower in Group 1. When patients in Group 1 were divided into two sub-groups according to hepatic involvement, the 5-year survival rate of patients who had hepatic involvement was significantly lower than those patients without hepatic involvement (0.7%, 27%, respectively) (P=0.002). The 5-year survival rate of patients who underwent hematopoietic stem cell transplantation was significantly higher than the patients who didn't (44%, 16%, respectively) (P=0.02). In conclusion, age two years and under, ferritin level above 1500 mg/dL, spinal cord or hepatic involvement should be considered as poor prognostic factors in familial hemophagocytic lymphohistiocytosis.

Correspondence: burcinbeken@gmail.com

Received: August 6, 2017. Accepted: November 13, 2017. Pre-published: November 16, 2017.

Introduction

doi:10.3324/haematol.2017.178038

Familial hemophagocytic lymphohistiocytosis (FHL) is a rare, autosomal recessive disease characterized by uncontrolled activation of T cells and macrophages, and excessive production of cytokines.1,2 The incidence of FHL in children varies in different studies. In Sweden, for example, it has been reported to be 1.2/1,000,000 while the frequency among hospitalized patients reported from Turkey is 1/1418.3 Although approximately 70% of patients with FHL are diagnosed before the age of one year, familial forms that manifest in late adulthood have also been reported.4,5 FHL is a genetically heterogeneous disorder and genotype-phenotype correlation is limited. So far, four genes (PRF1, UNC13D, STX11 and STXBP2) associated with FHL have been identified; unfortunately, in 10-50% of patients, none of these molecular defects can be determined.1,2 The most common symptoms of FHL are fever, hepatosplenomegaly and cytopenia. However, all organ systems can be involved at the time of diagnosis or during follow up. Jaundice, elevated liver enzymes, hepatomegaly, and coagulopathy are the signs of hepatobiliary involvement.6 Central nervous system (CNS) involvement has been reported to be associated with poor outcome and isolated CNS involvement can also be seen.7-10 In this study, we aimed to evaluate the demographic, clinical and laboratory findings of FHL patients followed in Hacettepe University Childrenâ&#x20AC;&#x2122;s Hospital and to

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/231

haematologica | 2018; 103(2)

ÂŠ2018 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.

231

B. Beken et al.

determine the risk factors affecting the treatment response, prognosis, and survival.

tate aminotransferase (ALT, AST) levels, low albumin levels, ascites and coagulopathy according to normal values by age. Complete remission and reactivation were decided on the basis of the criteria of the HLH society protocol.11

Methods Molecular genetic study for FHL Patients Retrospective, single center data were obtained from hemophagocytic lymphohistiocytosis (HLH) patients who were diagnosed between November 1994 and December 2012 in the Pediatric Hematology Department of Hacettepe University Children’s Hospital, Ankara, Turkey. The HLH-94 and HLH-2004 diagnostic criteria of the Histiocyte Society were applied to confirm the diagnosis.11,12 Written informed consent was provided by the patients’ parents and the study was approved by Hacettepe University Ethical Committee.

Materials A diagnosis of FHL was made when there were affected siblings/close relatives (family history) or a history of consanguineous marriage, relapse, or when the molecular study was consistent with FHL. Patients with an underlying malignancy, metabolic disorder or rheumatological disease were excluded. The medical records of all patients were evaluated for demographic features, clinical and laboratory findings, bone marrow aspiration evaluation, mutation analysis and treatment received including hematopoietic stem cell transplantation (HSCT), complications, and mortality. CNS disease was evaluated by neurological findings such as neck stiffness, a bulging fontanelle, seizure, cranial nerve paralysis, hemiparesis, ataxia, drowsiness, and coma. Computed tomography (CT) or magnetic resonance imaging (MRI) findings indicative of involvement, or cerebrospinal fluid (CSF) protein level more than 30 mg/dL and/or a cell count of more than 50x106/L observed in the CSF (more than 5 white blood cells/mL) and/or evidence of hemophagocytosis in the CSF.13 Hepatic involvement was defined as elevated bilirubin, alanine and aspar-

A molecular genetic study was performed on the DNA samples of 37 patients by direct sequencing of all the coding exons as previously described in the literature.14

Statistical analysis The software package SPSS v.16.0 was used in the statistical analysis of the data. Numerical variables were shown using mean±Standard Deviation (SD) and median (distribution) and qualitative variables were shown using numbers and percentages. The normality of the numerical variables was evaluated using the Shapiro Wilks test. To determine whether there was a difference with regard to qualitative variables between groups, the Χ2 test was used. To determine whether there was a difference with regard to numerical variables between two groups, the MannWhitney U test was used. Survival analyses were carried out using the Kaplan-Meier product-limit estimator method. Survival curves were compared using the log-rank test. To determine whether there was a difference with regard to numerical variables between more than two groups, the Kruskal-Wallis test was used. In determining the factors affecting survival, Cox regression analysis was used; P<0.05 was considered statistically significant.

Results Patients’ characteristics Patients’ characteristics and outcomes are shown in Table 1. A total of 37 patients were evaluated for mutations associated with FHL. It was determined that there were perforin mutations in a total of 10 patients, 4 of

Figure 1. Clinical findings of patients at the time of diagnosis. Fever and hepatosplenomegaly were the most common symptoms at the time of diagnosis.

232

haematologica | 2018; 103(2)

Hepatic and spinal cord involvement in FHL

whom had been previously published.14,15 UNC13D and STX11 mutations were detected in 11 and 3 patients, respectively. There was no statistical difference in terms of demographic, clinical and laboratory findings, and survival between the mutation positive and negative groups. A history of a deceased sibling was significantly higher in mutation-positive patients (P=0.027). Serological evaluation for the Epstein-Barr virus (EBV) was performed in 41 patients and was compatible with previous EBV infection in 32 patients (78%) and active infection in one patient (2%). The patient was also treated with intravenous ganciclovir. Clinical features of the patients are shown in Figure 1.

Comparison of clinical and laboratory findings according to the patientsâ&#x20AC;&#x2122; initial age at diagnosis Patients were categorized into two groups according to age: â&#x2030;¤ 2 years (Group 1) and >2 years (Group 2). The comparison of the laboratory findings and the 5-year survival rates of these two groups are shown in Table 2.

shown in Online Supplementary Table S1. The female/male ratio of patients with CNS involvement at the time of diagnosis was significantly higher than those without (3.1/1, 0.6/1, respectively) (P=0.015). Median age at diagnosis was higher in patients with CNS involvement than those without (36 months, 10.5 months, respectively) (P=0.028). There was no difference in 5-year survival rates between the two groups.

Comparison of HLH-94 and HLH-2004 treatment protocols, HSCT results and outcomes Thirty-two patients (56%) were treated according to the HLH-2004 protocol, 17 patients (30%) were treated according to the HLH-1994 protocol, and 8 patients (14%) received only supportive care (those patients in a poor clinical condition and who died before initiation of treatment). There were no statistical differences in remission

Table 1. Patients' characteristics, some clinical findings and outcomes.

Comparison of clinical and laboratory findings according to hepatic involvement Twenty-six (46%) patients had hepatic involvement at the time of diagnosis; 15 (58%) of these patients were two years old or younger, and the remaining 11 (42%) were older than two years of age when they were diagnosed. There was no statistical difference in 5-year survival rates between patients with hepatic involvement and without hepatic involvement (20%, 26%, respectively) (P=0.210). Analysis of the patients under two years old for hepatic involvement revealed that the 5-year survival rate of the patients with hepatic involvement was significantly lower than those without hepatic involvement (0.7%, 27%, respectively) (P=0.002).

Comparison of clinical and laboratory findings according to CNS involvement Twenty-five patients (44%) had CNS involvement at the time of diagnosis and 17 patients (30%) developed CNS involvement during follow up. While convulsion was the most common (48%) neurological symptom, the others reported were: cranial nerve involvement, hemiparesis, somnolence, confusion, headache, neck stiffness, delirium and ataxia. Five patients did not have neurological symptoms, but either MRI or CSF findings (elevated protein, histiocytes causing moderate pleocytosis, or hemophagocytosis) were compatible with HLH involvement. There were 6 patients diagnosed over ten years of age; 3 of these patients (10, 14 and 15 years old) had isolated CNS involvement at the time of diagnosis and all died despite undergoing HSCT. Two patients had both CNS and bone marrow infiltration (one of them is still alive) and one patient had only bone marrow infiltration (still alive). Spinal cord involvement was determined in 5 patients, including one at diagnosis and 4 during follow up; one of these patients has been previously published.16 Median age of the patients with spinal cord involvement was 11 years and all had neurological symptoms at the time of diagnosis. All of the patients with spinal cord involvement died despite 4 undergoing HSCT. Three patients had isolated CNS involvement at the time of diagnosis and 2 of them had concomitant spinal cord involvement. The clinical features of patients with spinal cord involvement are haematologica | 2018; 103(2)

N. of patients (%) Sex Female Male Consanguinity between parents (+) (-) History of an affected sibling (+) (-) CNS involvement on admission (+) (-) Molecular genetic analysis UNC13D PRF1 STX11 Undetectable Treatment protocol HLH-1994 HLH-2004 Supportive treatment HSCT (+) (-) Relapse Bone marrow CNS Bone marrow and CNS Outcome Alive without disease 11(19%) Dead 46(81%)

32 (56%) 25 (44%) 40 (70%) 17 (30%) 23 (40%) 34 (60%) 25 (44%) 32 (56%) 37 (65%) 11 (30%) 10 (27%) 3 (8%) 13 (35%) 17 (30%) 32 (56%) 8 (14%) 18 (32%) 39 (68%) 20 (35%) 3 (15%) 3 (15%) 14 (70%) Patients with HSCT: 9 Patients without HSCT: 2 Patients with HSCT: 9 Patients without HSCT: 37

N.: number; CNS: central nervous system; HSCT: hematopoietic stem cell transplantation.

233

B. Beken et al.

(35%, 56%, respectively) (P=0.27) and 5-year survival rates (37%, 17%, respectively) (P=0.17) between HLH1994 and HLH-2004 treatment protocols. Eighteen patients (32%) underwent HSCT. Nine were lost because of subsequent complications. The 5-year survival rate of patients with HSCT were significantly higher than those without HSCT (44%, 16%, respectively)

(P=0.02). Twenty-four patients went into remission with a median time of 76 days (15-705 days). Among the factors that could affect the remission time, a ferritin level higher than 1500 ng/mL was found to extend the time to remission by 2.3 times (95%CI: 0.9-5.7). When data were evaluated through multiple-variable analysis with triglycerides, fibrinogen, and sodium levels, a ferritin level more

Figure 2. Patient follow-up shown as a flow chart. EX: Exitus; BM: bone marrow; CNS: central nerveous system; HLH: hemophagocytic lymphohistiocytosis; HSCT: hematopoetic stem cell transplantation.

Table 2. Comparison of laboratory findings and survival according to age at diagnosis.

Laboratory findings

Group 1 (â&#x2030;¤2 years old) n=33 Median (min-max)

Group 2 (>2 years old) n=24 Median (min-max)

Hemoglobin (g/dL) Leukocyte (x109/L) Thrombocyte (x109/L) Ferritin (ng/dL) Triglyceride (mg/dL) Fibrinogen (mg/dL) ALT (U/L) AST (U/L) Total bilirubin (mg/dL) Conjugated bilirubin (mg/dL) LDH (U/L) Total protein (mg/dL) Albumin (g/dL) Sodium (mmol/L) aPTT (sn) 5-year survival rate

7.9 (3.6-11.8) 6.1 (0.8-16.5) 29 (7-683) 1963 (128-51788) 347 (41-1443) 123 (10-624) 73 (13-539) 104 (16-1894) 1.1 (0.1-25.2) 0.6 (0.1-17.4) 1080 (258-2583) 5.4 (3.6-7.0) 3.0 (1.9-4.0) 134 (125-150) 37 (22-180) 19%

8.3 (4.9-13.1) 4.3 (0.4-10.7) 41 (4-177) 1104 (133-7718) 358 (43-872) 223 (2-483) 69 (5-466) 71 (11-373) 1.3 (0.1-15.5) 0.6 (0.1-11.6) 790 (76-5730) 5.5 (3.6-7.6) 3.2 (1.7-4.2) 136 (120-145) 33 (24-66) 33%

0.682 0.672 0.477 0.001 0.959 0.023 0.884 0.043 0.722 0.686 0.156 0.945 0.597 0.841 0.057 0.017

ALT: alanine aminotransferase; AST: aspartate aminotransferase; aPTT: activated partial thromboplastine time; LDH: lactate dehydrogenase.

234

haematologica | 2018; 103(2)

Hepatic and spinal cord involvement in FHL

than 1500 ng/mL was found to extend the time to remission by 3.1 times (95%CI: 1.1-8.5). As for the factors that affect the 5-year survival rate of the patients who did not undergo HSCT, it was seen that sex, age of diagnosis, CNS involvement, and relapse did not have any effect on survival. When the data with P-values over 0.250 (aPTT, sodium, triglycerides, albumin and relapse) were analyzed with multivariate and Cox regression analysis, a low sodium level (<135 mEq/L) was found to increase the mortality risk by 2.46 times (95%CI: 1.15.3). Follow up of the patients is given as a flowchart in Figure 2.

Discussion Familial hemophagocytic lymphohistiocytosis is an autosomal recessive, life-threatening, inflammatory condition with excessive, prolonged, and ineffective immune response. It is expected to become frequent in Turkey because of the high rates of consanguineous marriages.3 Median age of our patients was 18 months and 47% of the patients were diagnosed before the age of one. According to the literature, 70% of FHL patients are diagnosed in the first year of life. The median age of diagnosis of our patients was thought to be higher because of the 6 patients over ten years of age in our study group. We think that the mean age of FHL diagnosis in the literature will increase with the growing number of adolescent and adult patients reported in recent years. The most common clinical findings on admission were fever, splenomegaly, focus of infection, lymphadenopathy, and rash, and these are similar to the FHL international registry (Figure 3).5 When we analyzed the mutations responsible for FHL, we found the rate of UNC13D mutation slightly higher than in the literature.17 The median ages of diagnosis were the same as in the literature: three months in PRF1 mutation and four months in UNC13D mutation.18 The median

age at diagnosis was 89 months in STX11 mutation. A study by Rudd et al. found STX11 mutation in 6 of 34 patients, half of whom were diagnosed after one year of age and who, unexpectedly, were able to survive for a long period without treatment.19 The 5-year survival rate of the patients who were two years old or under at the time of diagnosis was observed to be significantly lower than the patients over two years of age. On the other hand, when these two groups were compared in terms of laboratory findings, it became apparent that the patients aged two years or under had lower fibrinogen levels, and higher ferritin and AST levels. These parameters, which are related to disease activation have also been observed as being associated with a worse prognosis in other studies.3,20 Taking into consideration the fact that the liver is the site of fibrinogen synthesis, low levels of fibrinogen and elevated AST levels are considered to be associated with liver involvement. We also found that, in the age two years or under patient group, those with hepatic involvement had lower survival rates. In 2009, Flipovich had suggested that liver involvement should be included in the diagnostic criteria.21 There is also a study reporting poor prognosis in cases of secondary HLH presenting with acute hepatic failure.22 In 44% of our patients, CNS involvement was detected at the time of diagnosis, and in 14% it was detected during follow up. According to the literature, 10-73% of patients with FHL have CNS involvement at the time of diagnosis.2,8,15,16 Haddad et al. reported that CNS involvement occurs in particular during the reactivation of the disease, and that all patients without HSCT develop CNS complications. In their study, 39% of the patients had CNS involvement at the time of diagnosis, a percentage that increased to 54% when patients with relapse were included.8 A study by Horne et al. which focused on the frequency of CNS involvement in 193 HLH patients and the characteristics of these patients, reported that patients with CNS involvement were diagnosed at a median of nine months and that 55% of these were under one year of age when diagnosed.13

Figure 3. Comparison of the clinical findings with familial hemophagocytic lymphohistiocytosis (FHL) international registry.5

haematologica | 2018; 103(2)

235

B. Beken et al.

In our study, the median age of diagnosis in patients with CNS involvement was found to be 36 months. Three of our patients had isolated CNS involvement at the time of diagnosis. Even though a few patients were reported to have isolated CNS involvement, we believe that the number of such cases is not limited to those who have been reported because it is rather difficult to diagnose patients with isolated CNS symptoms, which leads to misdiagnosis and/or mistreatment.10,23,24 Although spinal cord involvement in FHL is a rare entity, and thus only a few patients have been reported in the literature to date,9,16 it was detected in 9% of our patients. We suggest that a cranial MRI, accompanied by a spinal MRI should be carried out in patients older than ten years of age and who have neurological symptoms in order to verify whether they have spinal cord involvement or not. Although it is a well-known fact that CNS involvement has a negative effect on FHL prognosis,8,13,15 we did not find any differences between the survival rates of patients with and those without CNS involvement. Approximately 33% of our patients underwent HSCT.

References 1. Gholam C, Grigoriadou S, Gilmour KC, Gaspar HB. Familial haemophagocytic lymphohistiocytosis: advances in the genetic basis, diagnosis and management. Clin Exp Immunol. 2011;163(3):271-283. 2. Janka GE. Familial and Acquired Hemophagocytic Lymphohistiocytosis. Annu Rev Med. 2012;63:233-246. 3. Gurgey A, Gogus S, Ozyurek E, et al. Primary hemophagocytic lymphohistiocytosis in Turkish children. Pediatr Hematol Oncol. 2003;20(5):367-371. 4. Henter JI, Elinder G, Söder O, Ost A. Incidence in Sweden and clinical features of familial hemophagocytic lymphohistiocytosis. Acta Paediatr Scand. 1991; 80(4):428435. 5. Aricò M, Janka G, Fischer A, et al. Hemophagocytic lymphohistiocytosis. Report of 122 children from the International Registry. FHL Study Group of the Histiocyte Society. Leukemia. 1996;10(2):197-203. 6. Guthery SL, Heubi JE. Liver involvement in childhood histiocytic syndromes. Curr Opin Gastroenterol. 2001;17(5):474-478. 7. Henter JI, Nennesmo I. Neuropathologic findings and neurologic symptoms in twenty-three children with hemophagocytic lymphohistiocytosis. J Pediatr. 1997;130(3):358-365. 8. Haddad E, Sulis ML, Jabado N, Blanche S, Fischer A, Tardieu M. Frequency and severity of central nervous system lesions in hemophagocytic lymphohistiocytosis. Blood. 1997;89(3):794-800. 9. Chiapparini L, Uziel G, Vallinoto C, et al. Hemophagocytic lymphohistiocytosis with neurological presentation: MRI findings and a nearly miss diagnosis. Neurol Sci. 2011;32(3):473-477.

236

Their 5-year survival rate was found to be 44%, similar to that reported in the literature (49-66%). When the factors affecting remission time were analyzed, it was seen that ferritin level higher than 1500 ng/mL extends the time to remission by 3.1 times. In 2012, Trottestam et al. demonstrated that elevated ferritin, bilirubin, creatinine, and aPTT levels together with low albumin levels, and the presence of pleocytosis in the CSF at the time of diagnosis, have an impact on early mortality, and a ferritin level more than 2000 ng/mL increased mortality by 3.2 times.20 It can be assumed that high mortality rates in patients with elevated ferritin levels are related to poor response to treatment. In conclusion, FHL is a disease with a fatal course, its only curative treatment being HSCT. Mutation analysis should be performed immediately in HLH patients in order to avoid any delay for HSCT. Patients aged two years and under, those with a ferritin level more than 1500 mg/dL at the time of diagnosis, and who have spinal cord or hepatic involvement, should be considered as having a poor prognosis.

10. Rostasy K, Kolb R, Pohl D, et al. Central nervous system disease as the main manifestation of hemophagocytic lymphohistiocytosis in two cildren. Neuropediatrics. 2004;35(1):45-49. 11. Henter JI, Horne A, Aricó M, et al. HLH2004: Diagnostic and Therapeutic Guidelines for Hemophagocytic Lymphohistiocytosis. Pediatr Blood Cancer. 2007;48(2):124-131. 12. Henter J-I, Arico M, Egeler M, et al. HLH-94: A treatment protocol for hemophagocytic lymphohistiocytosis. Med Pediatr Oncol. 1997;28(5):342-347. 13. Horne A, Trottestam H, Arico M, et al. Frequency and spectrum of central nervous system involvement in 193 children with haemophagocytic lymphohistiocytosis. Br J Haematol. 2008;140(3):327-335. 14. Okur H, Balta G, Akarsu N, et al. Clinical and molecular aspects of Turkish familial hemophagocytic lymphohistiocytosis patients with Perforin mutations. Leuk Res. 2008;32(6):972-975. 15. Gurgey A, Aytac S, Balta G, Oguz KK, Gumruk F. Central nervous system involvment in Turkish children with primary hemophagocytic lymphohistiocytosis. J Child Neurol. 2008;23(11):1293-1299. 16. Gokce M, Balta G, Unal S, Oguz KK, Cetin M, Gumruk F. Spinal cord involvement in a child with familial hemophagocytic lymphohistiocytosis. J Pediatr Neurosci. 2012;7(3):194-196. 17. Zur Stadt U, Beutel K, Kolberg S, et al. Mutation spectrum in children with primary hemophagocytic lymphohistiocytosis: molecular and functional analyses of PRF1, UNC13D, STX11, and RAB27A. Hum Mutat. 2006;27(1):62-68. 18. Sieni E, Cetica V, Santoro A, et al. Genotype-phenotype study of familial haemophagocytic lymphohistiocytosis type

3. J Med Genet. 2011;48(5):343-352. 19. Rudd E, Göransdotter Ericson K, Zheng C, et al. Spectrum and clinical implications of syntaxin 11 gene mutations in familial haemophagocytic lymphohistiocytosis: association with disease-free remissions and haematopoietic malignancies. J Med Genet. 2006;43(4):e14. 20. Trottestam H, Berglöf E, Horne A, et al. Risk factors for early death in children with haemophagocytic lymphohistiocytosis. Acta Paediatr. 2012;101(3):313-318. 21. Filipovich AH. The expanding spectrum of hemophagocytic lymphohistiocytosis. Curr Opin Allergy Clin Immunol. 2011; 11(6):512-516. 22. leri T, Azık F, Uysal Z, Ertem M, Kulo lu Z, Gözda o lu S. Severe acute hepatic failure as an initial manifestation of hemophagocyticlymphohistiocytosis. Ankara Üniversitesi Tıp Fakültesi Mecmuası. 2009; 62:119-123. 23. Rooms L, Fitzgerald N, McClain K.L. Hemophagocytic lymphohistiocytosis masquerading as child abuse: presentation of three cases and review of central nervous system findings in hemophagocytic lymphohistiocytosis. Pediatrics. 2003;111(5 Pt 1):e636-640. 24. Trutzo LC, Lin DD, Hartung H, et al. A neurologic presentation of familial hemophagocytic lymphohistiocytosis which mimicked septic emboli to the brain. J Child Neurol. 2007;22(7):863-868. 25. Ouachée-Chardin M, Elie C, de Saint Basile G, et al. Hematopoietic stem cell transplantation in hemophagocytic lymphohistiocytosis: a single-center report of 48 patients. Pediatrics. 2006;117(4):e743-750. 26. Yoon HS, Im HJ, Moon HN, et al. The outcome of hematopoietic stem cell transplantation in Korean children with hemophagocytic lymphohistiocytosis. Pediatr Transplant. 2010;14(6):735-740.

haematologica | 2018; 103(2)

ARTICLE

Myelodysplastic Syndrome

Outcome after relapse of myelodysplastic syndrome and secondary acute myeloid leukemia following allogeneic stem cell transplantation: a retrospective registry analysis on 698 patients by the Chronic Malignancies Working Party of the European Society of Blood and Marrow Transplantation Christoph Schmid,1* Liesbeth C. de Wreede,2,3* Anja van Biezen,2 Jürgen Finke,4 Gerhard Ehninger,5 Arnold Ganser,6 Liisa Volin,7 Dietger Niederwieser,8 Dietrich Beelen,9 Paolo Alessandrino,10 Lothar Kanz,11 Michael Schleuning,12 Jakob Passweg,13 Hendrik Veelken,14 Johan Maertens,15 Jan J. Cornelissen,16 Didier Blaise,17 Martin Gramatzki,18 Noel Milpied,19 Ibrahim Yakoub-Agha,20 Ghulam Mufti,21 Montserrat Rovira,22 Renate Arnold,23 Theo de Witte,24 Marie Robin25 and Nikolaus Kröger26

Department of Hematology and Oncology, Klinikum Augsburg, University of Munich, Augsburg, Germany; 2Department of Medical Statistics & Bioinformatics, Leiden University Medical Center, the Netherlands; 3DKMS, German Bone Marrow Donor Center, Germany; 4Department of Medicine 1, Hematology and Oncology, University of Freiburg, Germany; 5Medizinische Klinik und Poliklinik I, Universitaets-Klinikum Dresden, Germany; 6Department of Hematology, Hemostasis, Oncology and Stem Cell Transplantation, Hannover Medical School, Germany; 7Stem Cell Transplantation Unit, HUCH Comprehensive Cancer Center, Helsinki, Finland; 8Division of Hematology, Oncology and Hemostaseology, University Hospital Leipzig, Germany; 9Department of Bone Marrow Transplantation, University Hospital, Essen, Germany; 10Clinica Ematologica, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy; 11Medizinische Klinik II, Universität Tübingen, Germany; 12KMT Zentrum, Deutsche Klinik für Diagnostik, Wiesbaden, Germany; 13Department of Hematology, University Hospital, Basel, Switzerland; 14BMT Center Leiden, Leiden University Hospital, the Netherlands; 15 Department of Hematology, University Hospital Gasthuisberg, Leuven, Belgium; 16 Erasmus MC Cancer Institute, University Medical Center Rotterdam, the Netherlands; 17 Centre de Recherche en Cancérologie de Marseille, Institut Paoli Calmettes, Marseille, France; 18Division of Stem Cell Transplantation and Immunotherapy, University Hospital Schleswig-Holstein Campus, Kiel, Germany; 19CHU Bordeaux, Hôpital Haut-Leveque, Pessac, France; 20Hôpital Huriez, CHRU, Lille, France; 21Department of Hematological Medicine, GKT School of Medicine, London, UK; 22Institute of Hematology & Oncology, Hospital Clinic, Barcelona, Spain; 23Medizinische Klinik m. S. Hämatologie/Onkologie, Charité Universitätsmedizin Berlin, Germany; 24Department of Tumor Immunology, Radboud University Medical Center, Nijmegen, the Netherlands; 25Department of Hematology – BMT, Hôspital St. Louis, Paris, France and 26Department of Stem Cell Transplantation, University Hospital Eppendorf, Hamburg, Germany

Ferrata Storti Foundation

Haematologica 2018 Volume 103(2):237-245

*CS and LdW contributed equally to this manuscript

Correspondence: christoph.schmid@klinikum-augsburg.de

Received: March 13, 2017. Accepted: October 30, 2017. Pre-published: November 3, 2017. doi:10.3324/haematol.2017.168716

ABSTRACT

o standard exists for the treatment of myelodysplastic syndrome relapsing after allogeneic stem cell transplantation. We performed a retrospective registry analysis of outcomes and risk factors in 698 patients, treated with different strategies. The median overall survival from relapse was 4.7 months (95% confidence interval: 4.1-5.3) and the 2-year survival rate was 17.7% (95% confidence interval: 14.821.2%). Shorter remission after transplantation (P<0.001), advanced disease (P=0.001), older age (P=0.007), unrelated donor (P=0.008) and acute graft-versus-host disease before relapse (P<0.001) adversely influenced survival. At 6 months from relapse, patients had received no cellular treatment, (i.e. palliative chemotherapy or best supportive care, n=375), donor lymphocyte infusion (n=213), or a second transplant (n=110). Treatment groups were analyzed separately because of imbalanced characteristics and difficulties in retrospectively evaluating the reason for individual treatments. Of the patients who did not receive any cellular therapy, 109 were alive at 6 months after relapse, achieving a median haematologica | 2018; 103(2)

Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/237 ©2018 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.

237

C. Schmid et al.

overall survival from this landmark of 8.9 months (95% confidence interval: 5.1-12.6). Their 2-year survival rate was 29.7%. Recipients of donor lymphocytes achieved a median survival from first infusion of 6.0 months (95% confidence interval: 3.7-8.3) with a 2-year survival rate of 27.6%. Longer remission after first transplantation (P<0.001) and younger age (P=0.009) predicted better outcome. Among recipients of a second transplant, the median survival from second transplantation was 4.2 months (95% confidence interval: 2.5-5.9), and their 2-year survival rate was 17.0%. Longer remission after first transplantation (P<0.001), complete remission at second transplant (P=0.008), no prior chronic graft-versus-host disease (P<0.001) and change to a new donor (P=0.04) predicted better outcome. The data enabled identification of patients benefiting from donor lymphocyte infusion and second transplantation, and may serve as a baseline for prospective trials.

Introduction Relapse of the underlying disease is a major drawback of allogeneic hematopoietic stem cell transplantation (HSCT) for myelodysplastic syndrome (MDS) and secondary acute myeloid leukemia (sAML) evolved from MDS, in particular as a consequence of the increasing numbers of HSCT with reduced intensity conditioning.1,2 As in other diseases, there is no defined standard approach to the management of post-transplant relapse.3 Several studies addressing the outcome of post-transplant relapse in different myeloid diseases have included patients with MDS.4-6 However, a specified analysis for MDS was usually not performed, or included only limited numbers of patients. Hence, no largescale analysis of risk factors, different treatment strategies and outcomes of MDS relapse after HSCT is available. The Chronic Malignancies Working Party (CMWP) of the European Society for Blood and Marrow Transplantation (EBMT) performed a retrospective, registry-based analysis on adults with hematological relapse after allogeneic HSCT. Data are intended to serve as a baseline and comparison for future trials using innovative approaches.

Methods Inclusion criteria for patients from the CMWP registry comprised: (i) first allogeneic HSCT for MDS or sAML, using matched related, mismatched related or matched unrelated donors; (ii) age at HSCT ≥18 years; (iii) first hematological relapse after transplant (excluding decreasing chimerism or cytogenetic/molecular relapse); and (iv) reliable documentation about the management of posttransplant relapse. Patients had agreed to reporting data to national and international registries before transplantation. The study was approved by the ethical committee of the Medical Faculty, University of Essen. All procedures complied with the ethical standards of the responsible committees (institutional and national) and the revised version of the Helsinki Declaration of 1975. Based on the first treatment received during the first 6 months after relapse, three treatment groups were defined: (i) patients who did not receive any cellular therapy; (ii) patients who received donor lymphocyte infusion (DLI); and (iii) patients who underwent a second allogeneic HSCT (HSCT2). The 6-month cutoff was chosen, since >95% of HSCT2 and >98% of DLI reported in published studies were performed within the first 6 months after relapse. Hence, the cumulative use of these strategies, as well as the outcome of patients treated without cellular therapy, could be studied. Patients who both received DLI and underwent 238

a subsequent transplant proceeded to the DLI group, if HSCT2 was given >90 days after DLI, because 90 days were regarded as sufficient to evaluate the effect of DLI. Patients who received a second transplant <90 days after DLI entered the HSCT2 cohort, since HSCT was considered as the decisive intervention for the long-term outcome. Although being somewhat arbitrary, this classification enabled patients receiving both DLI and HSCT2 to be included in the analysis without uncontrolled bias. Conditioning,7 graft-versus-host disease (GvHD)8 and remission before HSCT9 were defined as described previously. As suggested5, the transfusion of peripheral blood stem cells or bone marrow was defined as DLI, if no prior conditioning and no prophylactic immunosuppression was given, whereas HSCT2 was defined as infusion of donor peripheral blood stem cells or bone marrow following a conditioning regimen and with prophylactic immunosuppression (refer to the Online Supplement for details).

Statistics Overall survival from relapse was the primary endpoint.10 Variables considered included characteristics of patients and their disease, donors, transplant procedure, and relapse (see the Online Supplement for details). Variables were compared among treatment groups, using the chisquare test for categorical variables and the Kruskal-Wallis test for continuous ones. In patients receiving DLI or HSCT2, cumulative incidence of relapse and non-relapse mortality were analyzed by competing risks models, with the starting time being the moment of DLI/HSCT2. In addition to the factors mentioned above, the characteristics of the DLI/HSCT2 were considered for risk analysis. Outcomes of subgroups were compared using a log-rank test. Cox proportional hazards regression models were used for multivariable analyses of factors for time-to-event endpoints. Variables were included if considered relevant based on the univariate analysis (P-value <0.2), or known to be so from the literature. Patients with missing predictor data were included in the analysis by assigning them to separate categories of the pertaining variables. Hazard ratios (HR) and 95% confidence intervals (95% CI) are reported. R Version 3.1.0, packages ‘survival’, ‘cmprsk’ and ‘mstate’11 and SPSS versions 18 and 23 (SPSS Inc. Chicago, IL, USA) were used.

Results Patients’ characteristics and overall outcome A total of 698 patients fulfilled the inclusion criteria (Table 1). The median interval between HSCT and hemahaematologica | 2018; 103(2)

MDS relapse after allogeneic transplantation

tological relapse was 6.3 months (range, 1-160.8). The median follow-up from relapse among survivors was 9.4 months (range, 0.7-119.8). The median overall survival from relapse of the entire cohort was 4.7 months (95% CI:

4.1-5.3 months). The overall survival rate was 27.6% (95% CI: 24.2-31.3%) at 1 year, 17.7% (95% CI: 14.821.2%) at 2 years and 11.4% (95% CI: 8.8-14.7%) at 4 years (Figure 1). Progression or another relapse of

Table 1. Baseline characteristics of 698 patients relapsing after allogeneic stem cell transplantation for myelodysplastic syndrome or secondary acute myeloid leukemia.

Entire cohort

Number of patients Year of HSCT, (range) Patients’ age at relapse (years) Patients’ sex (n)

698 (100%) 2003 (1994-2008) 52.2 (18.4-74.9) 328 (47%) 370 (53%) 6.5 (0.4-291.6) 47 (8%) 107 (19%) 418 (73%) 126 144 (21%) 304 (44%) 77 (11%)

Median (range) Female Male Time from diagnosis Median to HSCT (months) (range) Diagnosis at HSCT (n) RA/RARS RAEB sAML** Missing Stage at HSCT Untreated (n, %) CR Relapse/ progression Primary refractory 173 (25%) Donor (n, %) Matched family 398 (57%) Mismatched family 300 (43%) or unrelated Sex match donor/ Female in male 117 (17%) recipient (n, %) Other 571 (83) Missing 10 Conditioning Standard 413 (59%) (n, %) Reduced 285 (41%) T-cell depletion No 383 (56%) before HSCT (n, %) In vivo 199 (29%) Ex vivo 53 (8%) In vivo+ ex vivo 48 (7%) Missing 15 Stem cell source BM 176 (26%) at HSCT (n, %) PBSC 515 (75%) Missing 7 Acute GvHD grade 2-4 No 538 (79%) after first HSCT Yes 141 (21%) Missing 19 Chronic GvHD after first No 394 (73%) HSCT (n, %) Yes 143 (27%) Missing 161 Remission duration Median 6.3 after HSCT (months) (range) (1.0-160.8) (n, %) <6 months 329 (47%) 6-12 months 202 (29%) >12 months 167 (24%)

Patients receiving no cellular therapy

Patients given DLI ± prior chemotherapy as primary intervention

Patients given a second HSCT ± prior chemotherapy as primary intervention 110 (16%) 2002 (1994-2008) 48.7 (20.7-68.3) 48 (44%) 62 (56%) 5.5 (0.8-104.7) 4 (4%) 20 (22%) 66 (73%) 20 22 (20%) 49 (45%) 11 (10%)

Total

Alive without cell. therapy at 6 months from relapse*

375 (54%) 2003 (1994-2008) 53.3 (18.4-73.3) 164 (44%) 211 (56%) 7.0 (0.4-291.6) 24 (8%) 48 (16%) 231 (76%) 72 71 (19%) 171 (46%) 48 (13%)

109 (16%)

51.4 (21.3-72.3) 44 (40%) 65 (60%) 7.3 (1.0-291.6) 14 (17%) 15 (18%) 53 (65%) 27 29 (27%) 42 (39%) 15 (14%)

213 (31%) 2002 (1994-2008) 52.3 (18.7-74.9) 116 (55%) 97 (46%) 6.2 (0.8-143.2) 19 (11%) 39 (22%) 121 (68%) 34 51 (24%) 84 (39%) 18 (9%)

85 (23%) 190 (51%) 185 (49%)

23 (21) 59 (54%) 50 (46%)

60 (28%) 135 (63%) 78 (37%)

28 (26%) 73 (66%) 373 (34%)

72(20%) 296 (80%) 7 216 (58) 159 (42%) 213 (58%) 106 (29%) 27 (7%) 20 (6%) 9 92 (25%) 276 (75%) 7 266 (73%) 100 (27%) 9 204 (71%) 84 (29%) 87 5.8 (1.0-160.8) 193 (52%) 105 (28%) 77 (21%)

22(21%) 84 (79%) 3 63 (58%) 46 (42%) 58 (54%) 33 (31%) 10 (9%) 6 (6%) 2 36 (34%) 71 (66%) 2 82 (77%) 24 (23%) 3 60 (64%) 34 (36%) 15 9.6 (1.3-160.8) 28 (26%) 40 (37%) 41 (38%)

27 (13%) 183 (87%) 3 128 (60%) 85 (40%) 106 (51%) 66 (31%) 19 (9%) 19 (9%) 3 57 (27%) 156 (73%) 182 (88%) 25 (12%) 6 143 (77%) 42 (23%) 28 6.6 (1.0-148.3) 92 (43%) 63 (30%) 58 (27%)

18 (16%) 92 (84%) 69 (63%) 41 (37%) 64 (60%) 27 (25%) 7 (7%) 9 (8%) 3 27 (25%) 83 (76%) 90 (85%) 16 (15%) 4 47 (73%) 17 (27%) 46 7.1 (1.7-134.9) 44 (40%) 34 (31%) 32 (29%)

0.005 0.04 0.03 0.001 0.12

0.538

0.14

0.19

0.76 0.70

0.29

0.05

0.06

0.009

HSCT: hematopoietic stem cell transplantation; RA: refractory anemia; RARS: refractory anemia with ring sideroblasts; RAEB: refractory anemia with excess of blasts; sAML: secondary acute myeloid leukemia. CR: complete remission. GvHD: graft-versus-host disease. *The ‘no cellular therapy’ group contained patients who never received cellular therapy due to early death, as well as patients whose follow-up was not long enough to ascertain if they had received cellular therapy or not.To avoid bias, only those patients were included into the comparison among treatment cohorts, of whom it was certain that they survived for ≥6 months from relapse and had not received any cellular treatment.** Patients initially classified as having RAEB-T have been classified as having sAML, using the current WHO definition of AML (≥ 20% blasts in the bone marrow).

haematologica | 2018; 103(2)

239

C. Schmid et al.

MDS/sAML was the leading cause of death (82.1% of deaths). The remaining deaths were from causes related to HSCT or cellular therapy (14.9%), secondary malignancies (0.7%) or other causes (2.2%). A risk factor analysis for outcome after relapse was performed. Results from univariate analysis are shown in Online supplementary Table S1. Table 2 shows the results of the multivariate analysis for overall survival, based on variables available at the time of the relapse. In the multivariate model, a longer remission after HSCT (>12 months, HR 0.4, and 6-12 months HR 0.6, P<0.001), an earlier stage of MDS at time of HSCT (refractory anemia with excess blasts and sAML versus refractory anemia/refractory anemia with ring sideroblasts; HR 1.6 and 2.0, P=0.001), younger age at relapse (HR 1.01 per year, P=0.005), donor type (unrelated donor versus HLA-identical sibling, HR 1.3, P=0.005) and no history of acute GvHD before relapse (HR 0.6, P<0.001) were associated with better overall survival from relapse.

analysis, accounting also for loss from follow-up, was performed. All patients started in the cellular-treatment free group, and could subsequently proceed to one of the other groups. DLI, HSCT2 and death without any cellular therapy (whichever occurred first) were considered as competing events (Figure 2). Patientsâ&#x20AC;&#x2122; characteristics such as age or duration of remission after HSCT were significantly different among the three treatment groups (Table 1). Furthermore, treatment

Outcome according to the treatment applied In the first 6 months from relapse after HSCT, 213 patients were reported to have received DLI, and 110 to have undergone HSCT2. The other 375 patients had not received any cellular treatment: 109 of these were still alive and in follow-up 6 months after their relapse. Since patients who died early after relapse or whose follow-up was short did not have enough time for the transition to one of the two other groups, the comparison among treatment groups was only based on patients still in the group not having received any cellular therapy at 6 months after relapse, to avoid bias due to the high early mortality that by definition had to take place in this group. To assess the time-dependent probability of receiving a cellular intervention in the first 6 months after relapse, or remaining alive cellular-treatment free, a competing risks

Figure 1. Overall survival from relapse in 698 patients. Overall survival (OS) from relapse of the entire cohort (gray area denotes 95% confidence interval, CI, over time) The median overall survival was 4.7 months (95% CI: 4.1-5.3 months).

Table 2. Multivariate analysis of risk factors for overall survival from relapse in 698 patients.

HR for death Donor type HLA-identical sibling 1 Unrelated/mismatched 1.3 relative MDS subtype at HSCT* RA/RARS 1 RAEB 1.6 sAML 2.0 Acute GvHD before relapse* No 1 Yes 1.6 Age at relapse (as continuous variable, 1.010 per year)** Remission after HSCT < 6 months 1 6-12- months 0.6 > 12 months 0.4

95% CI for HR lower upper

P 0.005

1.1

1.5 0.001

1.0 1.54

2.48 2.9

1.3

1.92

1.003

1.016

0.5 0.3

0.7 0.5

0.035 <0.001 <0.001

0.005

<0.001 <0.001 <0.001

HR: hazard ratio; CI: confidence interval; MDS: myelodysplastic syndrome; HSCT: hematopoietic stem cell transplantation; RA: refractory anemia; RARS: refractory anemia with ring sideroblasts; RAEB: refractory anemia with excess of blasts; sAML: secondary acute myeloid leukemia; GvHD: graft-versus-host disease.*Patients with missing data were retained in the analysis by assigning them to a separate category (hazard ratios not shown). **Impact of an age difference at relapse of (e.g.) 10 years translates into a hazard ratio of 1.10.

240

Figure 2. Cumulative incidence of treatments applied during the first 6 months. The plots are stacked: the distance between two lines (and, for the uppermost curve, the distance from the curve to 100%) indicates the cumulative incidence as a function of time. At 6 months after relapse, the cumulative probability of having received a DLI (bottom group) was 31% (95% CI: 28-35%) and that of having undergone HCT2 (second group from the bottom) was 17% (95% CI: 14-19%). Thirty-five percent had died without having received DLI or HSCT2 (second group from top), whereas 18% (95% CI: 15-21%) of patients were still in the cellular treatment-free group (uppermost group).

haematologica | 2018; 103(2)

MDS relapse after allogeneic transplantation

decisions in the reporting centers had not been based on a general strategy, but had been made individually for each patient. Therefore, the reason for assigning a given patient to one of the treatment modalities could not be evaluated retrospectively at a reliable level. The three treatment cohorts were, therefore, evaluated separately, and no comparison of outcome was performed.

Patients not receiving any cellular treatment (n=375) The treatment approaches applied among patients not receiving any cellular intervention ranged from palliation only to mild and intensive chemotherapy. However, detailed information on choice and dosage of the drugs used was not available. At last follow-up, 266 patients had died without having received DLI or HSCT2: reported causes of death were MDS/sAML in 92%, and HSCTrelated events in 7%. For the 109 patients alive at 6 months after relapse, the median overall survival from this landmark was 8.9 months (95% CI: 5.1-12.6). The 2year overall survival rate was 29.7% (95% CI: 20.139.3%). Since the vast majority of DLI and HSCT2 had been given within the first 3 months after relapse, another landmark analysis was performed, including 221 patients who had been alive without any cellular therapy by day 90. The median overall survival from this landmark was 6.1 months and the 2-year overall survival rate was 23.8%

Patients receiving a second stem cell transplant (n=110) The characteristics of second transplants are shown in Table 3. Only 9% of patients underwent HSCT2 while in complete remission, whereas 90% had active disease. The median interval from relapse to HSCT2 was 1.7 months (range, 0.2-6.0). Seventy-six percent received the HSCT2 from the original donor, whereas a new donor was used in 24%. Following HSCT2, 45 patients (46% of informative cases) achieved complete remission; 16 of them developed another relapse at a median of 4.6 months (range, 1.8-37.8) after HSCT2. Acute GvHD grade II-IV and chronic GvHD developed in 24.8% and 34.6%, respectively, of informative patients. The median follow-up was 11.0 months among survivors. The median overall survival from HSCT2 was 4.2 months (95% CI: 2.5-5.9), while the overall survival rates at 1, 2 and 4 years were 22.3% (95% CI: 13.9-30.7%), 17.0% (95% CI: 10.7-27.1%) and 12.4% (95% CI: 5.119.7%), respectively. (Figure 3A). The cumulative incidence of relapse/progression was 35% (95% CI: 26-45%) at both 1 and 2 years, whereas the cumulative incidence of non-relapse mortality was 45% (95% CI: 35-55%) at 1 year and 49.3% (95% CI: 39-59%) at 2 years after HSCT2. A risk factor analysis for overall survival from HSCT2 was performed, including variables known at the time of HSCT2. Results of the univariate analysis are provided in Online Supplementary Table S2. Among other factors, a history of DLI given prior to HSCT2 did not influence the overall outcome. In a multivariate Cox model (Table 4A) a sibling donor for HSCT1, no history of chronic GvHD after HSCT1, a longer remission after HSCT1, and HSCT2 in complete remission were strongly associated with better overall survival after HSCT2. With respect to disease stage, the median overall survival after HSCT2 in complete remission was 37.8 months, as compared to only 2.9 months after HSCT2 in active disease (univariate comparison). There was also an advantage for those patients haematologica | 2018; 103(2)

undergoing HSCT2 from a different donor (P=0.044, HR 0.562, 95% CI: 0.321-0.984). The role of remission duration (P=0.002) and stage (P=0.022; univariate KaplanMeier estimates) is illustrated in Figure 3B,C.

Patients receiving donor lymphocyte infusion (n=213) The median interval from relapse to first DLI was 21 days (range, 0-170). The initial cell dose was 1x107 CD3+ cells/kg (range, 0.3-187). Of the informative patients, Table 3. Characteristics and early outcome of second transplant in 110 patients.

Characteristics Interval between HSCT1 and HSCT2 (months)

Median (range) Interval between relapse and HSCT2 (months) Median (range) DLI given for relapse <90 days before HSCT2 N. (%) Days between DLI and HSCT2 (median, range) Stage at HSCT2, n (%) CR Active disease Missing Donor for HSCT1 HLA identical family Unrelated/mismatched family Donor for HSCT2 HLA identical family Unrelated/mismatched family Missing Donor change* from HSCT1 to HSCT2, n (%) Same donor for HSCT2 New donor for HSCT2 Missing Conditioning for HSCT2, n (%) Standard Reduced Missing TBI for conditioning before HSCT2 Yes No Missing Source of stem cells for HSCT2 BM PB9 BM+PB Cord blood Missing Outcome Engraftment, n (%) Yes Yes, but secondary graft failure No Missing Time to neutrophil engraftment, days Median (range) Time to platelet engraftment, days Median (range) Response after HSCT2, n (%) Early death** CR No CR Unknown

N (range or %) 9.4 (2.3-135.8) 1.7 (0.2-6.0) 12 (11%) 49.5 (0-89) 9 (10%) 85 (90%) 16 73 (66%) 37 (34%) 70 (64%) 39 (36%) 1 83 (76%) 26 (24%) 1 46 (43%) 61 (57%) 3 24 (23%) 80 (77%) 6 8 (7%) 8 (91%) 1 (1%) 1 (1%) 2 80 (79%) 2 (2%) 19 (19%) 9 13 (0-102) 20 (7-263) 16 (16%) 45 (46%) 37 (38%) 12

HSCT: hematopoietic stem cell transplantation; DLI: donor lymphocyte infusion; BM: bone marrow; PB: peripheral blood; CR: complete remission. *Either from a matched family to an unrelated donor or from one unrelated donor to another. ** Early death was defined as death <1 month from HSCT2 and without re-occurrence of myelodysplastic syndrome or secondary acute myeloid leukemia.

241

C. Schmid et al.

68.6%, 17.1% and 14.3% received one, two and three infusions, respectively. The median follow-up of the 50 patients alive at last contact was 18.2 months (range, 0.03105.1). Following DLI, acute GvHD grade 2-4 was observed in 14 patients (26.4% of informative cases). Limited and extensive chronic GvHD were reported in 15 and 13 patients, respectively. While response rate to DLI was not reported for a considerable number of patients, the median overall survival from first DLI was 6.0±1.2 months (95% CI: 3.7-8.3); overall survival rates at 1, 2 and 4 years were 36.1±3.5%, 27.6±3.3% and 17.0±3.2%, respectively

(Figure 4A). Reported causes of death were MDS/sAML (81.7%), HSCT related causes (12.4%), secondary malignancies (2.6%), and other causes (3.3%). Results of the univariate analysis of risk factors for overall survival from first DLI are provided in Online Supplementary Table S3. A multivariate Cox model revealed a longer remission after HSCT (most significant), younger age, and male sex as significant protective parameters (Table 4B). Some patients achieved long-term survival after receiving a second transplant following the failure of DLI. Figure 4B illustrates the role of remission duration (P<0.001, univariate Kaplan-

Figure 3. Overall survival after second transplant. (A) Within the entire cohort of 110 patients (gray area denotes 95% confidence interval, CI, over time; 2-year overall survival: 17.0, 95% CI: 10.7-27.1%). (B) As of remission duration after first transplantation, (>12 months, solid line, 2-year overall survival 36.6%, 95% CI: 21.961.0 ; 6-12 months, dashed line, 2-year overall survival 14.9%, 95% CI: 6.3-35.7; <6 months, dotted line, 2-year overall survival 5.9%, 95% CI: 15.7-22.3) P=0.002. (C) As of remission status at time of second transplant (complete remission, solid line, 2-year overall survival 59.3%, 95% CI: 32.2-100%; active disease, dotted line, 2-year OS 11.1%, 95% CI: 5.5-22.4%) P=0.022.

Table 4A. Multivariate analysis of risk factors for overall survival from second transplant in 110 patients.

HR for death Donor type at HSCT1 HLA-identical sibling unrelated/mismatched Cell source at HSCT1 Bone marrow Peripheral blood Chronic GvHD before relapse* No Yes Remission after HSCT < 6 months < 6 months vs. 6-12 months < 6 months vs. > 12 months Stage at second HSCT* Complete remission Active disease Donor change for HSCT2 * Same donor for HSCT2 Donor change for HSCT2

95% CI for HR lower upper

P 0.018

1 1.8

1.1

2.9 0.185

1 1.5

0.8

2.5 <.001

1 4.4

2.2

8.9 <0.001

1 0.5 0.3

0.3 0.2

0.8 0.6

0.003 <0.001

1 3.8

1.49

10.4

0.008 0.044

1 0.6

0.3

1.0

HR: hazard ratio; CI: confidence interval; HSCT: hematopoietic stem cell transplantation; GvHD: graft-versus-host disease.

242

Table 4B. Multivariate analysis of risk factors for overall survival after first therapeutic donor lymphocyte infusion in 213 patients.

HR for death Patients’ sex Male 1 Female 1.5 Stage at HSCT1 Untreated 1 Complete remission 1.2 Relapsed/progressive disease 1.3 Donor type at HSCT1 HLA-identical sibling 1 Unrelated/mismatched 1.4 Chronic GvHD before relapse* No 1 Yes 1.0 Age at relapse (as continuous variable, per year) 1.017 Remission after HSCT < 6 months 1 6-12 months 0.7 > 12 months 0.4

95% CI for HR lower upper

P 0.014

1.1

2.2 0.521

0.9 0.8

1.9 1.9

1.0

1.9

0.6

1.6

1.004

1.030

0.313 0.303 0.080

0.904 0.009 <0.001

0.5 0.2

0.1.0 0.6

0.048 <.001

HR: hazard ratio; CI: confidence interval; HSCT: hematopoietic stem cell transplantation; GvHD: graft-versus-host disease.*Patients with missing data were retained in the analysis by assigning them to separate categories (hazard ratios not shown).

haematologica | 2018; 103(2)

MDS relapse after allogeneic transplantation

Meier estimate). If post-transplant remission exceeded 1 year, the median overall survival was 25.1 months (95% CI: 8.4-41.8) and the 2-year overall survival rate was 51.3% (95% CI: 39.1-67.2%).

Discussion In the largest group of patients relapsing after allogeneic HSCT for MDS and sAML analyzed so far, less advanced stage of MDS at transplant (refractory anemia/refractory anemia with ring sideroblasts and refractory anemia with excess blasts versus sAML), no history of acute GvHD after HSCT, and a longer remission after HSCT were associated with better overall survival after relapse. At 2 and 4 years from relapse, 6% and 4% of patients, respectively, were alive without having received a cellular treatment. During the first 6 months after relapse, 31% of patients were selected to receive DLI and 17% of patients received a second HSCT. The 2-year overall survival rates from the intervention in these selected subgroups were 28% for DLI recipients, and 17% for patients receiving HSCT2. While remission duration after HSCT1 was the most relevant prognostic parameter both after DLI and after HSCT2, no history of chronic GvHD after HSCT1, an HLA-identical family donor for HSCT1, and controlled disease at the time of HSCT2 were additional relevant factors for survival after a second transplant. Switching to an alternative donor for HSCT2 was associated with a better outcome. Despite the large number of patients included in this analysis, the nature of a retrospective registry study implies several limitations. Most importantly, cytogenetic data were rather incomplete, thereby precluding calculation of the revised International Prognostic Scoring System score and the analysis of outcome and efficacy of different treatments in biologically defined risk groups. On the other hand, cytogenetics has not been found to play a major role in outcomes following post-transplant relapse either in AML12 or MDS (with the exception of very poor risk cytogenetic characteristics).13 Second, the analysis is based on patients transplanted in the past, for whom sufficient data of reasonable quality to perform such a detailed analysis were available. To study a possible change in outcome of post-transplant MDS relapse

over time, we analyzed overall survival after relapse in a more recent cohort of MDS patients, also derived from the EBMT registry (transplantation years 2009-2012). The outcomes of this cohort were almost identical to those of our current study [median overall survival 5.0 months, 2-year overall survival rate 19% (95% CI 1621%), data not shown in detail]. Similarly, the 2-year overall survival rate following post-HSCT relapse was 16% in a more recent smaller study,13 which compares very well with our observation. We, therefore, believe that our data are still valid for current patients, also illustrating the limited progress made in recent years in the treatment of MDS relapsing after allogeneic HSCT, and the urgent need for new concepts. As discussed below, the broader use of hypomethylating agents might be a way to improve outcomes in the future. However, so far this has not been demonstrated in the various registry analyses. The lack of precise data on disease status at the time of relapse prevented distinguishing between relapse as MDS versus sAML. This is another drawback of the analysis, since bone marrow infiltration by leukemic blasts at the time of relapse was a relevant factor for overall survival in a recent analysis on patients relapsing after reduced intensity conditioning HSCT for de novo AML.14 The percentage of bone marrow blasts at relapse might also have influenced the treatment approach after relapse. The reason why no cellular therapy, DLI or a second transplant was preferred in a given patient could not be determined retrospectively. Given the differences in patientsâ&#x20AC;&#x2122; characteristics among the three treatment groups, and because the group not given cellular therapy was a heterogeneous mixture of patients who died early, patients with a short follow-up and patients alive and truly untreated with DLI or HSCT2, we decided to limit the overall analysis of risk factors to variables that were known at the time of relapse, and performed separate studies for the DLI and HSCT2 subgroups. Hence, general conclusions for selecting treatment strategies cannot be drawn on the basis of this analysis. As shown in many studies on relapse after allogeneic HSCT,5;12;14-17 remission duration after HSCT was the most important variable for outcome, irrespective of the applied treatment strategy. Patients with a history of acute GvHD after HSCT1 had an inferior outcome, a finding that has

Figure 4. Overall survival after first therapeutic donor lymphocyte infusion. (A) Within the entire cohort of 213 patients, 2year overall survival was 27.6%, 95% confidence interval (CI): 21.1-34.1.0%. Grey area denotes 95% CI, over time (B) As of remission duration after HSCT1 (>12 months, solid line, 2-year overall survival 51.3%, 95% CI: 39.1-67.2%; 6-12 months, dashed line, 2-year overall survival 30.8%, 95% CI: 20.2-46.8%; <6 months, dotted line, 2-year overall survival 11.0%, 95% CI: 5.9-20.4%) P<0.001.

haematologica | 2018; 103(2)

243

C. Schmid et al.

similarly been reported after relapse from reducing intensity conditioning HSCT for de novo AML14 and acute lymphocytic leukemia,17 and in a recent French study on relapsed MDS.13 However, in our cohort, 23% of patients who did not receive any cellular therapy for post-transplant relapse had developed acute GvHD of grade II-IV after HSCT1, in contrast to only 12% and 15% among patients who received DLI or HSCT2, respectively (P=0.05). Hence, a history of acute GvHD might not necessarily be a risk factor as such, but may have contributed to the decision not to consider DLI or HSCT2 in a given patient, or may have been associated with an inferior performance at the time of relapse, leading to less intensive treatment. Similarly, the history of a related donor transplant may have influenced the application of a donor cell based strategy (i.e. DLI or second transplant), and younger age may have been a criterion for offering HSCT2. Precise information on 110 patients undergoing HSCT2 allowed for a detailed analysis of this strategy. The role of disease control before HSCT2 was a striking observation. Although only 10% of patients received HSCT2 in complete remission, stage at HSCT2 was a significant factor for outcome in multivariate analysis, and median overall survival after HSCT2 in complete remission was 37 months, as compared to only 2 months after HSCT2 in active disease (Figure 3B; the large confidence intervals underscoring the need for confirmatory studies). Nevertheless, selection of patients during and after chemotherapy precludes firm recommendations. Similar results have been reported recently in a large German study on HSCT2 for acute leukemia in the related and unrelated donor setting,18 and in an EBMT analysis on DLI for post-transplant relapse of de novo AML.10 Unfortunately, data on remission status at time of DLI in our study were not sufficient to reproduce these findings among DLI recipients. Switching to a different donor for HSCT2 showed a limited, although statistically significant, advantage for overall survival (HR: 0.562, 95% CI: 0.321-0.984 in the multivariate model). This observation is in line with those of several studies addressing this issue in acute leukemia,16,18 indicating that a change of donor is definitely not disadvantageous, and seems to offer a slight improvement in certain subgroups. Hence, our data add to the growing evidence, that changing to another donor is a justifiable option for HSCT2. Donor switching might be more promising among patients receiving HSCT2 for controlled disease, whereas in patients with a short post-transplant remission or in uncontrolled disease, the aggressiveness of the underlying malignancy will most likely overwhelm a putatively improved graft-versus-leukemia reaction of a new donor. Change to a haploidentical family donor might be another option, given the more rapid availability and the greater HLA disparity.19

References 1. Lim Z, Brand R, Martino R, et al. Allogeneic hematopoietic stem-cell transplantation for patients 50 years or older with myelodysplastic syndromes or secondary acute myeloid leukemia. J Clin Oncol. 2010;28(3):405-411. 2. McClune BL, Weisdorf DJ, Pedersen TL, et al. Effect of age on outcome of reduced-

244

Finally, patients without a history of chronic GvHD after HSCT1 had a better outcome after HSCT2, suggesting that in these patients, the graft-versus-leukemia effect might not yet have been exploited before post-transplant relapse. In summary, we provide data on a large cohort of consecutively reported patients with relapsed MDS and sAML after allogeneic HSCT, discussing the difficulties and limitations of such a retrospective registry analysis. As in a recent French study, which showed comparable overall results in a smaller cohort,13 relapse or progression was by far the leading cause of death, underscoring the need for innovative strategies. Without a graft-versus-leukemiabased intervention (i.e., DLI or HSCT2), only a few patients survive more than 2 years after relapse. In contrast, DLI or second HSCT showed certain, albeit still limited, efficacy in selected patients, such as patients in whom the interval between HSCT and relapse was long, or patients who responded to chemotherapy. In the EBMT analysis on AML relapse after reduced intensity conditioning HSCT, a donor cell-based intervention was shown to be mandatory for long-term remission even in these positively selected patients.14 Similar findings emerged from the French study on post-HSCT MDS relapse mentioned above. In contrast, patients in whom the interval between HSCT and relapse is short or patients with overt AML and high blast counts have a grim prognosis and are not likely to benefit from the traditional interventions after relapse. Our results may serve as a baseline to which new approaches can be compared, as they give a large-scalebased estimate of the results to be expected after use of each approach in the treatment of MDS relapse after allogeneic HSCT. At present, hypomethylating agents (azacytidine, decitabine) alone or in combination with DLI seem to be among the most promising compounds for the treatment of post-transplant relapse in myeloid malignancies,20,21 because of both their direct antileukemic efficacy and their immunomodulatory capacity.22-26 Checkpoint inhibitors might be an option for the future.27 Strategies for prophylactic28-30 or preemptive31-33 treatment in highrisk patients are promising alternatives to avoid overt hematologic relapse, while targeting molecular aberrations such as Flt3-internal tandem duplication, or inhibition of histone deacetylase34 and prophylactic DLI are promising approaches to post-transplant maintenance. Acknowledgment Following EBMT publication rules, co-authorship was offered to centers contributing the highest number of patients. The authors also highly appreciate the contribution by many physicians and data managers throughout the EBMT, who made this analysis possible. A list of contributing centers and responsible physicians is provided in the Online Supplement.

intensity hematopoietic cell transplantation for older patients with acute myeloid leukemia in first complete remission or with myelodysplastic syndrome. J Clin Oncol. 2010;28(11):1878-1887. 3. Platzbecker U. Who benefits from allogeneic transplantation for myelodysplastic syndromes?: new insights. Hematology Am Soc Hematol Educ Program. 2013;522-528. 4. Kolb HJ, Schattenberg A, Goldman JM, et al.

Graft-versus-leukemia effect of donor lymphocyte transfusions in marrow grafted patients. Blood. 1995;86(5):2041-2050. 5. Levine JE, Braun T, Penza SL, et al. Prospective trial of chemotherapy and donor leukocyte infusions for relapse of advanced myeloid malignancies after allogeneic stem-cell transplantation. J Clin Oncol. 2002;20(2):405-412. 6. Ruutu T, de Wreede LC, van Biezen A, et al.

haematologica | 2018; 103(2)

MDS relapse after allogeneic transplantation

10. 11.

12.

13.

14.

15.

16.

Second allogeneic transplantation for relapse of malignant disease: retrospective analysis of outcome and predictive factors by the EBMT. Bone Marrow Transplant. 2015;50(12):1542-1550. Gratwohl A, Carreras E. Principles of Conditioning. In: Apperley JF, Carreras E, Gluckman E, Masszi T, eds. Haematopoietic Stem Cell Transplant. 2012:122-137. Apperley JF, Masszi T. Graft-versus-host disease. In: Apperley JF, Carreras E, Gluckman E, Masszi T, eds. Haematopoietic Stem Cell Transplantation. 2012:217-247. Cheson B, Benet JM, Kopecky KJ et al. Revised Recommendations of the International Working Group for Diagnostics, Standardization of Response Criteria, Treatment Outcomes, and Reporting Standards for Therapeutic Trials in Acute Myeloid Leukemia. J Clin Oncol. 2003;21(24):4642-4649. Kaplan E, Meier P. Non parametric estimation from incomplete observatinons. J Am Stat Assoc. 1958;53:457-418. de Wreede, LC, Fiocco, M, Putter, H. "mstate": An R package for the analysis of competing risk and multistate models. J Stat Software 2011;38(7):1-30 Schmid C, Labopin M, Nagler A, et al. Donor lymphocyte infusion in the treatment of first hematological relapse after allogeneic stem-cell transplantation in adults with acute myeloid leukemia: a retrospective risk factors analysis and comparison with other strategies by the EBMT Acute Leukemia Working Party. J Clin Oncol. 2007;25(31): 4938-4945. Romain GuiĂ¨ze M, Damaj, G, Pereira, B, et al. Management of myelodysplastic syndrome relapsing after allogeneic hematopoietic stem cell transplantation: a study by the French society of bone marrow transplantation and cell therapies. Biol Blood Marrow Transplant. 2016;22(2):240-247. Schmid C, Labopin M, Nagler A, et al. Treatment, risk factors, and outcome of adults with relapsed AML after reduced intensity conditioning for allogeneic stem cell transplantation. Blood. 2012;119(6): 1599-1606. Bosi A, Laszlo D, Labopin M, et al. Second allogeneic bone marrow transplantation in acute leukemia: results of a survey by the European Cooperative Group for Blood and Marrow Transplantation. J Clin Oncol. 2001;19(16):3675-3684. Eapen M, Giralt SA, Horowitz MM, et al.

haematologica | 2018; 103(2)

17.

18.

19.

20.

21.

22.

23.

24.

25.

Second transplant for acute and chronic leukemia relapsing after first HLA-identical sibling transplant. Bone Marrow Transplant. 2004;34(8):721-727. Spyridonidis A, Labopin M, Schmid C, et al. Outcomes and prognostic factors of adults with acute lymphoblastic leukemia who relapse after allogeneic hematopoietic cell transplantation. An analysis on behalf of the Acute Leukemia Working Party of EBMT. Leukemia. 2012;26(6):1211-1217. Christopeit M, Kuss O, Finke J, et al. Second allograft for hematologic relapse of acute leukemia after first allogeneic stem-cell transplantation from related and unrelated donors: the role of donor change. J Clin Oncol. 2013;31(26):3259-3271. Tischer J, Engel N, Fritsch S, et a. Second haematopoietic SCT using HLA-haploidentical donors in patients with relapse of acute leukaemia after a first allogeneic transplantation. Bone Marrow Transplant. 2014;49(7): 895-901. Jabbour E, Giralt S, Kantarjian H, et al. Lowdose azacitidine after allogeneic stem cell transplantation for acute leukemia. Cancer. 2009;115(9):1899-1905. Schroeder T, Czibere A, Platzbecker U, et al. Azacitidine and donor lymphocyte infusions as first salvage therapy for relapse of AML or MDS after allogeneic stem cell transplantation. Leukemia. 2013;27(6):12291235. Rohner A, Langenkamp U, Siegler U, Kalberer CP, Wodnar-Filipowicz A. Differentiation-promoting drugs up-regulate NKG2D ligand expression and enhance the susceptibility of acute myeloid leukemia cells to natural killer cell-mediated lysis. Leuk Res. 2007;31(10)1393-1402. Goodyear O, Agathanggelou A, NovitzkyBasso I, et al. Induction of a CD8+ T-cell response to the MAGE cancer testis antigen by combined treatment with azacitidine and sodium valproate in patients with acute myeloid leukemia and myelodysplasia. Blood. 2010;116(11):1908-1918. Goodyear OC, Dennis M, Jilani NY, et al. Azacitidine augments expansion of regulatory T cells after allogeneic stem cell transplantation in patients with acute myeloid leukemia (AML). Blood. 2012;119(14):33613369. Schroeder T, Frobel J, Cadeddu RP, et al. Salvage therapy with azacitidine increases regulatory T cells in peripheral blood of patients with AML or MDS and early

26.

27.

28.

29.

30.

31.

32.

33.

34.

relapse after allogeneic blood stem cell transplantation. Leukemia. 2013;27(9):1910-1913. Craddock C, Jilani N, Siddique S, et al. Tolerability and clinical activity of posttransplantation azacitidine in patients allografted for acute myeloid leukemia treated on the RICAZA trial. Biol Blood Marrow Transplant. 2016;22(2):385-390. Davids MS, Kim HT, Bachireddy P, et al. Ipilimumab for Patients with Relapse after allogeneic transplantation. N Engl J Med. 2016;375(2):143-153. de Lima M, Bonamino M, Vasconcelos Z. Prophylactic donor lymphocyte transfusion after moderately ablative chemotherapy and stem cell transplantation for hematological malignancies: high remission rate at the expense of of graft-versus-host disease. Bone Marrow Transplant. 2001;27(1):73-78. Barge RM, Osanto S, Marijt WA, et al. Minimal GVHD following in-vitro T celldepleted allogeneic stem cell transplantation with reduced-intensity conditioning allowing subsequent infusions of donor lymphocytes in patients with hematological malignancies and solid tumors. Exp Hematol. 2003;31(10):865-872. Jedlickova Z, Schmid C, Koenecke C, et al. Long-term results of adjuvant donor lymphocyte transfusion in AML after allogeneic stem cell transplantation. Bone Marrow Transplant. 2016;51(5):663-667 de Lima M, Giralt S, Thall PF, et al. Maintenance therapy with low-dose azacitidine after allogeneic hematopoietic stem cell transplantation for recurrent acute myelogenous leukemia or myelodysplastic syndrome: a dose and schedule finding study. Cancer. 2010;116(23):5420-5431. Mohamedbhai SG, Edwards N, Morris EC, et al. Predominant or complete recipient Tcell chimerism following alemtuzumabbased allogeneic transplantation is reversed by donor lymphocytes and not associated with graft failure. Br J Haematol. 2012;156 (4):516-522. Platzbecker U, Wermke M, Radke J, et al. Azacitidine for treatment of imminent relapse in MDS or AML patients after allogeneic HSCT: results of the RELAZA trial. Leukemia. 2012;26(3):381-389. Bug G, Burchert A, Wagner EM, et al. Phase I/II study of the deacetylase inhibitor panobinostat after allogeneic stem cell transplantation in patients with high-risk MDS or AML (PANOBEST trial). Leukemia. 2017 Jul 28. [Epub ahead of print]

245

ARTICLE

Acute Myeloid Leukemia

Ferrata Storti Foundation

Haematologica 2018 Volume 103(2):246-255

Micro-ribonucleic acid-155 is a direct target of Meis1, but not a driver in acute myeloid leukemia

Edith Schneider,1 Anna Staffas,2 Linda Röhner,1 Erik D. Malmberg,2 Arghavan Ashouri,3 Kathrin Krowiorz,1 Nicole Pochert,1 Christina Miller,1 Stella Yuan Wei,2,4 Laleh Arabanian,2 Christian Buske,5Hartmut Döhner,1 Lars Bullinger,1 Linda Fogelstrand,2,6 Michael Heuser,7 Konstanze Döhner,1 Ping Xiang,8 Jens Ruschmann,8 Oleh I. Petriv,9 Alireza Heravi-Moussavi,10 Carl L. Hansen,11 Martin Hirst,10,11 R. Keith Humphries,8 Arefeh Rouhi,1* Lars Palmqvist2,6* and Florian Kuchenbauer1,5*

Department of Internal Medicine III, University Hospital of Ulm, Germany; 2Department of Clinical Chemistry and Transfusion Medicine, Institute of Biomedicine, Sahlgrenska Academy at University of Gothenburg, Sweden; 3Institute of Biomedicine, University of Gothenburg, Sweden; 4Department of Medicine, Sahlgrenska University Hospital, Gothenburg, Sweden; 5Institute of Experimental Cancer Research, Comprehensive Cancer Centre Ulm, Germany; 6Department of Clinical Chemistry, Sahlgrenska University Hospital, Gothenburg, Sweden; 7Department of Hematology, Homeostasis, Oncology and Stem Cell Transplantation, Hannover Medical School, Germany; 8Terry Fox Laboratory, British Columbia Cancer Agency, Vancouver, BC, Canada; 9Department of Physics and Astronomy, University of British Columbia, Vancouver, BC, Canada; 10 Canada's Michael Smith Genome Sciences Centre, British Columbia Cancer Agency, Vancouver, BC, Canada and 11Centre for High-Throughput Biology, University of British Columbia, Vancouver, BC, Canada 1

*AR, LP and FK contributed equally to this work

ABSTRACT

Correspondence: florian.kuchenbauer@uni-ulm.de

Received: July 28, 2017. Accepted: November 30, 2017. Pre-published: December 7, 2017. doi:10.3324/haematol.2017.177485 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/246 ©2018 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.

246

icro-ribonucleic acid-155 (miR-155) is one of the first described oncogenic miRNAs. Although multiple direct targets of miR155 have been identified, it is not clear how it contributes to the pathogenesis of acute myeloid leukemia. We found miR-155 to be a direct target of Meis1 in murine Hoxa9/Meis1 induced acute myeloid leukemia. The additional overexpression of miR-155 accelerated the formation of acute myeloid leukemia in Hoxa9 as well as in Hoxa9/Meis1 cells in vivo. However, in the absence or following the removal of miR155, leukemia onset and progression were unaffected. Although miR155 accelerated growth and homing in addition to impairing differentiation, our data underscore the pathophysiological relevance of miR-155 as an accelerator rather than a driver of leukemogenesis. This further highlights the complexity of the oncogenic program of Meis1 to compensate for the loss of a potent oncogene such as miR-155. These findings are highly relevant to current and developing approaches for targeting miR-155 in acute myeloid leukemia. Introduction Leukemogenesis is a complex multistep process that impacts differentiation, proliferation and self-renewal. Large profiling approaches such as the Cancer Genome Project defined mutations in genes such as FLT3, NPM1, DMNT3A and NRAS as driver mutations in acute myeloid leukemia (AML).1,2 Mutations in miRNAs, a class of short non-coding RNAs, are rare, but quantitative expression studies have led to a better understanding of how deregulation of miRNAs associates with genetic subgroups of AML.1,3 However, only functional studies allow the elucidation of the potential of miRNAs as drivers in leukemia and as therapeutic targets. It is generally assumed that a single miRNA has hundreds of putative targets and can therefore simultaneously affect multiple pathways and processes. In hematopoiesis, specific miRNA expression patterns4 maintain a fine balance between hematopoietic stem and progenitor cell (HSPC) self-renewal and differentiation5 as well as between normal and malignant hematopoiesis.6 Therefore, the targeting of miRNAs holds promise for advancing targeted cancer therapies of currently undruggable genetic translocations and pathways. One of the first described oncogenic miRNAs, miR-155, was originally identihaematologica | 2018; 103(2)

microRNA-155 amplifies leukemogenesis

fied to be overexpressed in both lymphomas and in AML.7-9 However, the roles of miR-155 in leukemogenesis and hematopoiesis are more complex, as it impacts inflammatory processes, B-cell and T-cell function in addition to myeloid development.8,10-13 We have recently shown that miR-155 functions in HSPC mobilization,14 suggesting that miR-155 bears a more complex role in stem cell physiology than previously assumed. We also reported that miR-155 levels are correlated with MLL translocations, an AML subtype characterized by high HOX gene and MEIS1 levels.15 In AML, transcript levels of HOXA9 are highly correlated with poor prognosis,16 and engineered overexpression of Hox proteins in hematopoietic cells results in long latency leukemia in mice, indicating that collaborating genetic events are required for full leukemic transformation.17,18 The HOX cofactor, Meis1, is rate-limiting for MLL-rearranged AML and has been identified as collaborating with HOX proteins and HOX fusions (NUP98HOX) to induce a rapid disease onset of AML in mice.19-21 With the aim of identifying leukemia-contributing miRNAs and defining their roles in leukemogenesis, we sought to build a clinically relevant model system for AML. Using a Hoxa9 and Meis1 murine AML progression model,22 together with findings in human AML, herein we have identified deregulated miRNAs downstream of Hoxa9 and Meis1, and have further characterized the role of miR-155 in AML development as well as its potential as a therapeutic target both in vitro and in vivo.

Methods

MiRNA and messenger (m)RNA expression arrays RNA for the array analysis was prepared from independently generated cell lines three to four weeks post transduction, expressing Hoxa9/ctrl (n=9), Hoxa9/Meis1 (n=4), Hoxa9/ΔHDMeis1 (n=4), Hoxa9/Meis1155-/- (n=3) or Hoxa9/miR155 (n=3). A detailed data analysis is described in the Online Supplementary Methods. The Gene Expression Omnibus (GEO) accession numbers for miRNA array and mRNA array of Hoxa9/ctrl, Hoxa9/Meis1 and Hoxa9/ΔHDMeis1 are GSE74566 and GSE75272, respectively. Exon array data for Hoxa9/ctrl, Hoxa9/miR-155 and Hoxa9/Meis1155-/- are available at GSE76113.

Human samples from healthy donors and AML patients All samples were collected according to protocols approved by the Ethics Committee of the University Hospital of Ulm. All probands gave informed consent for genetic analysis according to the Declaration of Helsinki. Detailed descriptions of the patient samples used in this study are provided in the Online Supplementary Methods.

Real-time quantitative polymerase chain reaction (RT-qPCR) Detailed RT-qPCR experiments and primers are described in the Online Supplementary Methods.

Flow cytometric analysis and fluorescence activated cell sorting (FACS) Immunophenotype analysis was performed on either a LSRFortessa™ cell analyzer or FACSAria II and cell sorting on a FACSAria II or FACSAria III sorter (Becton Dickinson Biosciences). Antibodies used for immunophenotype determination and subpopulation sorting are described in the Online Supplementary Methods.

Retroviral and Lentiviral constructs Murine stem cell virus (MSCV)-based retroviral vectors carrying expression cassettes consisting of Hoxa923 Meis124 ΔHDMeis117 and miR-1558 have been described previously. The generation of recombinant retrovirus-producing GP-E86 cells was performed as previously described.25 The lentiviral miR-155 sponge vector BdLV.155pT and the scrambled control vector BdLV.Ctrl have been formerly elucidated.26

Generation of transduced murine bone marrow (BM) cells and transplantation assays A detailed description of the generation of cell lines and cell culture as well as proliferation and clonogenic assay experiments are described in the Online Supplementary Methods. All animal experiments were approved by the state government of Gothenburg, Sweden and Tuebingen, Germany. Transplantation assays were performed as previously described.17,25 Detailed engraftment analysis, sick mice workup, limiting dilution and homing assay are also described in the Online Supplementary Methods.

Chromatin immunoprecipitation and sequencing (ChIP-seq) ChIP-seq of the Hoxa9/Meis1 cell line was performed using standard operating procedures for ChIP-seq library construction as previously described.27 Details of antibodies that were used are described in the Online Supplementary Methods. Enriched Meis1 binding regions on chromosome 16 in Hoxa9/Meis1 cells are summarized in Online Supplementary Table S1. Identified peaks for the determined histone modifications on chromosome 16 are listed in Online Supplementary Table S2 for Hoxa9/ctrl cells and Online Supplementary Table S3 for Hoxa9/Meis1 cells. haematologica | 2018; 103(2)

Microfluidics analysis Mononuclear cells prepared from diagnostic peripheral blood (PB) of 17 newly diagnosed cytogenetically normal (CN)-AML (FLT-internal tandem duplication (ITD) pos, n=9) patient samples were sorted into CD34+CD38–, CD34+ and myeloid-enriched subpopulations and processed as previously described. Antibodies used for immunophenotype determination and cycle threshold (Ct) value calculations are described in the Online Supplementary Methods. Samples were collected at the British Columbia Cancer Agency (BCCA, Vancouver, BC, Canada) according to protocols approved by the Ethics Committee of the BCCA.

Western blot analysis Protein extraction and immunoblotting was performed via standard procedures using the following antibodies per manufacturer’s instructions: anti-CD13 (ANPEP, clone EPR4058) (Abcam), anti-JARID2 (Novus), and anti-β-actin (clone AC-15) (SigmaAldrich).

Statistics Pairwise comparisons were performed using the Mann-Whitney U test, unless otherwise specified. The KaplanMeier method with log-rank test was used to compare differences between survival curves. Spearman correlation was used for tests of relationships. Leukemia-initiating cell (LIC) frequencies were calculated with L-Calc™ Software Version 1.1 (STEMCELL Technologies). Unless otherwise indicated, data are expressed as mean ± standard error of the mean (s.e.m.). A P-value of less than 0.05 was defined as statistically significant; statistical significance is shown as *P<0.05, **P<0.01 and ***P<0.001. 247

E. Schneider et al.

Results

Table 1. Differentially expressed miRNAs between Hoxa9/Meis1 and Hoxa9/ctrl or Hoxa9/ΔHDMeis1.

MiR-155 is a direct target of Meis1

miRNAs

To identify miRNAs relevant for transforming HSPCs into AML cells, we compared the miRNA transcriptome of pre-leukemic cells overexpressing Hoxa9 and leukemic cells co-overexpressing Hoxa9 and Meis1.22 Murine BM cells were transduced to generate cell lines overexpressing Hoxa9 alone (with an empty vector control, Hoxa9/ctrl, n=9), Hoxa9 together with a mutant and inactive Meis1 lacking the homeodomain (Hoxa9/ΔHDMeis1, n=4) or wild-type (wt) Meis1 (Hoxa9/Meis1, n=4).17,28 Analysis of the miRNA transcriptome identified 16 significantly deregulated miRNAs (Table 1). Of these, miR155-5p (henceforth referred to as miR-155) was the most significantly upregulated miRNA in Hoxa9/Meis1 cells compared to the Hoxa9/ctrl and Hoxa9/ΔHDMeis1 cells. The upregulation of miR-155 was validated in independently generated Hoxa9/Meis1 cells (Figure 1A). Meis1 or Hoxa9 overexpression alone did not result in the upregulation of miR-155 compared to BM cells transduced with an empty control vector, indicating that miR-155 upregulation requires co-expression of Hoxa9 and Meis1 (Figure 1A). In line with this observation, the host gene of miR155, miR-155hg (Bic), was also more highly expressed in Hoxa9/Meis1 compared to Hoxa9/ctrl cells, but not in cells that ectopically expressed either Meis1 or Hoxa9 only (Figure 1B). We further validated the interaction of Meis1 with the miR-155hg/miR-155 locus in Hoxa9/Meis1 cells and characterized epigenetic changes (H3K4me, H3K27ac, H2K4me1, H3K36me3) by ChIP-seq. Our data demonstrate that Meis1 binds to a region approximately 4kb upstream of miR-155hg (Figure 1C and Online Supplementary Figure S1A), accompanied by a marked increase in H3K27ac. We further detected an increase in H3K4me3 at the promoter and 5’ region of miR-155hg as well as enrichment of H3K36me3 along the gene body in Hoxa9/Meis1 cells (Figure 1C). H3K27ac and H3K4me3 levels in Hoxa9/ΔHDMeis1 were similar to that of Hoxa9/ctrl cells (data not shown). The direct binding of Meis1 to a putative enhancer element and the enrichment of activating epigenetic marks is in line with increased transcript levels of miR-155hg and miR-155 in Hoxa9/Meis1 cells.

MiR-155 augments the leukemogenic potential of Hoxa9 Based on the upregulation of miR-155 in leukemic Hoxa9/Meis1 cells and its known oncogenic potency,29 we hypothesized that miR-155 may fully or partially account for the leukemogenic properties of Meis1 when coexpressed with Hoxa9. In addition, we tested the transforming potential of miR-155 alone by transduction of murine BM with miR-155 or an empty control vector. The overexpression levels of miR-155 in Hoxa9/miR-155 cells and cells overexpressing miR-155 alone are shown in Online Supplementary Figure S1B. In vitro analysis of cell lines overexpressing Hoxa9/miR-155 showed no difference in proliferation when compared to Hoxa9/ctrl, whereas Hoxa9/Meis1 cells grew significantly faster in liquid culture compared to Hoxa9/ctrl (Online Supplementary Figure S1C). The colony forming capacity of Hoxa9/miR155 was significantly elevated to a level similar to that found in Hoxa9/Meis1 cells when compared to Hoxa9/ctrl (Online Supplementary Figure S1D). The immunophenotype 248

miR-155-5p miR-708-5p miR-30a-3p miR-182-5p miR-449a-5p miR-155-3p miR-30c-2-3p miR-30a-5p miR-183-5p miR-501-5p miR-99b-5p miR-146a-5p miR-466b-3p miR-125a-5p miR-511-3p miR-466e-3p

P value 9.08E-03 1.04E-02 1.39E-02 1.77E-02 2.18E-02 3.32E-02 3.32E-02 3.77E-02 3.92E-02 9.41E-03 9.41E-03 1.83E-02 1.83E-02 3.39E-02 4.29E-02 4.89E-02

Fold change Fold change Hoxa9/Meis1 vs. Hoxa9/Meis1 vs. Hoxa9/Meis1 Hoxa9/ΔHDMeis1 2.4 6.7 5.2 2.1 3.9 2.8 2.3 2.0 4.0 -2.3 -5.2 -4.8 -2.0 -3.1 -4.2 -2.0

3.5 6.4 4.5 2.0 2.4 3.0 2.5 2.1 2.7 -2.0 -4.8 -6.1 -3.9 -3.9 -2.2 -3.9

of Hoxa9/miR-155 was also comparable to that of Hoxa9/Meis1 cells with significantly more c-kit-expressing cells and lower numbers of the more mature cells expressing Mac-1 and Gr-1 (Online Supplementary Figure S1E). MiR-155 overexpressing BM cells showed increased proliferation and colony formation in vitro (Online Supplementary Figure S1C,S1D), supporting the idea that miR-155 may possess transforming potential by increasing self-renewal capacity. Based on the recent report by Narayan et al.,30 we investigated the kinetics of miR-155 upregulation and of a possible selection process during Hoxa9/Meis1-mediated transformation by quantification of miR-155 in an in vitro time course experiment. MiR-155 upregulation was stable over time (Figure 1D), demonstrating that in our model miR-155 expression is not associated with differentiation and endogenous selection in Hoxa9/Meis1 cells. In vivo, mice transplanted with Hoxa9/miR-155 showed higher engraftment levels compared to Hoxa9/ctrl, but to a lower extent than mice transplanted with Hoxa9/Meis1 cells (Figure 2A). Overexpression of miR-155 alone did not enhance engraftment, and transplanted mice remained healthy (Figure 2A,B). The Hoxa9/miR-155 mice succumbed to leukemia with a median latency of 82 days (range: 32-153 days), corresponding to approximately twice the median latency of mice transplanted with Hoxa9/Meis1 (36 days, range: 28-62 days) and half the median latency of mice transplanted with Hoxa9/ctrl cells (162 days, range: 106-206 days) (Figure 2B). Mice transplanted with Hoxa9/miR-155 cells developed an acute myelomonocytic leukemia, based on the Bethesda criteria for non-lymphoid neoplasias31 (Figure 2C). Similar to Hoxa9/Meis1 induced leukemias that also show a myelomonocytic phenotype, mice transplanted with Hoxa9/miR-155 cells exhibited hepatomegaly and splenomegaly comparable to Hoxa9/ctrl (Online Supplementary Figure S2A). Blood cell counts of Hoxa9/miR-155 mice showed increased leukocytosis haematologica | 2018; 103(2)

microRNA-155 amplifies leukemogenesis

Figure 1. MiR-155 and its host gene, miR-155hg, are significantly upregulated in leukemic Hoxa9/Meis1 cells. (A) Relative expression of miR-155 in BM cells independently transduced with Hoxa9/Meis1 (n=7), Hoxa9 (n=5) or Meis1 (n=5) quantified by RT-qPCR and expressed relative to cells transduced with Hoxa9/ctrl (n=7) or an empty control vector (ctrl) (n=5), respectively. (B) Expression of miR-155hg in BM cells independently transduced with Hoxa9/Meis1 (n=7), Hoxa9 (n=3) or Meis1 (n=3) relative to expression in cells transduced with Hoxa9/ctrl (n=7) or an empty control (ctrl) (n=3), respectively. (C) ChIP-sequencing tracks for H3K4me3, H3K27ac, H3K4me1 and H3K36me3 in Hoxa9/Meis1 and Hoxa9/ctrl cells at the Meis1 binding site and miR-155hg locus are shown mapped to the mouse mm10 genome browser. Location of Meis1 binding site identified by Meis1 ChIP-sequencing is shown with a black bar. The black arrow depicts the transcriptional direction of miR-155hg. Open boxes indicate the region of the Meis1 binding site and the 5â&#x20AC;&#x2122;region of the miR-155hg. (D) The kinetics of miR-155 expression measured by RT-qPCR over time in n=3 biological replicates of Hoxa9/Meis1 transduced cells relative to Hoxa9/ctrl cells. Statistical significance was calculated using the Studentâ&#x20AC;&#x2122;s t-test (two-tailed). See also Online Supplementary Tables S5-S7.

when compared to Hoxa9/ctrl, whereas red blood cells (RBCs) and platelet levels were similar in deceased mice transplanted with Hoxa9/ctrl, Hoxa9/miR-155 and Hoxa9/Meis1 (Online Supplementary Figure S2B). Flow cytometry analysis of BM cells from deceased Hoxa9/miR-155 mice showed a lower proportion of c-kitexpressing cells in comparison to BM from Hoxa9/Meis1 mice, but elevated c-kit+ cells compared to Hoxa9/ctrl, verifying the immature blast accumulation in these mice (Figure 2D). These results show that miR-155 partially recapitulates the pro-leukemic effects of Meis1, leading to AML with intermediate latency.

MiR-155 promotes the Hoxa9 gene expression program To dissect the molecular differences between the longlatency Hoxa9/ctrl, the intermediate-latency Hoxa9/miR155 and the short-latency Hoxa9/Meis1 cells, we performed mRNA expression arrays. The fold change of the deregulated genes in Hoxa9/miR-155 and Hoxa9/Meis1 cells was defined relative to the Hoxa9/ctrl cells. In Hoxa9/miR-155 cells 223 genes and in Hoxa9/Meis1 cells 381 genes were differentially expressed when compared to Hoxa9/ctrl cells (Online Supplementary Table S4). Analysis of the overlapping genes deregulated in haematologica | 2018; 103(2)

Hoxa9/miR-155 and Hoxa9/Meis1 cells displayed a strikingly similar gene expression pattern with 64 common genes, of which 58 genes followed the same expression direction in both cell lines (Figure 3A and Online Supplementary Table S4), further indicating a role for miR155 in effecting the leukemogenic potential of Hoxa9/Meis1. Of the 223 differentially expressed genes in Hoxa9/miR-155 cells, only 29 were upregulated, and of the 194 downregulated genes, 114 were validated targets of miR-155 (Online Supplementary Table S5), based on an analysis using TarBase.32 Among these downregulated genes were a number of predicted miR-155 targets linked to hematological malignancies, including the previously validated miR-155 targets Csf1r, Jarid2 and Picalm8 as well as targets associated with myeloid cell differentiation, such as Anpep (CD13).33 Gene ontology and pathway analysis of all differentially expressed genes in Hoxa9/miR-155 cells revealed immune- and cell adhesionrelated processes as the main affected pathways (Online Supplementary Table S6). To further define the effects of miR-155 overexpression on Hoxa9-mediated pathways, we compared the differentially expressed genes of the Hoxa9/miR-155 and Hoxa9/Meis1 cells (normalized to Hoxa9/ctrl, as described above) to a previously published Hoxa9-medi249

E. Schneider et al.

Figure 2. MiR-155 cooperates with Hoxa9 to induce leukemia in vivo. (A) Engraftment kinetics in peripheral blood (PB) of mice transplanted in three to four individual experiments with independently generated cell lines for each experiment of Hoxa9/ctrl (week 4 n=25; week 8 n=23; week 12 n=8), Hoxa9/miR-155 (week 4 n=24; week 8 n=20; week 12 n=5), Hoxa9/Meis1 (n=23), miR-155 (n=16) or a control vector (n=16) for each experiment. Engraftment was measured by flow cytometry analysis and data is shown as mean of green fluorescent protein (GFP)+ or yellow fluorescent protein (YFP)+ cells at 4, 8 and 12 weeks after transplantation. (B) Survival curves for cohorts of mice transplanted with independently generated Hoxa9/ctrl (n=25), Hoxa9/miR-155 (n=25) and Hoxa9/Meis1 (n=24) cells via a minimum of four individual transplantation experiments and transplantation of cells overexpressing miR-155 (n=16) or control (n=16) from three individual experiments. (C) Representative microscopic photographs of Wright-stained bone marrow (BM) cells (magnification 50x) are shown for mice that succumbed to leukemia. (D) Percentage of c-kit-expressing cells measured via flow cytometry in the BM cells of mice that developed leukemia after transplantation of Hoxa9/ctrl (n=8), Hoxa9/miR-155 (n=13) or Hoxa9/Meis1 (n=13) cells. See also Online Supplementary Figure S1 and Online Supplementary Figure S2.

ated gene expression dataset derived from a conditional Hoxa9 cell line by Huang et al.34 There were 152 overlapping genes between Hoxa9/miR-155 and the Hoxa9-mediated gene expression dataset, of which 145 genes showed the same expression direction (i.e., pro Hoxa9-mediated program). Hoxa9/Meis1 cells and the Hoxa9-mediated gene expression dataset shared 219 overlapping genes, the majority of which (191 genes) followed the same expression direction as the Hoxa9-mediated program (Figure 3B and Online Supplementary Table S5). To evaluate if miR-155 expression is functionally redundant in the setting of Hoxa9 and Meis1 overexpression, as suggested by the expression analysis, we engineered overexpression of miR-155 in Hoxa9/Meis1 cells (Hoxa9/Meis1/miR-155) that were transplanted into syngeneic recipients. Additional miR-155 overexpression significantly accelerated the formation of AML (Figure 3C), further emphasizing the oncogenic contribution of miR-155 to leukemogenesis. Our data indicate that miR-155 partially activates a similar transcriptional program as Meis1 together with Hoxa9, leading to enhanced leukemogenesis, therefore suggesting that it is one of the key miRNAs within the Hoxa9/Meis1 leukemic program.

MiR-155 expression is not hierarchical in Hoxa9/Meis1 induced AML We further hypothesized that miR-155 expression reflects a hierarchical structure within Hoxa9/Meis1 leukemias, comparable to its differential expression during 250

normal hematopoiesis (Online Supplementary Figure S3A) where its levels decrease with differentiation. Thus, we sorted independently generated Hoxa9/Meis1 cells into subpopulations defined by c-kit, Gr-1 and Mac-1 expression (Online Supplementary Figure S3B), with LICs being enriched in the immature c-kit+Gr-1-Mac-1- population as previously reported by Gibbs et al.35 and validated in vivo (Online Supplementary Figure S3C). No difference in miR155 expression was found among the subpopulations (Figure 3D), indicating that miR-155 might not be critical for LICs in the context of Hoxa9/Meis1.

MiR-155 is dispensable for the onset and progression of AML Albeit miR-155 has become a pharmacological target in hematological malignancies,13,36-38 in vivo data showing the therapeutic benefit of direct depletion of miR-155 from AML cells is lacking. To explore the relevance of miR-155 in AML, we investigated the dependency of Hoxa9/Meis1-induced leukemias on miR-155 expression by overexpressing Hoxa9/Meis1 in BM cells of mice lacking miR-155 expression (Hoxa9/Meis1155–/–). Although the proliferation rate of Hoxa9/Meis1155–/– cells showed no difference compared to wt cells transduced with Hoxa9/Meis1 (Hoxa9/Meis1155+/+), colony formation capacity was impaired in Hoxa9/Meis1155–/– cells for the first two replates, but was compensated after the third replate and comparable to levels seen for the Hoxa9/Meis1155+/+ cells (Online Supplementary Figure S4A,S4B). The absence of haematologica | 2018; 103(2)

microRNA-155 amplifies leukemogenesis

Figure 3. Hoxa9/miR-155 and Hoxa9/Meis1 cells display a similar gene expression pattern. (A) Upper panel shows overlap of differentially expressed genes in Hoxa9/miR-155 and Hoxa9/Meis1 cells. The fold change of differentially expressed genes was calculated relative to Hoxa9/ctrl cells. The scatter plot below shows the fold change of common genes between Hoxa9/Meis1 and Hoxa9/miR-155. The significance and correlation coefficient (r) was calculated using Pearson’s correlation coefficient. (B) Overlap between gene expression profile of Hoxa9/miR-155 and Hoxa9/Meis1 cells with the Hoxa9-mediated gene expression dataset from a conditional Hoxa9 cell line (Huang et al).34 (C) Survival curves of mice transplanted with two biological replicates of Hoxa9/Meis1/ctrl or Hoxa9/Meis1/miR 155 cells (n=13/arm). (D) MiR-155 expression levels in the depicted sorted subpopulations of Hoxa9/Meis1 cells shown relative to expression in Hoxa9/Meis1 bulk cells. For each subpopulation n=4 biological replicates were analyzed, except for the c-kit Gr 1+Mac 1+ subpopulation, where only four replicates were available. See also Online Supplementary Figure S3, Online Supplementary Table S4 and Online Supplementary Table S5.

miR-155 led to lower engraftment levels in mice transplanted with Hoxa9/Meis1155–/– after four weeks, but at the time of death these levels reached that of the Hoxa9/Meis1155+/+ cells, mirroring the results from the colony-forming assay and resulting in similar survival kinetics as those for Hoxa9/Meis1155+/+ (Figure 4A). MiR155 has recently been linked to a leukemic stem cell signature.39 Therefore, to exclude differences in LIC frequency between Hoxa9/Meis1155–/– and Hoxa9/Meis1155+/+, we performed limiting dilution transplantations where no significant difference was detected (Figure 4B). In order to explore the possibility of impaired early leukemic cell and homing, we sacrificed Hoxa9/Meis1155–/– Hoxa9/Meis1155+/+ transplanted mice 12 hours post-transplantation. Hoxa9/Meis1155–/– cells showed significantly less yellow fluorescent protein (YFP)+ cells, indicating a role of miR-155 in homing of AML cells (Figure 4C, left panel). This observation was already reversed after one week, with similar YFP levels, despite lower c-kit levels in Hoxa9/Meis1155–/– cells (Figure 4C, middle and right panel), indicating differences in LIC pathophysiology. Molecular and differences between Hoxa9/Meis1155–/– Hoxa9/Meis1155+/+ cells were explored by mRNA expression arrays. Since miRNAs act as rheostats and induce small gene expression differences, we investigated genes with a fold change >1.2 and P-value cutoff of 0.05. In Hoxa9/Meis1155–/– cells 295 genes were downregulated and 115 upregulated compared to Hoxa9/Meis1155+/+ (Online Supplementary Table S7). As expected, miR-155 targets haematologica | 2018; 103(2)

Jarid2 and Anpep, which are involved in the regulation of hematopoietic cell differentiation and were downregulated in Hoxa9/miR-155 cells, were upregulated in Hoxa9/Meis1155–/– cells (Online Supplementary Table S7). To further investigate if wt Hoxa9/Meis1 cells are addicted to miR-155, we combined an immortalized Hoxa9/Meis1 cell line with a miR-155 sponge vector26 (Hoxa9/Meis1/155sp) or a scrambled vector (Hoxa9/Meis1/scr) as control. MiR-155 levels were significantly reduced by the miR-155 sponge (Online Supplementary Figure S4C). Reduced protein levels of the miR-155 targets, Jarid2 and Anpep,8 which we previously identified by mRNA expression arrays (Online Supplementary Table S4 and Online Supplementary Table S7), were confirmed with immunoblotting in Hoxa9/Meis1/155sp sponge as well as Hoxa9/Meis1155–/– cells when compared to wt Hoxa9/Meis1 cells (Figure 4D). However, no difference in survival was observed between Hoxa9/Meis1/scr and Hoxa9/Meis1/155sp transplanted mice (Figure 4E). Taken together, these results indicate that the oncogenic program of Meis1 is not dependent on miR-155, as shown in the knockout experiments, and can overcome a 50% reduction of miR-155 expression.

Lower expression of miR-155 in AML patients compared to HSPCs Based on our data in the Hoxa9/Meis1 model, we sought to corroborate our findings in human AML samples. Within The Cancer Genome Atlas Acute Myeloid 251

E. Schneider et al. Leukemia (TCGA-LAML) dataset,1 miR-155 expression levels were significantly increased in all AML patients when compared to the global mean of all miRNAs, and particularly in AML patients with mutated FLT3 (Figure 5A), reinforcing the fact that miR-155 is especially abundant in CN-AML. We further analyzed the correlation between HOXA9 and MEIS1 with miR-155 levels within the TCGA-LAML dataset. Supporting our hypothesis, CN-AML with mutated NPM1, known to have high

HOXA9 and MEIS1 levels,40 displayed a significant correlation between HOXA9, MEIS1 and miR-155 levels (Online Supplementary Figure S5A,S5B) within the TCGALAML dataset. To investigate the relationship of our findings in human AML to normal hematopoiesis, we quantified miR-155 levels by RT-qPCR in diagnostic samples from CN-AML and AML with t(11q23) in relation to human CD34+ cord blood cells, total BM and granulocyte samples from

Figure 4. Absence or depletion of miR-155 does not alter leukemogenicity of Hoxa9/Meis1 cells. (A) Engraftment in peripheral blood (PB) after 4 weeks (left panel), at the time of death (middle panel, n=7/arm) and survival (right panel) of mice transplanted in two independent experiments with two biological replicates of Hoxa9/Meis1155+/+ (n=9) or Hoxa9/Meis1155-/- (n=9) cells. Percentage of yellow fluorescent protein (YFP)+ cells was measured by flow cytometry in PB of mice. (B) Limiting dilution assay of Hoxa9/Meis1155+/+ and Hoxa9/Meis1155-/- transplanted cells with indicated numbers of mice transplanted for each arm. Leukemia initiating cell frequency was calculated using L-cal™. (C) Left: homing percentage of Hoxa9/Meis1155+/+ and Hoxa9/Meis1155-/- (n=5/arm) cells transplanted into non-irradiated recipient mice and assessed 12h post transplantation. Middle: percentage of Hoxa9/Meis1155+/+ and Hoxa9/Meis1155-/- (n=5/arm) cells transplanted into non-irradiated recipient mice and assessed one week post transplantation. Right: percentage of c-kit positive cells in Hoxa9/Meis1155+/+ and Hoxa9/Meis1155-/- (n=5/arm) cells transplanted into non-irradiated recipient mice and assessed one week post transplantation. (D) Left: western blot of Jarid2 and Anpep in Hoxa9/Meis1155+/+ and Hoxa9/Meis1155-/- as well as Hoxa9/Meis1/155 sponge and Hoxa9/Meis1/scr-transduced cells. Right: densitometry of Anpep and Jarid2 normalized to β-actin (gene/β-actin) using the ImageJ software (n=2 biological replicates). (E) Survival curves of two independent transplantation experiments, where 1000 cells of Hoxa9/Meis1/155 sponge (n=10) or Hoxa9/Meis1/scr, (n=7) cells were transplanted. See also Online Supplementary Figure S4 and Online Supplementary Table S7. BM: bone marrow; LIC: Leukemia-initiating cell.

252

haematologica | 2018; 103(2)

microRNA-155 amplifies leukemogenesis

healthy donors. Interestingly, the expression of miR-155 was significantly higher in CD34+ cells compared to the profiled AML subgroups (Figure 5B). We further investigated whether miR-155 is differentially expressed within AML subpopulations based on their differentiation status, reinforcing its potential as a therapeutic target in AML. For this purpose, we sorted subpopulations from seven randomly chosen AML patients according to their maturation states. In line with our findings in the Hoxa9/Meis1 cell lines, there were no differences in miR-155 expression among the subpopulations (Figure 5C). We confirmed this finding in CN-AML patient samples (n=17) (Online Supplementary Figure S5C), reinforcing the concept that miR-155 enhances, rather than drives, leukemogenesis.

Discussion Screening a Hoxa9 and Hoxa9/Meis1-based AML pro-

gression model, we found miR-155 to be the most significantly upregulated miRNA in leukemic Hoxa9/Meis1 cells. MiR-155 is a known oncogene that accelerates the formation of lymphomas and expansion of myeloid cells when overexpressed.8,29 Considering the relevance of the HOXA9/MEIS1 axis in AML, the full extent of pro-leukemic targets of HOXA9 and MEIS1 is not yet fully understood, which leaves this axis undruggable. Several of these targets, such as Sytl141 and Syk,42 support leukemic cell homing and engraftment and promote leukemogenesis. Thus far, few miRNAs, including miR-196b,43 miR-2144 and miR-146a42 have been linked with Hoxa9 and Meis1 regulation. Huang et al. identified a Meis1 binding site nearly 4kb upstream of the transcriptional start site of mir155hg (Bic), the host gene of miR-155.34 The data herein confirms, for the first time, the direct binding of Meis1 to this region that displays the hallmarks of an activated enhancer in Hoxa9/Meis1 cells. To our knowledge, miR-155 is the first discovered miRNA

Figure 5. MiR-155 expression is elevated in CD34+ cord blood cells but shows no difference between AML subpopulations. (A) MiR-155 expression levels across various genetic AML subgroups (n=187) from the LAML-TCGA dataset. The gray dashes indicate the global mean of all miRNAs for each sample. (B) MiR155 expression levels in CN-AML NPM1wt (n=10), CN-AML NPM1mut (n=10) and t(11q23) (n=8) AML patient samples as well as sorted granulocytes (n=4) and total BM from healthy donors (n=8), measured by RT-qPCR and compared to CD34+ cord blood cells (n=5). (C) Relative expression levels of miR155 in sorted subpopulations of AML patient samples (n=7) shown relative to expression in the CD34+CD117+CD38â&#x20AC;&#x201C; subpopulation. See also Online Supplementary Figure S5. AML: acute myeloid leukemia; miR: Micro-ribonucleic acid; CN-AML: cytogenetically normal

haematologica | 2018; 103(2)

253

E. Schneider et al.

which is a direct downstream target of Meis1, highlighting that the oncogenic potential of Meis1 is based on coding and non-coding genes. Our in vitro and in vivo studies demonstrate the collaborative and oncogenic potential of miR-155. However, the combination of Hoxa9 and miR-155 overexpression only partially recapitulated the leukemic potential of Hoxa9 and Meis1 overexpression, implying that additional genetic factors are needed for the generation of aggressive AML. For example, multiple profiling approaches have previously associated miR-155 with mutated FLT3 (FLT3ITD) in AML.3,13,45,46 MiR-155 has been attributed to oncogenic or tumor suppressor functions in AML. For example, Palma et al. highlighted a tumor suppressor role for miR155 through the induction of apoptosis and myeloid differentiation in the AML cell line OCI-AML3,47 whereas several groups have suggested the inhibition of miR-155 as a therapeutic approach in AML.37,46,48-50 It seems that the function of miR-155 could be context-dependent. Narayan et al. recently showed a dose-dependent role for miR-155 in AML, which might partially explain the published discrepancies.30 In the context of Hoxa9/Meis1-induced leukemia, a very potent AML model, we did not detect clonal selection based on endogenous miR-155 levels through enforced expression.30 This might be attributed to: 1) a very early selection of miR-155 clones with intermediate expression levels, 2) and/or to relatively constant ectopic expression levels, and 3) the specific (modeldependent) Hoxa9/Meis1 cell context in which the absence of miR-155 did not impact leukemogenesis. While miR-155 is overexpressed in a range of hematological malignancies, it is still unclear how miR-155 promotes leukemogenesis. We show that overexpression of miR-155 enhances the leukemic properties of Hoxa9 through downregulation of its targets, the majority of which are known tumor suppressor genes associated with processes such as myeloid differentiation, for example Jarid2 and Klf4.8,51,52 The absence or partial removal of miR155 did not impede Hoxa9 and Meis1-induced AML development, however, since we did not deplete miR-155 in other human AML models, a critical role for this miRNA in leukemogenesis cannot be fully excluded. Of note, although miR-155 is highly expressed in murine HSPCs, miR-155-/- mice did not show impaired myeloid differentiation or any perturbations in the HSPC compartment.11 Herein, we linked the absence of miR-155 to impaired early homing of Hoxa9/Meis1 cells, a previously unrecognized function of miR-155, which was reversed after only one week post transplantation, but which supports our recent findings demonstrating the necessity of miR-155 for the mobilization of HSPCs.14 However, the impact of miR-155 on homing in the context of human AML development is of yet undetermined. In order to translate our findings to AML patients, we further determined miR-155 levels in AML subtypes with elevated HOXA9 and MEIS1 transcript levels in addition to healthy donor BM cells. CD34+ cells exhibited significantly higher miR-155 levels compared to AML patients,

References 1. Miller CA, Wilson RK, Ley TJ. Genomic landscapes and clonality of de novo AML. N

254

suggesting that the profiled AML blasts were more differentiated than human HSPCs. Albeit high miR-155 expression levels were associated with an inferior overall survival in CN-AML samples49 and being upregulated in FLT3-ITD positive AML3,45,47 or French-American-British (FAB) M4/5 AML,53 we did not detect significantly altered miR-155 levels in sorted AML subpopulations. The fact that miR-155 is directly regulated by Hoxa9/Meis1 and accelerates Hoxa9/Meis1 AML, but is nevertheless dispensable for the initiation and progression of leukemia, suggests that miR-155 accelerates transformation rather than acting as a driver of leukemogenesis, supporting our previous findings in MLL-rearranged AML where the absence of miR-155 did not impact AML formation and progression.15 However, the function of miR-155 in AML still remains controversial, as Wallace et al. recently showed that miR-155 is relevant for human MV4-11 cell line growth38 as well as colony formation of AML samples with mutated FLT3 in vitro.13 Of note, inhibition of FLT3 signaling with SU14813 did not alter miR-155 expression in the same MV4-11 AML cell line,45 suggesting that the transcriptional activation of miR-155 may be independent of FLT3 signaling. MiR-155 has recently been included in a predictive 4-miRNA signature,39 which is in line with its leukemia-enhancing role, as shown in the context of Hoxa9 and Hoxa9/Meis1 in vivo. However, its potential as an in vivo driver has not yet been validated. Taken together, our results show that miR-155 accelerates, but is not required for the induction of AML and that its absence leads to impaired engraftment/homing of AML cells without affecting the onset of AML. Acknowledgments The authors would like to thank Carolin Ludwig, Britta Kopp, Ann Jansson, Carina Wasslavik and Sara Ståhlman for their technical assistance. We thank Vera Martins for technical expertise and helpful discussions. We thank Bernhard Gentner for providing viral vectors and constant support. We thank Tomer Itkin and Tvsee Lapidot for providing reagents and helpful discussions. We thank David Baltimore for providing us with the miR155 construct. Funding FK was supported by grants from Deutsche Krebshilfe grant 109420 (Max-Eder program); fellowship 2010/04 by the European Hematology Association; and by the Deutsche Forschungsgemeinschaft (DFG) (SFB 1074, project A5) and the Wilhelm Sander Stiftung (2015.153.1). AR was supported by the DFG (SFB 1074, project A5) as well as the gender equality program by the DFG (SFB 1074, project Z2), a fellowship from the Canadian Institutes of Health Research and the Baustein Startförderung Program of the Medical Faculty, Ulm University. LP was supported by grants from the Swedish Cancer Society (CAN2014/525), the Swedish Childhood Cancer Foundation (PR2014-0125), and Västra Götalandsregionen (ALFGBG431881). LB was supported in part by the German Research Foundation (Heisenberg-Stipendium BU 1339/3-1). The authors declare no conflict of interest.

Engl J Med. 2013;369(15):1473. 2. Papaemmanuil E, Dohner H, Campbell PJ. Genomic classification in acute myeloid leukemia. N Engl J Med. 2016;375(9):900901.

3. Jongen-Lavrencic M, Sun SM, Dijkstra MK, Valk PJ, Lowenberg B. MicroRNA expression profiling in relation to the genetic heterogeneity of acute myeloid leukemia. Blood. 2008;111(10):5078-5085.

haematologica | 2018; 103(2)

microRNA-155 amplifies leukemogenesis

4. Petriv OI, Kuchenbauer F, Delaney AD, et al. Comprehensive microRNA expression profiling of the hematopoietic hierarchy. Proc Natl Acad Sci USA. 2010; 107(35):1544315448. 5. Copley MR, Babovic S, Benz C, et al. The Lin28b-let-7-Hmga2 axis determines the higher self-renewal potential of fetal haematopoietic stem cells. Nat Cell Biol. 2013;15(8):916-925. 6. Starczynowski DT, Kuchenbauer F, Wegrzyn J, et al. MicroRNA-146a disrupts hematopoietic differentiation and survival. Exp Hematol. 2011;39(2):167-178.e164. 7. Eis PS, Tam W, Sun L, et al. Accumulation of miR-155 and BIC RNA in human B cell lymphomas. Proc Natl Acad Sci USA. 2005; 102(10):3627-3632. 8. O'Connell RM, Rao DS, Chaudhuri AA, et al. Sustained expression of microRNA-155 in hematopoietic stem cells causes a myeloproliferative disorder. J Exp Med. 2008;205(3):585-594. 9. Garzon R, Volinia S, Liu CG, et al. MicroRNA signatures associated with cytogenetics and prognosis in acute myeloid leukemia. Blood. 2008; 111(6): 3183-3189. 10. O'Connell RM, Taganov KD, Boldin MP, Cheng G, Baltimore D. MicroRNA-155 is induced during the macrophage inflammatory response. Proc Natl Acad Sci USA. 2007;104(5):1604-1609. 11. Thai TH, Calado DP, Casola S, et al. Regulation of the germinal center response by microRNA-155. Science. 2007; 316(5824):604-608. 12. Hu YL, Fong S, Largman C, Shen WF. HOXA9 regulates miR-155 in hematopoietic cells. Nucleic Acids Res. 2010; 38(16):5472-5478. 13. Wallace JA, Kagele DA, Eiring AM, et al. miR-155 promotes FLT3-ITD-induced myeloproliferative disease through inhibition of the interferon response. Blood. 2017;129(23):3074-3086. 14. Itkin T, Kumari A, Schneider E, et al. MicroRNA-155 promotes G-CSF-induced mobilization of murine hematopoietic stem and progenitor cells via propagation of CXCL12 signaling. Leukemia. 2017; 31(5):1247-1250. 15. Schneider E, Staffas A, Rohner L, et al. MicroRNA-155 is upregulated in MLLrearranged AML but its absence does not affect leukemia development. Exp Hematol. 2016;44(12):1166-1171. 16. Golub TR, Slonim DK, Tamayo P, et al. Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science. 1999; 286(5439):531-537. 17. Argiropoulos B, Palmqvist L, Yung E, et al. Linkage of Meis1 leukemogenic activity to multiple downstream effectors including Trib2 and Ccl3. Exp Hematol. 2008; 36(7):845-859. 18. Palmqvist L, Pineault N, Wasslavik C, Humphries RK. Candidate genes for expansion and transformation of hematopoietic stem cells by NUP98-HOX fusion genes. PLoS One. 2007;2(8):e768. 19. Quentmeier H, Dirks WG, Macleod RA, Reinhardt J, Zaborski M, Drexler HG. Expression of HOX genes in acute leukemia cell lines with and without MLL translocations. Leuk Lymphoma. 2004; 45(3):567574. 20. Kawagoe H, Humphries RK, Blair A, Sutherland HJ, Hogge DE. Expression of HOX genes, HOX cofactors, and MLL in phenotypically and functionally defined subpopulations of leukemic and normal

haematologica | 2018; 103(2)

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

human hematopoietic cells. Leukemia. 1999;13(5):687-698. Wong P, Iwasaki M, Somervaille TC, So CW, Cleary ML. Meis1 is an essential and rate-limiting regulator of MLL leukemia stem cell potential. Genes Dev. 2007; 21(21):2762-2774. Thorsteinsdottir U, Kroon E, Jerome L, Blasi F, Sauvageau G. Defining roles for HOX and MEIS1 genes in induction of acute myeloid leukemia. Mol Cell Biol. 2001;21(1):224234. 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. Pineault N, Buske C, Feuring-Buske M, et al. Induction of acute myeloid leukemia in mice by the human leukemia-specific fusion gene NUP98-HOXD13 in concert with Meis1. Blood. 2003;101(11):4529-4538. Kuchenbauer F, Mah SM, Heuser M, et al. Comprehensive analysis of mammalian miRNA* species and their role in myeloid cells. Blood. 2011;118(12):3350-3358. Zonari E, Pucci F, Saini M, et al. A role for miR-155 in enabling tumor-infiltrating innate immune cells to mount effective antitumor responses in mice. Blood. 2013; 122(2):243-252. Lorzadeh A, Bilenky M, Hammond C, et al. Nucleosome density ChIP-Seq identifies distinct chromatin modification signatures associated with MNase accessibility. Cell Rep. 2016;17(8):2112-2124. Argiropoulos B, Yung E, Xiang P, et al. Linkage of the potent leukemogenic activity of Meis1 to cell-cycle entry and transcriptional regulation of cyclin D3. Blood. 2010;115(20):4071-4082. Costinean S, Zanesi N, Pekarsky Y, et al. Pre-B cell proliferation and lymphoblastic leukemia/high-grade lymphoma in E(mu)miR155 transgenic mice. Proc Natl Acad Sci USA. 2006;103(18):7024-7029. Narayan N, Morenos L, Phipson B, et al. Functionally distinct roles foar different miR-155 expression levels through contrasting effects on gene expression, in acute myeloid leukemia. Leukemia. 2016;31(4): 808-820. Kogan SC, Ward JM, Anver MR, et al. Bethesda proposals for classification of nonlymphoid hematopoietic neoplasms in mice. Blood. 2002;100(1):238-245. Vergoulis T, Vlachos IS, Alexiou P, et al. TarBase 6.0: capturing the exponential growth of miRNA targets with experimental support. Nucleic Acids Res. 2012;40(Database issue):D222-229. Winnicka B, O'Conor C, Schacke W, et al. CD13 is dispensable for normal hematopoiesis and myeloid cell functions in the mouse. J Leukoc Biol. 2010;88(2):347359. Huang Y, Sitwala K, Bronstein J, et al. Identification and characterization of Hoxa9 binding sites in hematopoietic cells. Blood. 2012;119(2):388-398. Gibbs KD, Jr., Jager A, Crespo O, et al. Decoupling of tumor-initiating activity from stable immunophenotype in HoxA9-Meis1driven AML. Cell Stem Cell. 2012; 10(2):210-217. Zitzer NC, Ranganathan P, Dickinson BA, et al. Preclinical development of LNA antimir155 (MRG-106) in acute myeloid leukemia. Blood. 2015;126(23):3802-3802. Khalife J, Radomska HS, Santhanam R, et al. Pharmacological targeting of miR-155 via

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

the NEDD8-activating enzyme inhibitor MLN4924 (Pevonedistat) in FLT3-ITD acute myeloid leukemia. Leukemia. 2015;29(10):1981-1992. Wallace J, Hu R, Mosbruger TL, et al. Genome-wide CRISPR-Cas9 screen identifies microRNAs that regulate myeloid leukemia cell growth. PLoS One. 2016; 11(4):e0153689. Lechman ER, Gentner B, Ng SW, et al. miR126 regulates distinct self-renewal outcomes in normal and malignant hematopoietic stem cells. Cancer Cell. 2016;29(2):214-228. Verhaak RG, Goudswaard CS, van Putten W, et al. Mutations in nucleophosmin (NPM1) in acute myeloid leukemia (AML): association with other gene abnormalities and previously established gene expression signatures and their favorable prognostic significance. Blood. 2005;106(12):37473754. Yokoyama T, Nakatake M, Kuwata T, et al. MEIS1-mediated transactivation of synaptotagmin-like 1 promotes CXCL12/CXCR4 signaling and leukemogenesis. J Clin Invest. 2016;126(5):1664-1678. Mohr S, Doebele C, Comoglio F, et al. Hoxa9 and Meis1 cooperatively induce addiction to Syk signaling by suppressing miR-146a in acute myeloid leukemia. Cancer Cell. 2017;31(4):549-562.e511. Li Z, Huang H, Chen P, et al. miR-196b directly targets both HOXA9/MEIS1 oncogenes and FAS tumour suppressor in MLLrearranged leukaemia. Nat Commun. 2012; 21;3:688. Velu CS, Chaubey A, Phelan JD, et al. Therapeutic antagonists of microRNAs deplete leukemia-initiating cell activity. J Clin Invest. 2014;124(1):222-236. Garzon R, Garofalo M, Martelli MP, et al. Distinctive microRNA signature of acute myeloid leukemia bearing cytoplasmic mutated nucleophosmin. Proc Natl Acad Sci U S A. 2008;105(10):3945-3950. Gerloff D, Grundler R, Wurm AA, et al. NFkappaB/STAT5/miR-155 network targets PU.1 in FLT3-ITD-driven acute myeloid leukemia. Leukemia. 2015;29(3):535-547. Palma CA, Al Sheikha D, Lim TK, et al. MicroRNA-155 as an inducer of apoptosis and cell differentiation in Acute Myeloid Leukaemia. Mol Cancer. 2014;13:79. Chuang MK, Chiu YC, Chou WC, Hou HA, Chuang EY, Tien HF. A 3-microRNA scoring system for prognostication in de novo acute myeloid leukemia patients. Leukemia. 2015;29(5):1051-1059. Marcucci G, Maharry KS, Metzeler KH, et al. Clinical role of microRNAs in cytogenetically normal acute myeloid leukemia: miR155 upregulation independently identifies high-risk patients. J Clin Oncol. 2013; 31(17):2086-2093. Lee DW, Futami M, Carroll M, et al. Loss of SHIP-1 protein expression in high-risk myelodysplastic syndromes is associated with miR-210 and miR-155. Oncogene. 2012;31(37):4085-4094. Feinberg MW, Wara AK, Cao Z, et al. The Kruppel-like factor KLF4 is a critical regulator of monocyte differentiation. EMBO J. 2007;26(18):4138-4148. Norfo R, Zini R, Pennucci V, et al. miRNAmRNA integrative analysis in primary myelofibrosis CD34+ cells: role of miR155/JARID2 axis in abnormal megakaryopoiesis. Blood. 2014;124(13):e21-32. Xue H, Hua LM, Guo M, Luo JM. SHIP1 is targeted by miR-155 in acute myeloid leukemia. Oncol Rep. 2014;32(5):2253-2259.

255

ARTICLE

Acute Myeloid Leukemia

Ferrata Storti Foundation

Haematologica 2018 Volume 103(2):256-265

Allogeneic stem cell transplantation benefits for patients ≥ 60 years with acute myeloid leukemia and FLT3 internal tandem duplication: a study from the Acute Leukemia Working Party of the European Society for Blood and Marrow Transplantation Xavier Poiré,1 Myriam Labopin,2,3 Emmanuelle Polge,2,3 Jakob Passweg,4 Charles Craddock,5 Didier Blaise,6 Jan J. Cornelissen,7 Liisa Volin,8 Nigel H. Russell,9 Gérard Socié,10 Mauricette Michallet,11 Nathalie Fegueux,12 Patrice Chevallier,13 Arne Brecht,14 Mathilde Hunault-Berger,15 Mohamad Mohty,2,3* Jordi Esteve16* and Arnon Nagler2,17*

Section of Hematology, Cliniques Universitaires Saint-Luc, Brussels, Belgium; 2Acute Leukemia Working Party of the EBMT; 3Service d’Hématologie, Hôpital Saint-Antoine, Paris, France; 4Hematology, University Hospital, Basel, Switzerland; 5Center for Clinical Haematology, Queen Elizabeth Hospital, Birmingham, UK; 6Programme de Transplantation et Thérapie Cellulaire, Centre de Recherche en Cancérologie de Marseille, Institut Paoli Calmettes, France; 7Daniel den Hoed Cancer Centre, Erasmus Medical Center, Rotterdam, the Netherlands; 8Stem Cell Transplantation Unit, HUH Comprehensive Cancer Center, Helsinki, Finland; 9Nottingham City Hospital, UK; 10 Department of Hematology, Hôpital Saint-Louis, Paris, France; 11Service Hématologie, Centre Hospitalier Lyon Sud, France; 12Département d’Hématologie Clinique, CHU Lapeyronie, Montpellier, France; 13Département d’Hématologie, CHU Nantes, France; 14 Deutsche Klinik für Diagnostik, KMT Zentrum, Wiesbaden, Germany; 15Service des Maladies du Sang, CHRU, Angers, France; 16Hematology Department, IDIBAPS, Hospital Clinic, Barcelona, Spain and 17Chaim Sheba Medical Center, Tel-Hashomer, Israel ' 1

*MM, JE and AN contributed equally to this work

ABSTRACT

Correspondence: xavier.poire@uclouvain.be

Received: August 8, 2017. Accepted: December 7, 2017. Pre-published: December 14, 2017. doi:10.3324/haematol.2017.178251 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/256 ©2018 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.

256

ntermediate-risk cytogenetic acute myeloid leukemia with an internal tandem duplication of FLT3 (FLT3-ITD) is associated with a high risk of relapse, and is now a standard indication for allogeneic stem cell transplantation. Nevertheless, most studies supporting this strategy have been performed in young patients. To address the benefit of allogeneic transplantation in the elderly, we made a selection from the European Society for Blood and Marrow Transplantation registry of de novo intermediate-risk cytogenetic acute myeloid leukemia harboring FLT3-ITD in patients aged 60 or over and transplanted from a related or unrelated donor between January 2000 and December 2015. Two hundred and ninety-one patients were identified. Most patients received a reducedintensity conditioning (82%), while donors consisted of an unrelated donor in 161 (55%) patients. Two hundred and twelve patients received their transplantation in first remission, 37 in second remission and 42 in a more advanced stage of the disease. The 2-year leukemia-free survival rate was 56% in patients in first remission, 22% in those in second remission and 10% in patients with active disease, respectively (P<0.005). Non-relapse mortality for the entire cohort was 20%. In multivariate analysis, disease status at transplantation was the most powerful predictor of worse leukemia-free survival, graft-versus-host disease and relapse-free survival, and overall survival. In this elderly population, age was not associated with outcome. Based on the current results, allogeneic transplantation translates into a favorable outcome in fit patients ≥ 60 with FLT3-ITD acute myeloid leukemia in first remission, similarly to current treatment recommendations for younger patients.

Introduction Internal tandem duplication in the juxtamembrane domain of the tyrosine kinase receptor gene FLT3 (FLT3-ITD) is one of the most frequent recurrent mutations in acute myeloid leukemia (AML),1-3 and translates into early relapse and worse surhaematologica | 2018; 103(2)

Stem cell transplantation in elderly patients with FLT3-ITD AML

vival in young and older AML patients with normal karyotype or other intermediate-risk cytogenetics (IRC).4-6 Allogeneic stem cell transplantation (SCT), which has been shown to be beneficial in first remission (CR1) in most studies,3,7-11 has emerged as the best consolidation strategy in these patients. However, the vast majority of these studies were performed in patients under 60 years of age, transplanted with a myeloablative conditioning and using a matched sibling donor, while data for SCT in patients over 60 years of age harboring FLT3-ITD AML in CR1, especially with a reduced-intensity conditioning (RIC) regimen, are rather limited. The benefit of RIC SCT in AML patients with FLT3-ITD in CR1 has been observed in a previously reported smallscale single center study.12 Howbeit, the median age in this study was 55, ranging from 19 to 64 years, which may not represent a true elderly population. In a subsequent retrospective European Society for Blood and Marrow Translation (EBMT) study by Schmid et al., the authors confirmed the significant negative impact of FLT3-ITD on outcome.13 Moreover, in this cohort, which included patients of up to 71 years old, advanced age was found to be a significant negative factor, associated with worse leukemia-free survival (LFS) and increased non-relapse mortality (NRM).3,13 Nevertheless, the improvement in supportive care, human leukocyte antigen (HLA) typing and the development of new RIC regimens substantially reduce NRM, extend the eligibility criteria for SCT, and signify that age should no longer be a barrier to SCT.14,15 Albeit, relapse remains the major cause of treatment failure with RIC regimens.16,17 Due to the early relapse incidence (RI) in FLT3-ITD AML4 and the intrinsic chemoresistance and poor tolerance to therapy in elderly patients,18 the role of SCT in this older population may appear questionable. To evaluate the potential benefit of SCT in elderly patients with FLT3-ITD AML, we decided to conduct a retrospective study based on the EBMT registry in order to address the outcomes of FLT3-ITD AML in patients aged 60 or over and undergoing SCT.

Methods Patient selection and data collection Herein is a retrospective study performed by the Acute Leukemia Working Party (ALWP) of the EBMT group. The EBMT registry is a voluntary working group of more than 500 transplant centers, the participants of which are required to report all consecutive SCT and follow-up from their respective centers once a year. Patients aged 60 or over with a diagnosis of de novo AML transplanted between 1st January 2000 and 31st December 2015 with a related or unrelated donor (10/10 or 9/10) who were reported to the EBMT registry were included in this analysis. We selected only those patients with normal karyotype or other intermediate-risk karyotype. according to the European LeukemiaNet (ELN) classification,19,20 and harboring a FLT3-ITD mutation at the time of diagnosis. Patients with second SCT have been excluded, as have those who underwent a cord blood or haploidentical transplantation. All patients provided informed consent for the use of their data in retrospective studies. The Review Board of the ALWP as well as the ethic committee of the EBMT approved this study. A total of 291 patients from 100 centers met the criteria and were selected for further analysis. Myeloablative conditioning (MAC), RIC and non-myeloablative conditioning regimen (NMA) have been defined elsewhere.21 haematologica | 2018; 103(2)

The following variables were selected and included in the analysis: year of transplantation, age, sex, white blood cell count (WBC) at diagnosis, status at transplantation, time from diagnosis to CR, time from CR to SCT, the number of induction courses to reach CR, type of conditioning regimen, in vivo T-cell depletion (including both anti-thymocyte globulins and alemtuzumab), cytomegalovirus (CMV) status of donor and recipient, donor type, source of stem cells, Karnofsky performance status (KPS) at transplantation, engraftment, presence of acute and chronic graft-versus-host disease (GvHD), grade of acute GvHD, and NPM1 status. The molecular remission status at the time of SCT is center-dependent and was not defined in the registry.

Statistical analysis and endpoints definitions Endpoints included LFS, RI, NRM, overall survival (OS), acute and chronic GvHD, and GvHD-free/relapse-free survival (GRFS). All outcomes were measured from the time of transplant. LFS was defined as survival without relapse; patients alive without relapse were censored at the time of last contact. OS was based on death from any cause. NRM was defined as death without previous relapse. GRFS was defined as survival without grade 3-4 acute GvHD, extensive chronic GvHD, relapse or death. Surviving patients were censored at the time of last contact. The probabilities of OS, LFS and GRFS were calculated by the Kaplan-Meier test, and those of acute and chronic GvHD, NRM, and relapse were determined by the cumulative incidence estimator to accommodate competing risks. Results are expressed with a 95% confidence interval (CI). For NRM, relapse was the competing risk, while for relapse the competing risk was NRM. For acute and chronic GvHD, death without the event and relapse were the competing risks. For all prognostic analyses, continuous variables were categorized and the median was used as a cut-off point, excepting that of age which was analyzed as a continuous variable in multivariate analysis. A Cox proportional hazards model was used for multivariate regression. Factors associated with a P-value less than 0.15 by univariate analysis, and other clinically meaningful variables were included in the model. Results were expressed as the hazard ratio (HR) with 95% CI. All tests were two-sided. The type-1 error rate was fixed at 0.05 for the determination of factors associated with time-to-event outcomes. Statistical analyses were performed with SPSS 19 (SPSS Inc./IBM, Armonk, NY, USA) and R 3.0.1 (R Development Core Team, Vienna, Austria) software packages.

Results Patientsâ&#x20AC;&#x2122; characteristics Characteristics of the 291 selected patients are listed in Table 1. Median age at SCT was 63.7 years (range: 6075.4). Only 12 patients were over 70 years old. The most frequent RIC was fludarabine and busulfan (N=118), followed by fludarabine and melphalan (N=42). Twentythree patients, including 15 with active disease, one in second complete remission (CR2) and seven in CR1, received a fludarabine, amsacrine, and cytarabine (FLAMSA)-RIC preparative regimen.22 In vivo T-cell depletion (TCD) included 162 patients with anti-thymocyte globulins and 35 patients with alemtuzumab. At the time of SCT, most patients (252, 94%) had a KPS of more than 80%, and 71% had more than 90%. The characteristics of CR1 patients are summarized in Table 2. Molecular status at the time of SCT was available for 104 out of 202 CR1 patients; 80 (77%) were in molecular CR1. 257

X. Poiré et al.

Engraftment and graft-versus-host disease Engraftment was successful in 268 patients (98%) with a median time to neutrophils engraftment of 17 days (range: 6-64). The cumulative incidence of grade II-IV acute GvHD was 22% (95% CI: 17.7-27.6) and the 2-year cumulative incidence of chronic GvHD was 34% (95% CI: 28.4-40.5) (Figure 1). The cumulative incidence of grade III-IV acute GvHD was 8% (95% CI: 5-12) and the 2-year cumulative incidence of extensive chronic GvHD was 15% (95% CI: 10.5-19.8). In the multivariate analysis performed in the entire population, a lower performance status was associated with more grade III-IV acute GvHD (16% [95% CI: 8.2-25.2] vs.

Table 1. Patient characteristics of the entire cohort.

Patient’s characteristics N=291 Median age at SCT (range) Median follow-up (range) WBC at diagnosis (range) Median year of SCT Remission status at SCT, N(%) CR1 CR2 Not in CR Sex, N(%) Male Female Donor type, N(%) Sibling Unrelated Cytogenetics, N(%) Normal Abnormal NPM1 status, N(%) Unmutated Mutated Missing Source of SC BM PB In vivo T-cell depletion, N(%) Conditioning regimen, N(%) MAC RIC NMA Karnofsky > 80%, N(%) CMV patient+, N(%) CMV donor+, N(%) Co-morbidity score (HCT-CI) 0 1-2 3+ Missing

63.7 years old (60-75.4) 23 months (2-173) 44.0 x 109/L (1-575) 2012 (2002-2015) 212 (72.9%) 37 (12.7%) 42 (14.4%) 150 (51.5%) 141 (48.5%) 130 (44.7%) 161 (55.3%) 254 (87.3%) 37 (12.7%) 50 (24.6%) 153 (75.4%) 88 27 (9.3%) 264 (90.7%) 197 (68.4%) 52 (17.9%) 200 (68.7%) 39 (13.4%) 252 (94%) 199 (69%) 153 (53.3%) 65 (58%) 22 (19.6%) 25 (22.3%) 179

N: number; SCT: stem cell transplantation; WBC: white blood cell count; CR: complete remission; SC: stem cell; BM: bone marrow; PB: peripheral blood; MAC: myeloablative conditioning; RIC: reduced-intensity conditioning; NMA: non myeloablative conditioning; CMV: cytomegalovirus; HCT-CI: hematopoietic cell transplant co-morbidity index.

258

6% [95% CI: 3.2-6.2], for patients with KPS of <90% vs. ≥90%, respectively, HR=0.4, 95% CI: 0.17-0.93, P=0.03). The age of both the patient and donor, type of donor, source of stem cells, TCD and conditioning intensity were Table 2. CR1 patients’ characteristics.

Patient’s characteristics N=212 Median age at SCT (range) WBC at diagnosis (range) Median year of SCT Interval from diagnosis to CR1 (range) Interval from CR1 to SCT (range) Interval from diagnosis to SCT (range) Number of induction courses to CR1, N(%) 1 2 or more Missing Sex , N(%) Male Female Female to male, N(%) Donor type, N(%) Sibling Unrelated Cytogenetics, N(%) Normal Abnormal NPM1 status, N(%) Unmutated Mutated Missing Molecular CR at SCT No molecular CR Molecular CR Missing Source of SC BM PB In vivo T-cell depletion, N(%) Conditioning regimen, N(%) MAC RIC NMA Karnofsky > 80%, N(%) CMV patient+, N(%) CMV donor+, N(%) Co-morbidity score (HCT-CI) 0 1-2 3+ Missing

63.5 years old (60-72.4) 42.3 x 109/L (1-380) 2012 (2002-2015) 42 days (13-149) 98 days (15-300) 5 months (2-17) 126 (73.7%) 45 (26.3%) 41 110 (51.9%) 102 (48.1%) 41 (19.3%) 103 (48.6%) 109 (51.4%) 183 (86.3%) 29 (13.7%) 35 (23.5%) 114 (76.5%) 63 24 (23.1%) 80 (76.9%) 108 23 (10.9%) 189 (89.2%) 142 (67.6%) 33 (15.6%) 146 (68.9%) 33 (15.6%) 186 (96.4%) 140 (66.4%) 114 (54.3%) 47 (57.3%) 18 (22%) 17 (20.7%) 130

N: number; SCT: stem cell transplantation; WBC: white blood cell count; CR1: first complete remission; SC: stem cell; BM: bone marrow; PB: peripheral blood; MAC: myeloablative conditioning; RIC: reduced-intensity conditioning; NMA: non myeloablative conditioning; CMV: cytomegalovirus; HCT-CI: hematopoietic cell transplant co-morbidity index.

haematologica | 2018; 103(2)

Stem cell transplantation in elderly patients with FLT3-ITD AML

not significantly associated with the incidence of acute GvHD. Focusing on the 212 patients transplanted in CR1, only a better KPS (>90%) at SCT correlated with less grade II-IV and grade III-IV acute GvHD (HR=0.43, 95% CI: 0.22-0.82, P=0.01 and HR=0.13, 95% CI: 0.04-0.41, P=0.0005, respectively). Regarding chronic GvHD, no correlation was observed between the cumulative incidence of chronic GvHD and age, type of donor or source of stem cells. In the multivariate analysis performed in the entire population, better KPS was associated with more overall chronic GvHD (HR=1.95, 95% CI: 1.08-3.51, P=0.03), while TCD correlated with less overall chronic GvHD (HR=0.51, 95% CI: 0.32-0.83, P=0.006) and less extensive chronic GvHD (HR=0.29, 95% CI: 0.14-0.59, P<0.001). We also found significantly less extensive chronic GvHD with RIC (P=0.02), and more extensive chronic GvHD with a female donor to a male recipient (P=0.02). Among patients transplanted in CR1, we found a significant impact of donorâ&#x20AC;&#x2122;s age, with a higher incidence of chronic GvHD when an older donor was used (46% [95% CI: 35-56.9] vs. 31% [95% CI: 21.5-41.7] with a donor aged >47 and â&#x2030;¤47 years old, respectively, P=0.04).

GvHD was not associated with less relapse (HR=0.96, 95% CI: 0.48-19, P=0.9). Focusing on patients transplanted in CR1, a significant correlation between RI and the interval from diagnosis to CR1 (P=0.003) was demonstrated in univariate analysis, in line with the significant association between relapse and the number of induction courses to achieve CR1 (P<0.001). Thus, RI was 17.5% (95% CI: 10-26.9) when the interval from diagnosis to CR1 was less than 42 days, and 34.4% (95% CI: 24.1-44.9) when this interval was greater than 42 days; this difference was confirmed in multivariate analysis (HR: 2.32, 95% CI: 1.15-4.7, P=0.02). Being in molecular remission at the time of SCT was also significantly associated with less relapse in CR1 patients (17% vs. 44%, P=0.001). Five out of 24 patients with persistent molecular disease at the time of SCT received donor lymphocyte infusion (DLI) compared to 11 out of 80 patients with molecular remission (P=0.4). Increasing age (as a continuous variable), NPM1 status, type of donor, and conditioning intensity did not influence RI in multivariate analysis. TCD showed a trend toward less relapse in multivariate analysis (HR=0.53, 95% CI: 0.27-1.04, P=0.06). Molecular status at the time of SCT was not included in the multivariate analysis due to an excess of missing data (N=108).

Non-relapse mortality Overall survival, leukemia-free survival and graft-versus-host/relapse-free survival Among the 291 patients, the 2-year probability of OS was 46.7% (95% CI: 40.4-53.1). Disease status at SCT was the most powerful factor influencing survival, with a 2-year OS of 58.7% (95% CI: 51.2-66.1) in patients transplanted in CR1, 28.8% (95% CI: 12.3-45.4) in those transplanted in CR2, and only 9.5% (95% CI: 0.6-18.4) when transplantation was performed in active disease (P<0.001) (Figure 2C). In multivariate analysis, only disease status at the time of SCT (CR2 and active disease compared to CR1) was significantly associated with decreased OS

cGvHD

CI of chronic GvHD

The 2-year cumulative incidence of NRM for the whole cohort was 20% (95% CI: 15.6-25.4). In multivariate analysis, active disease at the time of SCT was significantly associated with increased NRM (P=0.01), while unrelated donors showed a trend toward a higher NRM (Table 3, Figure 2A). Thus, 2-year NRM was 18% (95% CI: 12.823.9) in CR1 patients and 29% (95% CI: 15.6-43) in patients with active disease who underwent transplantation (HR=2.38, 95%CI: 1.17-4.84, P=0.02). The presence of chronic GvHD was significantly associated with more NRM in multivariate analysis (HR=2.38, 95% CI: 1.045.49, P=0.04). In patients transplanted in CR1, the interval from CR1 to SCT was significantly associated with NRM in univariate analysis, being 7% (95% CI: 2.9-14.3) for patients transplanted within 98 days from CR1 and 24% (95% CI: 15.2-33.5) for patients transplanted more than 98 days from CR1. In multivariate analysis, a 9/10 unrelated donor was significantly associated with more NRM compared to a sibling donor, whereas NRM from a 10/10 unrelated donor SCT was comparable to a sibling donor (P=0.03 and P=0.42, respectively). Thus, 2-year NRM was 32% (95% CI: 12-53.8) when the SCT was performed with a 9/10 unrelated donor, 17% (95% CI: 9.7-25.3) when a sibling donor was used, and 16% (95% CI: 9.225.6) when the donor was a 10/10 unrelated donor. A longer interval from CR1 to SCT remained significantly associated with higher NRM in multivariate analysis (P=0.04) (Table 4).

Relapse incidence The 2-year cumulative RI in the overall series was 35.4% (95% CI: 29.6-41.3). RI strongly correlated with disease status at the time of SCT, at 26.1% (95% CI: 19.932.6), 56.8% (95% CI: 37.1-72.3) and 61.9% (95% CI: 44.9-77) for patients transplanted in CR1, CR2 and not in remission at the time of SCT, respectively (P<0.001) (Figure 2B). In multivariate analysis, both CR2 and active disease were significantly associated with increased RI compared to patients transplanted in CR1 (P<0.001 and P<0.001, respectively) (Table 3). The presence of chronic haematologica | 2018; 103(2)

Figure 1. Cumulative incidence of chronic graft-versus-host disease (GvHD). The 2-year cumulative incidence of chronic GvHD was 34% (95% CI: 28.4-40.5) in the entire cohort (N=291).

259

X. Poiré et al.

(P<0.001; Table 3). Chronic GvHD was not associated with OS (HR=1.14, 95% CI: 0.74-1.79, P=0.57). In the cohort of CR1 patients, the 2-year probability of OS was 69.1% (95% CI: 58.9-79.4) in patients with an interval from diagnosis to CR1 of less than 42 days, and 54.9% (95% CI: 43.5-66.2) in patients with a longer interval to diagnosis of CR1 (P=0.06). We also found that a donor age of more than 47 years old was significantly associated with improved OS (P=0.02), however, donor’s age was associated with donor type, being significantly older in HLA-identical siblings compared with unrelated donors (53 vs. 36 years old, P<0.001). When we compared the oldest sibling donors to the youngest unrelated donors according to median age in each group, we consistently found better OS with sibling donors, which confirms the stronger impact of donor type over donor’s age. In multivariate analysis adjusted for patients’ age, performance status, conditioning intensity, donor CMV status and

in vivo TCD, older sibling donors were still associated with better OS (HR=0.38, P=0.008) compared to younger unrelated donors. Time from CR1 to SCT was significantly longer for the youngest unrelated donors (107 days, range: 9-198) compared to the oldest sibling donors (83 days, range: 13-186, P=0.04). The 2-year probability of OS was 62.7% (95% CI: 52.2-73.2) after SCT from a sibling donor, 57.7% (95% CI: 46.3-69.1) after SCT from a 10/10 unrelated donor and 42.2% (95% CI: 17.4-67) after SCT from a 9/10 unrelated donor, respectively, but those differences did not reach statistical significance across groups (P=0.27). Age (> or < 65 years old), NPM1 status, molecular status at SCT, KPS, conditioning intensity, donor CMV positivity, and TCD were not correlated with OS in univariate analysis. In multivariate analysis, increasing patient’s age as a continuous variable was significantly associated with better OS (HR=0.89, 95% CI: 0.8-0.99, P=0.03), and SCT from 9/10 unrelated donors compared

Figure 2. Non-relapse mortality (NRM), relapse incidence (RI), overall survival (OS) and leukemia-free survival (LFS) per disease status (first complete remission (CR1), second remission (CR2) and active disease (Active D)). (A) The 2-year cumulative incidence of NRM was 18% (95% CI: 12.8-23.9) in CR1 patients, 21.6% (95% CI: 8.9-37.9) in CR2 patients and 28.6% (95% CI: 15.6-43) in Active D patients. (B) The 2-year cumulative incidence of relapse was 26.1% (95% CI: 19.9-32.6) in CR1 patients, 56.8% (95% CI: 37.1-72.3) in CR2 patients and 61.9% (95% CI: 44.9-75) in Active D patients. (C) The 2-year probability of OS was 58.7% (95% CI: 51.2-66.1) in CR1 patients, 28.8% (95% CI: 12.3-45.4) and 9.5% (95% CI: 0.6-18.4) in Active D patients. (D) The 2-year probability of LFS was 55.9% (95% CI: 48.663.3) in CR1 patients, 21.6% (95% CI: 6.4-36.8) and 9.5% (95% CI: 0.6-18.4).

260

haematologica | 2018; 103(2)

Stem cell transplantation in elderly patients with FLT3-ITD AML

to sibling donors showed a trend toward worse OS (P=0.09) (Table 4). The 2-year probability of LFS was 44.3% (95% CI: 38.150.5) for the whole patient cohort. Disease status had the strongest impact on LFS, with a 2-year LFS of 55.9% (95% CI: 48.6-63.3) in patients transplanted in CR1, 21.6% (95% CI: 6.4-36.8) if transplanted in CR2 and 9.5% (95% CI: 0.6-18.4) in patients not in CR at the time of SCT (P<0.001) (Figure 2D). In multivariate analysis, the significant influence of disease status at SCT was confirmed (P<0.001), and donor CMV positivity was also significant-

ly associated with worse LFS (P=0.04) (Table 3). Chronic GvHD did not correlate with LFS (HR=1.51, 95% CI: 0.922.48, P=0.1). In the cohort of patients transplanted in CR1, an interval from diagnosis to CR1 of less than 42 days was significantly associated with better LFS (64.6% vs. 52.5%, P=0.03). On the contrary, other variables such as age (> or < 65 years old), NPM1 status, TCD, donor CMV positivity or conditioning intensity did not show a prognostic impact on LFS. Similarly to OS, we observed a better LFS in SCT from donors aged over 47 years old in univariate (P=0.02) and multivariate analysis (P=0.01). The 2-year

Table 3. Multivariate analysis using a Cox proportional hazards model, N=291. Shown are variables with P<0.15 in univariate analysis. Nonrelapse mortality, relapse incidence, overall survival and leukemia-free survival. NRM

LFS

Age (per year) Status at SCT (CR1 as reference) CR2 Advanced Type of donor (MSD as reference) 10/10 UD 9/10 UD Karnoksky > 90% RIC Donor CMV+ In vivo TCD Age (per year) Status at SCT (CR1 as reference) CR2 Advanced Type of donor (MSD as reference) 10/10 UD 9/10 UD Karnoksky > 90% RIC Donor CMV+ In vivo TCD Age (per year) Status at SCT (CR1 as reference) CR2 Advanced Type of donor (MSD as reference) 10/10 UD 9/10 UD Karnoksky > 90% RIC Donor CMV+ In vivo TCD Age (per year) Status at SCT (CR1 as reference) CR2 Advanced Type of donor (MSD as reference) 10/10 UD 9/10 UD Karnoksky > 90% RIC Donor CMV+ In vivo TCD

0.57

0.97

0.89

1.07

0.36 0.02

1.55 2.38

0.61 1.17

3.90 4.84

0.06 0.11 0,15 0.39 0.07 0.90 0.23

1.85 2.02 0.66 0.75 1.76 1.04 0.96

0.97 0.85 0.37 0.38 0.95 0.56 0.89

3.52 4.82 1.16 1.47 3.26 1.92 1.03

0.00001 <0.00001

4.59 4.23

2.37 2.42

8.87 7.39

0.81 0.68 0.72 0.63 0.21 0.15 0.31

0.94 0.86 1.10 1.16 1.37 0.71 0.97

0.56 0.41 0.67 0.64 0.84 0.44 0.91

1.57 1.78 1.79 2.08 2.25 1.13 1.02

0.0004 <0.00001

2.64 3.35

1.53 2.13

4.55 5.26

0.17 0.26 0.41 0.53 0.06 0.43 0.19

1.34 1.40 0.85 1.16 1.47 0.85 0.96

0.88 0.79 0.58 0.72 0.99 0.58 0.91

2.03 2.48 1.25 1.87 2.19 1.26 1.02

0.00004 <0.00001

3.04 3.30

1.79 2.13

5.16 5.11

0.31 0.51 0.51 0.87 0.04 0.28

1.23 1.21 0.89 0.97 1.51 0.82

0.83 0.69 0.61 0.62 1.03 0.56

1.83 2.11 1.28 1.50 2.23 1.18

95% CI

N: number; NRM: non-relapse mortality; RI: relapse incidence; OS: overall survival; LFS: leukemia-free survival; HR: hazard ratio; CI: confidence interval; SCT: stem cell transplantation; CR1: first complete remission; CR2: second remission; UD: unrelated donor; MSD: matched sibling donor; CMV: cytomegalovirus; TCD: T-cell depletion; RIC: reduced-intensity conditioning.

haematologica | 2018; 103(2)

261

X. PoirĂŠ et al.

probability of LFS was 62.7% (95% CI: 52.2-73.2) after SCT from a sibling donor, 57.7% (95% CI: 46.3-69.1) after SCT from a 10/10 unrelated donor, and 42.2% (95% CI: 17.4-67) after SCT from a 9/10 unrelated donor, respectively, however, as with OS, those differences did not reach significance (P=0.27). In multivariate analysis, increasing patientâ&#x20AC;&#x2122;s age as a continuous variable and a shorter interval from diagnosis to CR1 were both significantly associated with better LFS (P=0.03 and P=0.05, respectively) (Table 4). The 2-year probability of GRFS was 32.3% (95% CI: 26.3-38.3) in the study population. A worse GRFS was seen with more advanced disease; 41.7% (95% CI: 34.349.2) in patients transplanted in CR1, 18.1% (95% CI: 3.632.6) and 2.4% (95% CI: 0-7.2) in patients with CR2 and active disease at the time of SCT, respectively (P<0.001) (Figure 3). Multivariate analysis performed on GRFS within the entire cohort and the CR1 patients is available in the Online Supplementary Material. We also focused on the two main conditioning regimens within CR1 patients, which were fludarabine and busulfan (N=95), followed by fludarabine and melphalan (N=30). We found no significant differences in terms of acute and chronic GvHD incidence, NRM, RI, OS and LFS between those two regimens in univariate analysis (data not shown).

Figure 3. Graft-versus-host disease and relapse-free survival (GRFS) per disease status (first complete remission (CR1), second remission (CR2) and active disease (Active D)). The 2-year probability of GRFS was 41.7% (95% CI: 34.349.2) in CR1 patients, 18.1% (95% CI: 3.6-32.6) in CR2 patients and 2.4% (95% CI: 0-7.2) in Active D patients.

Discussion SCT is becoming a routine standard of care consolidation strategy for younger patients with AML and FLT3ITD.3,8-11,23 Nonetheless, its potential benefit for older patients has not been specifically addressed, and there is currently no strong evidence which supports SCT for elderly patients with FLT3-ITD AML.13,24 Against this background, our study demonstrated that in patients with an age equal to or over 60 years old, SCT performed in CR1 translates into a 2-year OS and LFS of 59% and 56%, respectively. A NRM and RI rate of 18% and 25%, respectively, are acceptable in this population. Interestingly, these results are only slightly inferior than those reported in younger patients, and suggest the relevance of graft-versus-leukemia (GvL) for disease control in this entity.8,13,24 Moreover, increasing age was not associated with NRM and other outcomes within our population, probably reflecting a careful and adequate selection process in the elderly AML population submitted to SCT in CR1. On the contrary, we found very poor outcomes when SCT was performed beyond CR1, thus, the benefit of SCT in these situations remains questionable. The inferior results obtained in CR2 patients are in accordance with previous publications.25,26 Based on our sizable dataset, we strongly recommend that SCT is offered as the best consolidation strategy for eligible patients in early disease phase with AML and FLT3-ITD.27-29 It is possible that we have to concede a selection bias in our study, howbeit this bias supports the need for a thorough evaluation of each older candidate prior to SCT. We did not find any difference in characteristics between the youngest and oldest patients from our population, such as time from diagnosis to CR1, conditioning regimen, hematopoietic cell transplant co-morbidity index (HCTCI) or KPS. The superior OS and LFS observed among the oldest patients of our study may be explained by individ262

ual characteristics not reported in the registry. KPS and HCT-CI are well-described tools used to evaluate patients before SCT, and are reliable even in older patients.15,17 In addition to HCT-CI, other tools have been described in the assessment of elderly patients undergoing antileukemic therapy, and found that chronologic age is definitively not a limiting factor.16,30-32 This assessment included the functional, cognitive, biological, nutritional and medical evaluation of each patient, and helped us to discern the best candidate for intensive therapy or SCT.33,34 NPM1 mutation had no impact on any outcome parameter in this study. The favorable prognostic influence of NPM1 has been demonstrated in patients of up to 65 years of age, although it is less pronounced or even lost in older subjects.35,36 Several studies focused on younger patients have shown that NPM1 mutation may influence OS and LFS in FLT3-ITD AML,37-39 however, this impact is observed primarily in patients with a low allelic ratio of FLT3-ITD.37,39,40 In NPM1-mutated AML, only patients harboring a high level of FLT3-ITD may benefit unequivocally from SCT.41-44 Information regarding the FLT3-ITD allelic burden was not available in our study, and, given the lack of current standardization, it remains extremely difficult to analyze it in a multicenter registry setting.45 Moreover, the concurrent mutation of DNMT3A, frequently found in combination with FLT3-ITD and NPM1 mutation, may have a profound adverse prognostic impact, and the capacity for SCT to overcome this poor prognosis is currently unknown.2,46-48 We also found that patients transplanted in molecular remission at the time of SCT had a better outcome after SCT, with a decreased relapse risk and a trend toward improved LFS and GRFS, a fact which has been recently addressed by Gaballa et al.49 Nonetheless, the role of upfront SCT for patients who fail to achieve a molecular remission before SCT is unknown, and the benefit of donor lymphocyte infusion or other haematologica | 2018; 103(2)

Stem cell transplantation in elderly patients with FLT3-ITD AML Table 4. Multivariate analysis using a Cox proportional hazards model, N=212 (CR1 patients). Shown are variables with P<0.15 in univariate analysis. Non-relapse mortality, relapse incidence, overall survival and leukemia-free survival. NRM

LFS

Age (per year) Type of donor (MSD as reference) 10/10 UD 9/10 UD Kanofsky > 90% RIC Diagnosis to CR1 > 42 days CR1 to SCT > 98 days In vivo TCD Age (per year) Type of donor (MSD as reference) 10/10 UD 9/10 UD Kanofsky > 90% RIC Diagnosis to CR1 > 42 days CR1 to SCT > 98 days In vivo TCD Age (per year) Type of donor (MSD as reference) 10/10 UD 9/10 UD Kanofsky > 90% RIC Diagnosis to CR1 > 42 days CR1 to SCT > 98 days In vivo TCD Age (per year) Type of donor (MSD as reference) 10/10 UD 9/10 UD Kanofsky > 90% RIC Diagnosis to CR1 > 42 days CR1 to SCT > 98 days In vivo TCD

0.15

0.88

0.75

1.05

0.42 0.03 0.10 0.28 0.61 0.04 0.75 0.10

1.46 3.58 0.52 0.56 1.23 2.50 1.15 0.90

0.58 1.15 0.23 0.20 0.56 1.06 0.50 0.79

3.68 11.13 1.14 1.61 2.68 5.91 2.66 1.02

0.41 0.71 0.37 0.77 0.02 0.21 0.06 0.03

1.36 0.76 1.45 1.18 2.32 0.63 0.53 0.89

0.65 0.17 0.64 0.40 1.15 0.31 0.27 0.80

2.87 3.41 3.25 3.48 4.70 1.28 1.04 0.99

0.29 0.10 0.52 0.87 0.16 0.71 0.36 0.03

1.39 2.14 0.83 0.94 1.47 1.11 0.78 0.89

0.76 0.88 0.47 0.43 0.86 0.64 0.45 0.80

2.54 5.23 1.47 2.06 2.52 1.93 1.34 0.99

0.29 0.13 0.73 0.64 0.05 0.59 0.25

1.36 1.9 0.91 0.84 1.67 1.15 0.74

0.77 0.82 0.52 0.40 1.02 0.69 0.44

2.42 4.40 1.57 1.75 2.79 1.93 1.23

95% CI

N: number; NRM: non-relapse mortality; RI: relapse incidence; OS: overall survival; LFS: leukemia-free survival; HR: hazard ratio; CI: confidence interval; CR1: first complete remission; UD: unrelated donor; MSD: matched sibling donor; RIC: reduced-intensity conditioning; TCD: T-cell depletion.

maintenance therapy in this setting should be specifically addressed in future studies. A relevant issue concerning FLT3-ITD AML is the potential benefit of the use of FLT3 inhibitors in combination with intensive chemotherapy in patients undergoing SCT. In this regard, the addition of midostaurin to chemotherapy in newly diagnosed FLT3-ITD AML has been showed to significantly improve survival, even in patients undergoing SCT in CR1, suggesting that a deeper anti-leukemic response before SCT can translate into an improved outcome after transplant.50 Since information on the use of FLT3 inhibitors pre- and post-transplant was not available in this registry study, we were unable to specifically analyze their effects. The prevention of relapse after SCT8,13 with post-transplant maintenance therapy based on FLT3 inhibitors is an area of current preferential interest, and is being investigated via ongoing clinical trials using agents such as sorafenib, midostaurin or gilteritinib. Small retrospective studies on maintenance with sorafenib have resulted in reduced RI and improved survival withhaematologica | 2018; 103(2)

out increased toxicity.51-54 Nevertheless, SCT remains the best consolidation therapy offered to date, and the use of FLT3 inhibitors may only increase the proportion of patients, including frailer subjects, who might benefit from SCT. Herein, transplantation from 9/10 unrelated donors was associated with significant higher NRM and a trend toward inferior OS and LFS compared to sibling donors and 10/10 unrelated donors, as previously reported.55,56 However, an unrelated donor may be preferable due to the fact that older patients have older sibling donors, and donor age has been associated with decreased survival due to an excess of acute and chronic GvHD.57 Of note, improved survival has been observed with the use of younger unrelated donors compared to older sibling donors in an EBMT retrospective study,58 while another large study from the Center for International Blood and Marrow Transplant Research (CIBMTR) reported no difference in terms of outcomes between younger matched unrelated donors and older sibling donors.59 Among our 263

X. PoirĂŠ et al.

CR1 patients, we found that older donor age (>47 years old) was associated with more chronic GvHD, but better OS and LFS, and had no effect on RI. However, the age of the donor was strongly associated with the type of donor. The effect of donor type on outcomes was more potent than that of donor age, and our observations favor the use of a sibling donor, if eligible for stem cell donation. If a sibling donor is not available, a 10/10 fully matched unrelated donor is a suitable option, but caution must be applied with the use of a 9/10 unrelated donor. Nevertheless, we found a shorter interval from CR1 to SCT with sibling donors compared to older donors. We suggest an early donor search in order to further improve the results obtained with unrelated donors. TCD was significantly

References 1. Cancer Genome Atlas Research N, Ley TJ, Miller C, et al. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med. 2013;368(22):2059-2074. 2. Papaemmanuil E, Gerstung M, Bullinger L, et al. Genomic classification and prognosis in acute myeloid leukemia. N Engl J Med. 2016;374(23):2209-2221. 3. Schlenk RF, Dohner K, Krauter J, et al. Mutations and treatment outcome in cytogenetically normal acute myeloid leukemia. N Engl J Med. 2008;358(18):1909-1918. 4. Kottaridis PD, Gale RE, Frew ME, et al. The presence of a FLT3 internal tandem duplication in patients with acute myeloid leukemia (AML) adds important prognostic information to cytogenetic risk group and response to the first cycle of chemotherapy: analysis of 854 patients from the United Kingdom Medical Research Council AML 10 and 12 trials. Blood. 2001;98(6):1752-1759. 5. Singh H, Asali S, Werner LL, et al. Outcome of older adults with cytogenetically normal AML (CN-AML) and FLT3 mutations. Leuk Res. 2011;35(12):1611-1615. 6. Whitman SP, Maharry K, Radmacher MD, et al. FLT3 internal tandem duplication associates with adverse outcome and geneand microRNA-expression signatures in patients 60 years of age or older with primary cytogenetically normal acute myeloid leukemia: a Cancer and Leukemia Group B study. Blood. 2010;116(18):3622-3626. 7. Gale RE, Hills R, Kottaridis PD, et al. No evidence that FLT3 status should be considered as an indicator for transplantation in acute myeloid leukemia (AML): an analysis of 1135 patients, excluding acute promyelocytic leukemia, from the UK MRC AML10 and 12 trials. Blood. 2005; 106(10):3658-3665. 8. Brunet S, Labopin M, Esteve J, et al. Impact of FLT3 internal tandem duplication on the outcome of related and unrelated hematopoietic transplantation for adult acute myeloid leukemia in first remission: a retrospective analysis. J Clin Oncol. 2012; 30(7):735-741. 9. DeZern AE, Sung A, Kim S, et al. Role of allogeneic transplantation for FLT3/ITD acute myeloid leukemia: outcomes from 133 consecutive newly diagnosed patients from a single institution. Biol Blood

264

associated with less chronic GvHD and less extensive chronic GvHD, with no effect on OS and LFS. We did not find a beneficial effect of chronic GvHD on RI and LFS to support the existence of a GvL effect in our population. In conclusion, SCT emerges as a recommended consolidation strategy for fit patients aged 60 or over with AML and FLT3-ITD in CR1. Based on our study, and in view of the inferior results observed when SCT is performed in CR2 and beyond, we do not recommend postponing SCT until relapse. However, the few patients over 70 years of age included herein preclude firm recommendations for older patients. Sibling donors or fully matched unrelated donors remain the best donor choice in this older population.

Marrow Transplant. 2011;17(9):1404-1409. 10. Lin PH, Lin CC, Yang HI, et al. Prognostic impact of allogeneic hematopoietic stem cell transplantation for acute myeloid leukemia patients with internal tandem duplication of FLT3. Leuk Res. 2013; 37(3):287-292. 11. Meshinchi S, Arceci RJ, Sanders JE, et al. Role of allogeneic stem cell transplantation in FLT3/ITD-positive AML. Blood. 2006; 108(1):400; author reply 400-1. 12. Laboure G, Dulucq S, Labopin M, et al. Potent graft-versus-leukemia effect after reduced-intensity allogeneic SCT for intermediate-risk AML with FLT3-ITD or wildtype NPM1 and CEBPA without FLT3-ITD. Biol Blood Marrow Transplant. 2012; 18(12):1845-1850. 13. Schmid C, Labopin M, Socie G, et al. Outcome of patients with distinct molecular genotypes and cytogenetically normal AML after allogeneic transplantation. Blood. 2015;126(17):2062-2069. 14. Gooley TA, Chien JW, Pergam SA, et al. Reduced mortality after allogeneic hematopoietic-cell transplantation. N Engl J Med. 2010;363(22):2091-2101. 15. Sorror ML, Sandmaier BM, Storer BE, et al. Long-term outcomes among older patients following nonmyeloablative conditioning and allogeneic hematopoietic cell transplantation for advanced hematologic malignancies. JAMA. 2011;306(17):1874-1883. 16. Lim Z, Brand R, Martino R, et al. Allogeneic hematopoietic stem-cell transplantation for patients 50 years or older with myelodysplastic syndromes or secondary acute myeloid leukemia. J Clin Oncol. 2010;28(3):405-411. 17. Sorror ML, Storb RF, Sandmaier BM, et al. Comorbidity-age index: a clinical measure of biologic age before allogeneic hematopoietic cell transplantation. J Clin Oncol. 2014;32(29):3249-3256. 18. Klepin HD, Geiger AM, Tooze JA, et al. Geriatric assessment predicts survival for older adults receiving induction chemotherapy for acute myelogenous leukemia. Blood. 2013;121(21):4287-4294. 19. Dohner H, Estey E, Grimwade D, et al. Diagnosis and management of AML in adults: 2017 ELN recommendations from an international expert panel. Blood. 2017;129(4):424-447. 20. Dohner H, Estey EH, Amadori S, et al. Diagnosis and management of acute myeloid leukemia in adults: recommenda-

21.

22.

23.

24.

25.

26. 27.

28.

29.

30.

tions from an international expert panel, on behalf of the European LeukemiaNet. Blood. 2010;115(3):453-474. Bacigalupo A, Ballen K, Rizzo D, et al. Defining the intensity of conditioning regimens: working definitions. Biol Blood Marrow Transplant. 2009;15(12):16281633. Schmid C, Schleuning M, Ledderose G, et al. Sequential regimen of chemotherapy, reduced-intensity conditioning for allogeneic stem-cell transplantation, and prophylactic donor lymphocyte transfusion in high-risk acute myeloid leukemia and myelodysplastic syndrome. J Clin Oncol. 2005;23(24):5675-5687. Berman E, Maloy M, Devlin S, et al. Stem cell transplantation in adults with acute myelogenous leukemia, normal cytogenetics, and the FLT3-ITD mutation. Leuk Res. 2016;40:33-37. Oran B, Cortes J, Beitinjaneh A, et al. Allogeneic transplantation in first remission improves outcomes irrespective of FLT3ITD allelic ratio in FLT3-ITD-positive acute myelogenous leukemia. Biol Blood Marrow Transplant. 2016;22(7):1218-1226. Burnett AK, Goldstone A, Hills RK, et al. Curability of patients with acute myeloid leukemia who did not undergo transplantation in first remission. J Clin Oncol. 2013;31(10):1293-1301. Forman SJ, Rowe JM. The myth of the second remission of acute leukemia in the adult. Blood. 2013;121(7):1077-1082. Michelis FV, Messner HA, Atenafu EG, et al. Benefit of allogeneic transplantation in patients age >/= 60 years with acute myeloid leukemia is limited to those in first complete remission at time of transplant. Biol Blood Marrow Transplant. 2014;20(4):474-479. Walter RB, Sandmaier BM, Storer BE, et al. Number of courses of induction therapy independently predicts outcome after allogeneic transplantation for acute myeloid leukemia in first morphological remission. Biol Blood Marrow Transplant. 2015; 21(2):373-378. Deol A, Sengsayadeth S, Ahn KW, et al. Does FLT3 mutation impact survival after hematopoietic stem cell transplantation for acute myeloid leukemia? A Center for International Blood and Marrow Transplant Research (CIBMTR) analysis. Cancer. 2016;122(19):3005-3014. Brunner AM, Kim HT, Coughlin E, et al.

haematologica | 2018; 103(2)

Stem cell transplantation in elderly patients with FLT3-ITD AML

31.

32.

33.

34.

35.

36.

37.

38.

39.

Outcomes in patients age 70 or older undergoing allogeneic hematopoietic stem cell transplantation for hematologic malignancies. Biol Blood Marrow Transplant. 2013;19(9):1374-1380. Federmann B, Faul C, Meisner C, et al. Influence of age on outcome after allogeneic hematopoietic cell transplantation: a single center study in patients aged 60. Bone Marrow Transplant. 2015;50(3):427-431. Lim Z, Brand R, Martino R, et al. Allogeneic hematopoietic stem-cell transplantation for patients 50 years or older with myelodysplastic syndromes or secondary acute myeloid leukemia. J Clin Oncol. 2010; 28(3):405-411. Muffly LS, Boulukos M, Swanson K, et al. Pilot study of comprehensive geriatric assessment (CGA) in allogeneic transplant: CGA captures a high prevalence of vulnerabilities in older transplant recipients. Biol Blood Marrow Transplant. 2013;19(3):429434. Muffly LS, Kocherginsky M, Stock W, et al. Geriatric assessment to predict survival in older allogeneic hematopoietic cell transplantation recipients. Haematologica. 2014;99(8):1373-1379. Lazenby M, Gilkes AF, Marrin C, et al. The prognostic relevance of flt3 and npm1 mutations on older patients treated intensively or non-intensively: a study of 1312 patients in the UK NCRI AML16 trial. Leukemia. 2014;28(10):1953-1959. Ostronoff F, Othus M, Lazenby M, et al. Prognostic significance of NPM1 mutations in the absence of FLT3-internal tandem duplication in older patients with acute myeloid leukemia: a SWOG and UK National Cancer Research Institute/Medical Research Council report. J Clin Oncol. 2015;33(10):1157-1164. de Jonge HJ, Valk PJ, de Bont ES, et al. Prognostic impact of white blood cell count in intermediate risk acute myeloid leukemia: relevance of mutated NPM1 and FLT3-ITD. Haematologica. 2011;96(9):1310-1317. Gale RE, Green C, Allen C, et al. The impact of FLT3 internal tandem duplication mutant level, number, size, and interaction with NPM1 mutations in a large cohort of young adult patients with acute myeloid leukemia. Blood. 2008;111(5):2776-2784. Schnittger S, Bacher U, Kern W, et al. Prognostic impact of FLT3-ITD load in NPM1 mutated acute myeloid leukemia.

haematologica | 2018; 103(2)

Leukemia. 2011;25(8):1297-1304. 40. Pratcorona M, Brunet S, Nomdedeu J, et al. Favorable outcome of patients with acute myeloid leukemia harboring a low-allelic burden FLT3-ITD mutation and concomitant NPM1 mutation: relevance to postremission therapy. Blood. 2013; 121(14):2734-2738. 41. Ho AD, Schetelig J, Bochtler T, et al. Allogeneic stem cell transplantation improves survival in patients with acute myeloid leukemia characterized by a high allelic ratio of mutant FLT3-ITD. Biol Blood Marrow Transplant. 2016;22(3):462-469. 42. Schlenk RF, Kayser S, Bullinger L, et al. Differential impact of allelic ratio and insertion site in FLT3-ITD-positive AML with respect to allogeneic transplantation. Blood. 2014;124(23):3441-3449. 43. Versluis J, In 't Hout FE, Devillier R, et al. Comparative value of post-remission treatment in cytogenetically normal AML subclassified by NPM1 and FLT3-ITD allelic ratio. Leukemia. 2017;31(1):26-33. 44. Linch DC, Hills RK, Burnett AK,et al. Impact of FLT3(ITD) mutant allele level on relapse risk in intermediate-risk acute myeloid leukemia. Blood. 2014;124(2):273276. 45. Pratz KW, Levis M. How I treat FLT3mutated AML. Blood. 2017;129(5):565-571. 46. Ahn JS, Kim HJ, Kim YK, et al. DNMT3A R882 mutation with FLT3-ITD positivity is an extremely poor prognostic factor in patients with normal-karyotype acute myeloid leukemia after allogeneic hematopoietic cell transplantation. Biol Blood Marrow Transplant. 2016;22(1):6170. 47. Garg M, Nagata Y, Kanojia D, et al. Profiling of somatic mutations in acute myeloid leukemia with FLT3-ITD at diagnosis and relapse. Blood. 2015; 126(22):2491-2501. 48. Metzeler KH, Herold T, RothenbergThurley M, et al. Spectrum and prognostic relevance of driver gene mutations in acute myeloid leukemia. Blood. 2016;128(5):686698. 49. Gaballa S, Saliba R, Oran B, et al. Relapse risk and survival in patients with FLT3 mutated acute myeloid leukemia undergoing stem cell transplantation. Am J Hematol. 2017;92(4):331-337. 50. 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. 51. Antar A, Kharfan-Dabaja MA, Mahfouz R, et al. Sorafenib Maintenance Appears Safe and Improves Clinical Outcomes in FLT3ITD Acute Myeloid Leukemia After Allogeneic Hematopoietic Cell Transplantation. Clin Lymphoma Myeloma Leuk. 2015;15(5):298-302. 52. Brunner AM, Li S, Fathi AT, et al. Haematopoietic cell transplantation with and without sorafenib maintenance for patients with FLT3-ITD acute myeloid leukaemia in first complete remission. Br J Haematol. 2016;175(3):496-504. 53. Chen YB, Li S, Lane AA, et al. Phase I trial of maintenance sorafenib after allogeneic hematopoietic stem cell transplantation for fms-like tyrosine kinase 3 internal tandem duplication acute myeloid leukemia. Biol Blood Marrow Transplant. 2014; 20(12):2042-2048. 54. Metzelder SK, Schroeder T, Finck A, et al. High activity of sorafenib in FLT3-ITD-positive acute myeloid leukemia synergizes with allo-immune effects to induce sustained responses. Leukemia. 2012; 26(11):2353-2359. 55. Verneris MR, Lee SJ, Ahn KW, et al. HLA mismatch is associated with worse outcomes after unrelated donor reduced-intensity conditioning hematopoietic cell transplantation: an analysis from the Center for International Blood and Marrow Transplant Research. Biol Blood Marrow Transplant. 2015;21(10):1783-1789. 56. Piemontese S, Ciceri F, Labopin M, et al. A comparison between allogeneic stem cell transplantation from unmanipulated haploidentical and unrelated donors in acute leukemia. J Hematol Oncol. 2017;10(1):24. 57. Kollman C, Howe CW, Anasetti C, et al. Donor characteristics as risk factors in recipients after transplantation of bone marrow from unrelated donors: the effect of donor age. Blood. 2001;98(7):2043-2051. 58. Kroger N, Zabelina T, de Wreede L, et al. Allogeneic stem cell transplantation for older advanced MDS patients: improved survival with young unrelated donor in comparison with HLA-identical siblings. Leukemia. 2013;27(3):604-609. 59. Alousi AM, Le-Rademacher J, Saliba RM, et al. Who is the better donor for older hematopoietic transplant recipients: an older-aged sibling or a young, matched unrelated volunteer? Blood. 2013; 121(13):2567-2573

265

ARTICLE

Acute Lymphoblastic Leukemia

Ferrata Storti Foundation

WT1 loss attenuates the TP53-induced DNA damage response in T-cell acute lymphoblastic leukemia

Fulvio Bordin,1* Erich Piovan,1,2* Elena Masiero,2 Alberto Ambesi-Impiombato,3,4 Sonia Minuzzo,1 Roberta Bertorelle,2 Valeria Sacchetto,2 Giorgia Pilotto,1 Giuseppe Basso,5 Paola Zanovello,1,2 Alberto Amadori,1,2 and Valeria Tosello2

Dipartimento di Scienze Chirurgiche, Oncologiche e Gastroenterologiche, Università degli Studi di Padova, Italy; 2U.O.C. Immunologia e Diagnostica Molecolare Oncologica, Istituto Oncologico Veneto IOV - IRCCS, Padova, Italy; 3Institute for Cancer Genetics, Columbia University, New York, NY, USA; 4PsychoGenics Inc., Tarrytown, New York, NY, USA and 5Dipartimento di Salute della Donna e del Bambino, Università degli Studi di Padova, Italy 1

Haematologica 2018 Volume 103(2):266-277

*FB and EP contributed equally to this work.

ABSTRACT

Correspondence: valeria.tosello@iov.veneto.it

Received: April 12, 2017. Accepted: November 15, 2017. Pre-published: November 23, 2017. doi:10.3324/haematol.2017.170431 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/266 ©2018 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.

266

oss-of-function mutations and deletions in Wilms tumor 1 (WT1) gene are present in approximately 10% of T-cell acute lymphoblastic leukemia. Clinically, WT1 mutations are enriched in relapsed series and are associated to inferior relapse-free survival in thymic T-cell acute lymphoblastic leukemia cases. Here we demonstrate that WT1 plays a critical role in the response to DNA damage in T-cell leukemia. WT1 loss conferred resistance to DNA damaging agents and attenuated the transcriptional activation of important apoptotic regulators downstream of TP53 in TP53-competent MOLT4 T-leukemia cells but not in TP53-mutant T-cell acute lymphoblastic leukemia cell lines. Notably, WT1 loss positively affected the expression of the X-linked inhibitor of apoptosis protein, XIAP, and genetic or chemical inhibition with embelin (a XIAP inhibitor) significantly restored sensitivity to γ-radiation in both T-cell acute lymphoblastic leukemia cell lines and patient-derived xenografts. These results reveal an important role for the WT1 tumor suppressor gene in the response to DNA damage, and support the view that anti-XIAP targeted therapies could have a role in the treatment of WT1-mutant T-cell leukemia.

Introduction The Wilms tumor 1 (WT1) tumor suppressor gene plays an important role in embryonic development and tumorigenesis.1 The WT1 protein contains a N-terminal transactivation domain and a C-terminus with four zinc-fingers that can act as an activator or a repressor depending on the cell context.2,3 The complexity of the protein is augmented by the generation of alternative isoforms that translates into a different capacity of DNA-binding. In mammals, exon 5 and nine nucleotides at the end of exon 9 are subjected to alternative splicing, generating four different predominant splice isoforms. Alternative splicing of exon 5 gives rise to proteins that differ in the presence or absence of a 17-amino-acid insertion (Ex5+ and Ex5-). A different splicing, which cuts three amino acids [Lysine, Serine, Threonine, (KTS)] at 3’ end of exon 9, generates proteins which vary for the KTS insert (KTS+ and KTS-). These two isoforms (KTS+ and KTS−) are conserved in all vertebrates and imbalanced expression of these variants are associated to developmental abnormalities. Importantly, WT1 isoforms lacking the KTS insertion bind to DNA more efficiently, and significantly affect transcription (reviewed by Hastie4). Notably, WT1 has been demonstrated to interact with a variety of proteins, such as TP53, STAT3 and TET2, that play crucial roles in transformation.5-8 Mutations and deletions in the WT1 gene were first described in inherited and sporadic Wilms tumors, a pediatric malignancy resulting from the transformation of pluripotent embryonic renal precursor cells.9,10 Subsequently, WT1 gene mutations were also found in acute myeloid and bi-phenotypic leukemia subtypes.11 haematologica | 2018; 103(2)

WT1 role in DNA damage in T-cell leukemia

More recently, WT1 mutations and/or deletions were also reported in approximately 10% of both pediatric and adult T-cell acute lymphoblastic leukemia (T-ALL).12 Leukemia-associated WT1 mutations typically consist of heterozygous frameshift-generating deletions and insertions in exon 7 leading to premature stop codons which may ultimately result in truncated proteins lacking the Cterminal DNA-binding domain or in loss-of-function due to nonsense-mediated RNA decay.13 WT1 mutations are particularly prevalent in patients with relapsed T-ALL,14 and have been associated with inferior relapse-free survival in cases with standard risk thymic T-ALL.15 Here we describe a previously unrecognized direct mechanistic role of WT1 loss in the attenuation of DNA damage-induced apoptosis in T-ALL.

RPMI medium supplemented with 10% FCS (Gibco). T-ALL patient-derived xenografts (T-ALL PDX) had been previously established from pediatric T-ALL samples in non-obese/severe combined immunodeficiency mice (NOD/SCID).16,17 T-ALL PDX were expanded in vivo via intravenous (i.v.) injection into NODscidIL2Rnull immunodeficient mice (NSG mice, Jackson Laboratory). For prolonged in vitro culture, T-ALL xenografts were maintained in complete RPMI medium supplemented with 20% FCS, cytokines (10 ng/mL IL-7, 20 ng/mL FLT3-L, and 50 ng/mL SCF, all from Peprotech) and 20 nM insulin (Sigma Aldrich). Procedures involving animals and their care conformed with institutional guidelines and were authorized by the animal ethical committee (Italian Ministry of Health).

Statistical analysis

Methods

Results were expressed as mean value±Standard Deviation (SD). Unpaired Student t-test was used to analyze data. P<0.05 was considered statistically significant.

Cell lines and patient-derived xenografts

Results

MOLT4, PF382 and CCRF-HSB2 T-ALL cells and U2OS cells were obtained from the American Type Culture Collection (ATCC). The P12-Ichikawa T-ALL cells were from the German Collection of Microorganisms and Cell Cultures (Leibniz Institute DSMZ). T-ALL cell lines were cultured in vitro with complete

haematologica | 2018; 103(2)

WT1 alterations confer resistance to DNA damage in T-ALL cells Given the association of WT1 mutations and loss with relapsed T-ALL, we hypothesized that WT1 inactivation

Figure 1. WT1 mutations are associated with increased resistance to γ-radiationinduced apoptosis in T-ALL PDX. (A) Cell viability analysis of WT1 wild-type TP53 wildtype (wt-WT1/wt-TP53), WT1 wild-type TP53-mutant (wtWT1/mut-TP53), WT1-mutant TP53 wild-type (mut-WT1/wtTP53) and WT1-mutant TP53mutant (mut-WT1/mut-TP53) T-ALL PDX samples after 24 hours (h) from 0.5, 1, 2, 4 and 6 Gray of γ-radiation; analysis is shown as fold change (untreated cells fixed at 1). (B) Group column scatter analysis showing the specific apoptosis in wt-WT1/wt-TP53 and mutWT1/wt-TP53 T-ALL PDX after 24 h from 1 Gray of γ-radiation (*P<0.05). (C) Representative FACS analysis of apoptosis, using Annexin-V/Sytox Red staining of wt-WT1/wt-TP53 (n=2; sample ns. 8 and 12) and mut-WT1/wt-TP53 (n=2; sample ns. 48 and 51) T-ALL PDX; apoptosis analysis was performed at 24 h from 1 Gray of γ-radiation.

267

F. Bordin et al.

could result in impaired response to DNA damaging agents in this disease. To test this, we investigated the effects of γ-radiation in a panel of T-ALL patient-derived xenografts (T-ALL PDX) including both WT1 wild-type [sample ns. 8, 9, 10, 11, 12, 15, previously obtained from T-ALL cells at diagnosis; sample 46R, previously obtained from T-ALL cells at relapse (R)] and WT1-mutant (sample ns. 48 and 51, and sample ns. 47R and 51R). Consistent with previous reports, all WT1 mutations in these samples consisted of truncating nonsense or frameshift alterations in exon 7 (Table 1). Of note, only 2 of these specimens (samples 46R, WT1 wild-type and 47R, WT1-mutant) presented inactivating mutations in TP53 (Table 1). Additional data, such as NOTCH1, FBXW7 mutations and PTEN expression, that are frequently altered in T-ALL samples, showed that alterations were homogenously distributed amongst the WT1 wild-type and WT1-mutant

specimens (Online Supplementary Table S1). Cell viability assays in response to different doses of γ-radiation (0.5, 1, 2, 4 and 6 Gray) divided these T-ALL PDX into sensitive (median lethal dose = LD50<1.5 Gray) and resistant (LD50>1.5 Gray) (Figure 1A). Importantly, the T-ALL PDX resistant to DNA damage included all the WT1-mutant TP53 wild-type xenografts together with TP53-mutant samples 46R and 47R. Sample ns. 9 and 16 also clustered in the resistant group which, in contrast, did not present deletions in TP53 and WT1 loci, as assessed by Array-based Comparative Genomic Hybridization analysis (data not shown). In line with cell viability assays, apoptosis analysis at 24 hour (h) following 1 Gray of γ-radiation showed increased apoptosis in WT1 wild-type TP53 wild-type samples compared with WT1-mutant TP53 wild-type tumors (P<0.05) (Figure 1B and C). Overall, these results showed an association between

Figure 2. WT1 loss protects from DNA damage in MOLT4 T-ALL cells. (A) Western blot analysis of WT1-knockdown in MOLT4 cells (upper panel); Western blot quantification of WT1 protein is reported on top of the gel. Values are normalized with respect to the loading control and relative intensity signal calculated with respect to Control (Ctrl) cells (fixed at 1). Cell viability assay and apoptosis analysis in MOLT4 cells infected either with sh-Scramble (Ctrl) or shRNA-WT1 (WT1-KD) after 24 hours (h)-treatment with increasing doses of γ-radiation (left and right panels, respectively). Quantitative data are shown as mean±Standard Deviation (SD); assays were performed in triplicates and reproduced at least three times. *P<0.05; **P<0.005; ***P<0.001. (B) FACS analysis of active proliferation (S-phase, EDU positive) in Ctrl and WT1-KD MOLT4 cells through direct measurement of DNA synthesis. The Click-iT® EdU Flow Cytometry Assay kit was used in combination with cell cycle analysis performed with Sytox Red staining. Analysis was performed after 12 h from 2 Gray of γ-radiation. Analysis is shown as fold change (untreated cells fixed at 1). Quantitative data are shown as mean±SD; assays were performed in triplicates. Two independent experiments were performed. (Right) A representative experiment is shown. Square regions identify cells in active proliferation.

268

haematologica | 2018; 103(2)

WT1 role in DNA damage in T-cell leukemia

WT1 mutations and resistance to DNA damage, thus suggesting a putative role of WT1 in DNA damage response. In order to further investigate a potential mechanistic role of WT1 in DNA damage-induced apoptosis, we performed WT1-knockdown in MOLT4 T-ALL cells (Figure

2A, left panel). We used MOLT4 cells that, among several T-ALL cell lines tested, were reported to be wild-type for TP53 locus, even though results were contradictory.19 Sequencing analysis of our MOLT4 cells showed a heterozygous nonsense mutation in TP53 locus (R306*) that

Figure 3. WT1-knockdown affects the expression of apotosis-related proteins following γ-radiation and chemotherapy. (A) Whole cell extracts from Control (Ctrl) and WT1-knockdown (WT1-KD) MOLT4 cells after 24 hours (h) from 6 Gray of γ-radiation were hybridized to human apoptosis arrays. Apoptotic proteins that were consistently modulated are labeled in red (left panel). Immunoblot analysis of cleaved-Caspase3 and XIAP in Ctrl or WT1-KD MOLT4 cells; analysis was performed after 24 h from 6 Gray of γ-radiation. β−Actin is shown as loading control (right panel). Western blot quantification of XIAP protein is reported on top of the gel. Values are normalized with respect to the loading Ctrl and the numbers represent the relative intensity signal calculated respect to untreated Ctrl cells (fixed at 1). (B) Cell viability assays in Ctrl or WT1-KD MOLT4 cells after 24 h treatment with increasing doses of etoposide (left). Analysis is shown as fold change (untreated cells fixed at 1). Quantitative data are shown as mean±Standard Deviation (SD); assays were performed in triplicates and reproduced at least three times. Apoptosis analysis using AnnexinV/PI staining in Ctrl or WT1-KD MOLT4 cells after 24 h treatment with increasing doses of etoposide (middle panel). Error bars represent SD for triplicate experiments. *P<0.05; **P<0.005; ***P<0.001. Immunoblot analysis of PARP, cleaved-Caspase 3 and XIAP in Ctrl or WT1-KD MOLT4 cells; analysis was performed after 24 h treatment with 0.5 mM etoposide (right panel). β−Actin is shown as loading Ctrl. (C) Cell viability assay in Ctrl or WT1-KD MOLT4 cells after 24 h treatment with increasing doses of cytarabine (ARA-C) (0.05, 0.25, 0.5 and 1 mM), vincristine (0.01, 1, 10 and 100 ng/mL), or methotrexate (0.01, 0.1, 0.3 and 3 mM). Analysis is shown as fold change (untreated cells fixed at 1). Quantitative data are shown as mean±SD; assays were performed in triplicates. *P<0.05; **P<0.005; ***P<0.001.

haematologica | 2018; 103(2)

269

F. Bordin et al.

was not detectable at the cDNA level, as previously described.20 Analysis of cell viability in WT1-knockdown and shRNA scramble control cells treated with increased doses of γ-radiation showed significantly increased survival at 24 h in WT1-knockdown cells (Figure 2A, left panel). Similarly, analysis of γ-radiation-induced apoptosis showed significant protection from DNA damage-induced programmed cell death in WT1-knockdown T-ALL cells (Figure 2A, right panel). Moreover, analysis of cell cycle and DNA synthesis demonstrated a decreased number of cells in S phase in response to γ-radiation in Control but not in WT1-shRNA expressing cells (Figure 2B). These results were further validated using the CRIPSR-Cas9 strategy that allows us to directly target WT1 locus leading to complete WT1-knockout (Online Supplementary Figure S1A and B, left panels). Overall, these results implicate WT1 loss with increased resistance to DNA damage in T-ALL.

WT1 resistance to DNA damage results in increased expression of pro-survival factors To explore potential mechanisms implicated in the attenuation of DNA-damage-induced apoptosis following WT1 loss, we analyzed the expression of apoptosis-related proteins in Control and WT1-knockdown MOLT4 cells

270

following 6 Gray of γ-radiation. These analyses revealed increased cleaved-Caspase3 in WT1 control cells compared with WT1-knockdown samples (Figure 3A and Online Supplementary Figure S2A). However, early stages of DNA-damage response upstream of TP53 seem to be intact in the absence of WT1, as TP53 phosphorylation was detected in both WT1 Control and WT1-knockdown T-ALL lymphoblasts. In addition, we noted increased levels of pro-survival factors survivin and XIAP as well as upregulation of HMOX2 (HO-2), an important factor in protection against oxidative stress, in response to γ-radiation in WT1-knockdown cells, as demonstrated by peptide array quantification (Online Supplementary Figure S2A). Western blot analysis using different validated antibodies and independent lysates, documented the strong downregulation of XIAP in MOLT4 cells treated with γ-radiation and the effective and sustained abrogation of this effect upon WT1-knockdown (Figure 3A, right panel, and Online Supplementary Figure S2B). This result was further confirmed in WT1-knockout MOLT4 cells (Online Supplementary Figure S2C). Analysis of the DNA-damage response in response to etoposide, a topoisomerase II inhibitor, showed similar results. In fact, WT1-knockdown cells showed increased cell viability and decreased apop-

Figure 4. WT1 loss does not impair DNA damage recognition in MOLT4 cells. (A) Immunoblot of cleaved-Caspase3, P-TP53 (S15), Ac-TP53 (K382), and total TP53. β−Actin is shown as loading Control (Ctrl). (B) Immunoblot analysis of phosphorylated ataxia telangiectasia mutated (P-ATM; S1981), and P-CHK2 (T68) in Ctrl and WT1-KD MOLT4 cells following 1, 3 and 6 hours (h) from 6 Gray of γ-radiation. Tubulin is shown as loading Ctrl. Western blot quantification of P-TP53 (S15), Ac-TP53 (K382), TP53, PATM and P-CHK2 proteins is reported on top of the gel. Values are normalized with respect to the loading Ctrl. The numbers represent the relative intensity signal calculated with respect to untreated Ctrl cells (fixed at 1). For P-TP53, PATM and P-CHK2 proteins, that were not detected in untreated Ctrl cells, the relative intensity signal is calculated respected to 1 h (Ctrl cells fixed at 1). (C) Flow cytometry analysis of Histone H2AX phosphorylation (S139) in combination with PI incorporation and cell cycle analysis. Ctrl and WT1-KD cells were treated with 6 Gray of γ-radiation for 10 and 30 minutes. Three independent experiments were performed. One representative experiment is shown.

haematologica | 2018; 103(2)

WT1 role in DNA damage in T-cell leukemia

tosis compared with Controls when treated with increasing doses of etoposide (Figure 3B, left and middle panels). As in the case of γ-radiation, WT1 inactivation resulted in decreased cleaved-Caspase3 and impaired XIAP downregulation following etoposide treatment (Figure 3B, right panel). In line with these results, increased survival of WT1 deficient MOLT4 T-ALL cells was also observed after cytarabine (ARA-C), vincristine and methotrexate treatment (Figure 3C).

WT1 loss promotes cell survival by attenuating TP53 apoptotic response The TP53 pathway is extremely efficient in detecting DNA lesions in cells. Once induced, TP53 regulates the expression of a wide range of genes, leading to the biological outcomes of repair, growth arrest or apoptosis.21 We thus analyzed a possible interaction between WT1 and TP53 pathways in Control or WT1-knockdown MOLT4

T-ALL cells treated with γ-radiation. Western blot analysis of phosphorylated TP53 (P-TP53, S15), acetylated TP53 (Ac-TP53, K382) and total TP53 demonstrated induction and activation of TP53 protein in both Control and WT1-knockdown MOLT4 cells (Figure 4A). Importantly, at early time points, TP53 activation resulted more strongly induced in Control with respect to WT1-knockdown MOLT4 cells. These results were confirmed also using WT1-knockout cells (Online Supplementary Figure S3A). This effect did not depend on impaired recognition of DNA damage due to WT1 loss. Indeed, phosphorylation of crucial proteins involved in the initial phases of DNA damage recognition and repair subsequent to DNA double strand breaks such as ATM, CHK2 and histone H2A variant X (P-H2AX, S139; also known as γH2AX), was comparable between Control and knockdown cells treated with γ-radiation (Figure 4B and C, and Online Supplementary Figure S3B).

Figure 5. WT1-knockdown impairs TP53 DNA damage response in MOLT4 cells. (A) Relative expression of specific genes involved in the DNA-damage response. Analysis is performed in both Control (Ctrl) and WT1-KD cells after different times [3, 6 and 12 hours (h)] from 6 Gray of γ-radiation. Expression is calculated relative to untreated sh-Scramble MOLT4 cells (Ctrl 0 h) fixed as 1. (B) Western blot analysis of CDKN1A, BBC3, BAX and XIAP in Ctrl and WT1-KD MOLT4 cells after 1, 3, 6 and 12 h from γ-radiation. β−Actin is shown as loading control. Western blot quantification of CDKN1A, BBC3, BAX and XIAP proteins is reported on top of the gel. Values are normalized respect to the loading control. The numbers represent the relative intensity signal calculated respect to untreated Ctrl cells (fixed at 1). For CDKN1A and BBC3 proteins, that were not detected in untreated Ctrl cells, the relative intensity signal is calculated respected to 3 h and 1 h, respectively (Ctrl cells fixed at 1).

haematologica | 2018; 103(2)

271

F. Bordin et al.

Following these results, we analyzed the expression of genes involved in DNA repair, apoptosis and cell cycle progression in control and WT1-knockdown MOLT4 cells under basal conditions and at 3, 6 and 12 h after γ-radiation. These analyses revealed that WT1-knockdown

affected the expression of prominent genes involved in apoptosis, including BAX, BBC3, FAS and GADD45A (P<0.001) (Figure 5A). Similarly, upregulation of CDKN1A, a key TP53 target gene involved in cell cycle regulation, was also impaired in WT1-knockdown cells

272

Figure 6. WT1-knockdown does not affect survival in TP53-mutated T-cell acute lymphoblastic leukemia (T-ALL) cells. (A) Western blot analysis of WT1-knockdown in P12-Ichikawa and PF382 cells (top panels); cell viability analysis of Control (Ctrl) and WT1-knockdown P12-Ichikawa and PF382 cells after 24 hours (h) treatment with increasing doses of γ-radiation (middle panels). Analysis is shown as fold change (untreated cells fixed at 1). Western blot quantification of WT1 protein is reported on top of the gel. Values are normalized respect to the loading. The numbers represent the relative intensity signal calculated with respect to Ctrl cells (fixed at 1). Immunoblot analysis of PARP and XIAP in Ctrl and WT1-knockdown P12-Ichikawa and PF382 cells (lower panels); analysis was performed after 24 h from 6 Gray of γ-radiation. β−Actin is shown as a loading control. Western blot quantification of XIAP protein is reported on top of the gel. Values are normalized respect to the loading control. The numbers represent the relative intensity signal calculated respect to untreated Ctrl cells (fixed at 1). (B) Cell viability analysis in Ctrl or WT1-KD PF382 cells after 1, 3, 6, 12 and 24 h after 6 Gray of γ-radiation (left panel). Analysis is shown as fold change. Immunoblot of PARP, P-TP53 (S15), Ac-TP53 (K382), and total TP53. β−Actin is shown as loading control (right panel). Western blot quantification of P-TP53 (S15), Ac-TP53 (K382), and TP53, P-ATM and P-CHK2 proteins is reported on top of the gel. Values are normalized respect to the loading control. The numbers represent the relative intensity signal calculated respect to untreated Ctrl cells (fixed at 1). (C) Relative expression of specific genes involved in the DNA damage response. Analysis was performed in both Ctrl and WT1-KD after different times (3, 6 and 12 h) from 6 Gray of γ-radiation. Expression is calculated relative to untreated Ctrl PF382 cells (Ctrl 0 h) fixed as 1.

haematologica | 2018; 103(2)

WT1 role in DNA damage in T-cell leukemia treated with γ-radiation compared with WT1 scramble shRNA Control (P<0.001). Among these, the BBC3 proapoptotic factor showed markedly impaired protein upregulation in WT1-knockdown irradiated cells (Figure 5B). We thus hypothesized that WT1 directly regulates BBC3 expression co-operating with TP53 in the DNA damage response. To test this possibility, we analyzed the effects of overexpression of the WT1 isoforms on the activity of a WT1-regulatory element located in intron 1-2 of the BBC3 gene in a luciferase reporter assay. In these experiments, expression of both WT1-(KTS-) isoforms induced an approximate 9-fold increase of BBC3 reporter activity (Online Supplementary Figure S4). Consistent with data in literature (reviewed by Hastie4), overexpression of both WT1-(KTS+) isoforms determined a weak transcriptional response (approx. 2-fold induction). Luciferase activity was seen to be abolished following transfection of mutant WT1 (E384*) isoforms, the expression of which was found to be barely detectable by Western blot analysis. Overall, these data suggest that, in response to DNA damage, WT1 loss de-regulates prominent TP53 targets affecting cell survival.

Viability of TP53-mutant T-ALL cell lines is not significantly affected by WT1 loss when exposed to γ-radiation

In order to further validate the resistance to DNA damage conferred by WT1 loss in MOLT4 cells, we extended our analysis to additional T-ALL cell lines with known TP53 status (Online Supplementary Table S2). To this end, we infected CCRF-HSB2, which resulted wild-type for TP53 with a specific hairpin for WT1 (WT1-KD CCRFHSB2). These cells showed decreased levels of WT1 as

compared to control cells (Online Supplementary Figure S5A). Importantly, treatment of CCRF-HSB2 cells with increasing doses of γ-radiation and etoposide, showed that loss of WT1 conferred increased resistance to apoptosis, even if this was to a lesser extent than MOLT4 cells (P<0.005) (Online Supplementary Figure S5A and B). Notably, Western blot of XIAP protein showed higher levels upon WT1 loss at 1 mM etoposide (Online Supplementary Figure S5B, right panel). These experiments were also carried out on TP53-mutated T-ALL cell lines, PF382 and P12-Ichikawa, that presented deleterious mutations in the TP53 gene (Online Supplementary Table S2). Treatment of TP53-mutated P12-Ichikawa cells with increasing doses of γ-radiation revealed that WT1-knockdown cells were not protected from apoptosis compared to control cells (Figure 6A, top left panel). Molecular analysis showed that cleaved-PARP and XIAP were not significantly differentially expressed between Control and WT1knockdown cells upon DNA damage (Figure 6A, bottom left). Similar results were obtained in TP53-mutated PF382 cells (Figure 6A, right panel). As expected from a TP53-mutated cell line, PF382 cells displayed TP53 stabilization in the absence of DNA damage and failed to induce the transcription of crucial apoptotic genes involved in TP53 response, such as BAX, BBC3, FAS, and GADD45A, following γ-radiation treatment (Figure 6B and C). Consistently, loss of WT1 in PF382 cells did not alter this profile (Figure 6C). CDKN1A resulted similarly upregulated following γ-radiation in Control and WT1knockdown PF382 cells (Figure 6C). In conclusion, differently to MOLT4 and CCRF-HSB2 cells that displayed functional TP53 activation and WT1 loss increased cell viability following DNA damage, TP53-mutated T-ALL cell

Table 1. WT1 and TP53 genetic status in T-cell acute lymphoblastic leukemia PDX. HGVS-nomenclature was used for the description of sequence variants.18

T-ALL samples

WT1

TP53

8 9 10 11 12 15 16 46R 47R

wt wt wt wt wt wt wt wt Mut c. 1142C>A; p. S381*

Mut c.1109_1110insGGGCCGG; p.V371fs Mut c.1102_1104delinsCGGCCACTCCCCGGGGGTCC; p.V368fs Mut c.1102_1104delinsCGGCCACTCCCCGGGGGTCC; p.V368fs

wt wt wt wt wt wt wt c.743G>A; p.R248Q Mut c.708_709insC; p.M237fs c.638G>A; p.R213Q. wt

51R

wt: wild-type; Mut: mutant; R: *.

haematologica | 2018; 103(2)

273

F. Bordin et al.

lines, PF382 and P12 Ichikawa, were not significantly affected by WT1 loss when exposed to γ-radiation, suggesting a possible cross-talk between WT1 and functional TP53.

Increased resistance of WT1-deficient T-ALL cells to DNA damage can be rescued by pre-treatment with XIAP inhibitors Molecular characterization of DNA damage response in WT1 deficient MOLT4 cells showed an impaired TP53 response characterized by de-regulation of several proteins involved at different steps of the apoptotic cascade downstream of TP53 activation. To further validate these

results, we molecularly characterized the effects of γ-radiation on WT1 wild-type TP53 wild-type T-ALL PDX samples (n=2; sample ns. 8 and 12) and on WT1-mutated TP53 wild-type samples (n=2; sample ns. 48 and 51). All the tumors showed TP53 protein stabilization upon DNA damage (Figure 7A). Importantly, WT1 wild-type samples showed higher levels of cleaved-Caspase3 compared to WT1-mutant tumors. Following γ-radiation, BAX, survivin and HO-2 proteins were not significantly regulated either in WT1 wild-type nor in WT1-mutated T-ALL xenograft samples, while BBC3 was not differentially regulated between WT1 wild-type and WT1-mutated T-ALL, with WT1-mutated T-ALL PDX sample n. 51 undergoing a

274

Figure 7. XIAP is a critical player in WT1 resistance to DNA damage in T-cell acute lymphoblastic leukemia (T-ALL) cells. (A) Western blot analysis of TP53, cleaved-Caspase3, BBC3, BAX, survivin, HO-2 and XIAP in wt-WT1 (n=2; sample ns. 8 and 12) and mut-WT1 (n=2; sample ns. 48 and 51) TALL PDX following 3 hours (h) from 6 Gray of γ-radiation. Histogram plots representing Western blot quantification of TP53, cleaved-Caspase3, BBC3, BAX and XIAP proteins is reported on the right. Values are normalized respect to the loading control (Ctrl). The plots show the relative intensity signal calculated with respect to untreated Control cells (fixed at 1). (B) Apoptosis assay in WT1-mutated PDX sample n. 48 pre-treated for 24 h with 15 μM embelin and subjected to γ-radiation or etoposide treatment (0.5 Gray and 0.1 μΜ, respectively) in the presence or absence of embelin. (C) Apoptosis assay in MOLT4 cells pre-treated for 24 h with 7.5 mM or 10 mM embelin and subjected to γ-radiation or etoposide treatment (2 Gray and 0.2 mM, respectively). Analysis is shown at 48 h following γ-radiation and presented as percentage of apoptotic cells. *P<0.05; **P<0.005; ***P<0.001.

haematologica | 2018; 103(2)

WT1 role in DNA damage in T-cell leukemia

strong induction of BBC3 (Figure 7A and Supplementary Figure S6). On the other hand, XIAP protein resulted grossly down-regulated following γ-radiation in WT1 wild-type but not in WT1-mutant T-ALL PDX, in line with results obtained in MOLT4 and CCRF-HSB2. XIAP is one of the best characterized members of the inhibitor of apoptosis (IAP) family proteins and is considered to be a key regulator of apoptosis. We thus investigated whether inhibition of XIAP could contribute to rescue the effects of WT1 loss in T-ALL cells upon DNA damage. T-ALL PDX samples were treated with increasing doses of embelin, a natural compound that binds to XIAP and promotes apoptosis.22 Amongst 5 T-ALL PDX samples analyzed, 2 resulted resistant at 30 mM embelin (n. 12 and n. 15) whereas TALL PDX sample ns. 8, 48 and 51 showed decreased viability and increased apoptosis starting from 15 mM of embelin (Online Supplementary Figure S7). We thus pretreated WT1-mutated PDX sample n. 48 with 15 mM of embelin, followed by treatment with γ-radiation and etoposide at 0.5 Gray and 0.1 mM etoposide, respectively. Apoptosis analysis showed a significant rescue of resistance to both γ-radiation and etoposide treatments (P<0.001) (Figure 7B). Importantly, a complete rescue was achieved by γ-radiation and etoposide treatments in MOLT4 cells at 10 mM embelin. Indeed, apoptosis analysis revealed that pre-treatment with 7.5 mM and especially 10 mM embelin before DNA damaging agents (γ-radiation and etoposide), resulted in a significant increase in apoptosis with respect to that induced by γ-radiation or etoposide alone in WT1-knockdown MOLT4 cells (P<0.001) (Figure 7C). Notably, XIAP-knockdown in MOLT4 cells deficient for WT1 (WT1-KD) efficiently rescued resistance to apoptosis in the presence of increasing doses of etoposide (P<0.001 at 0.25 mM etoposide) (Online Supplementary Figure S8). Overall, these results identified XIAP as a critical player in regulating DNA damage resistance induced by WT1 loss, opening new therapeutic opportunities in WT1-mutant patients.

Discussion The introduction of intensive combination chemotherapy protocols in T-ALL treatment has led to remarkable improvements in survival.23 Unfortunately, primary resistance to chemotherapy or acquired resistance are strongly associated with poor outcome and the molecular mechanisms that allow leukemia cells to resist chemotherapy are not yet completely understood. A recent study has demonstrated that refractory T-ALL clones can develop mutations in drug-resistance genes under chemotherapy pressure.24 Moreover, resistant clones may originate from minor sub-clones present in patients at diagnosis through a dramatic selection during chemotherapy.25,26 A better understanding of the genetic lesions that may contribute to chemo-resistance is imperative to improve treatment protocols and for the identification of more effective antileukemic drugs. In particular, it is necessary to better characterize genetic alterations associated to genes with poorly characterized function. The WT1 gene is mutated in approximately 10% of TALL and AML cases, and these alterations are mainly heterozygous frameshift mutations that cluster in exon 7 and are predicted to lead to a truncated protein which lacks the zinc finger domain and the property to bind DNA.11,12,27 haematologica | 2018; 103(2)

Monoallelic or subclonal deletions have also been observed in some cases in association with mutations.12,28,29 Even if WT1 alterations clearly suggest a role as a tumor suppressor mediated at least in part by de-regulation of its transcriptional activity, its role in T-ALL is still poorly understood. Most studies have focused their attention on acute myeloid leukemia (AML). Recently, WT1 inactivating mutations have been found to inversely correlate with TET2/IDH1/IDH2 mutations; indeed, WT1-mutant AML patients have reduced 5-hydrossy-methyl-cytosine levels, similar to TET2/IDH1/IDH2 mutant AML, suggesting a critical role for WT1 as an epigenetic regulator.7 Moreover, always in the context of AML, WT1 mutations were found to be associated with an increased risk of relapse and an inferior outcome, even if contradictory results have been reported,30-36 probably due to the difference between the AML subgroups analyzed. Interestingly, a case report of AML demonstrated that a minor sub-clone, characterized by a mutation in the WT1 gene, emerges as a dominant clone after repetitive chemotherapy, including stem cell transplant.37 In T-ALL, studies on the role of WT1 mutations are still limited. In pediatric and adult T-ALL, the presence of WT1 mutations in the entire cohort were not predictive of poor clinical outcome.12,27 Interestingly, one study observed that within the standard risk group of thymic T-ALL, the small group of WT1-mutant patients had inferior relapse-free survival compared to wild-type patients.15 Notably, a very recent study in a group of highrisk T-ALL composed of paired diagnosis and relapse samples reported that WT1 mutations resulted highly enriched with respect to unselected T-ALL cases.14 Moreover, WT1 was shown to be among the frequently altered genes in the Early T Progenitor ALL (ETP) subgroup.38 ETP ALL were originally identified as a high-risk subtype of leukemia that lacks several T-cell markers and is characterized by aberrant expression of myeloid and stem cell markers.39 Our study highlights an important role for WT1 as a critical regulator of the TP53-dependent apoptotic response following DNA damage, and indicates an association between WT1 loss-of-function and resistance both in TALL PDX and MOLT4 T-ALL cells. In fact, in the presence of DNA-damaging stimuli, such as γ-radiation and chemotherapy, loss of WT1 in MOLT4 cells conferred increased survival associated with an impaired induction of TP53 apoptotic response. Amongst the de-regulated genes, some of which have been previously described as WT1 targets, such as BAX and GADD45A, we identified BBC3 as a new important WT1 target. BBC3, which encodes for the pro-apoptotic factor Puma, is a potent antagonist of anti-apoptotic BCL2-related proteins contributing to the release and activation of the pro-apoptotic factors BAX and BAK. Although BBC3 was found to be a critical regulator of apoptosis in primary thymocytes in response to DNA damage and glucocorticoid treatment40 and a crucial factor in MOLT4 cells following WT1 loss, one mut-WT1 T-ALL xenograft in our series (n. 51) showed strong induction of BBC3 much like the wt-WT1 PDX samples, suggesting that other pro-apoptotic factors may be important in this sample. An analysis of the WT1-knockout mouse first pointed out the involvement of WT1 in regulating apoptosis, which demonstrated that lack of WT1 expression determined severe abnormalities of renal development and that the metanephric blastema of WT1-null embryos was char275

F. Bordin et al.

acterized by massive apoptosis compared to wild-type embryos.41 A number of subsequent studies have demonstrated that WT1 can directly or indirectly regulate several BCL-2 family members including pro-apoptotic factors, such as BAK and BAX, and the anti-apoptotic proteins BCL2 and BCL2A1 with different outcomes depending on the cell context (reviewed by Toska and Roberts2 and by Yang et al.3). Contradictory results regarding WT1 regulation may be due in part to the complex structure of the WT1 protein and its multiple isoforms. Another level of complexity is due to co-operativity between WT1 and its interacting proteins. WT1 was found to physically interact with TP53, and this interaction was demonstrated to modulate the ability of WT1 and TP53 to trans-activate their respective targets.5,42,43 In this scenario, it is plausible that the cellular context may substantially determine WT1 effects on specific targets. In our study, in fact, WT1 significantly directly or indirectly affects the transcription of important mediators of TP53-apoptotic response only in the presence of a functional TP53. MOLT4 cells carry a heterozygous nonsense mutation in TP53 locus (p.R306*) that was not detectable at the cDNA level as previously described by others;20 concerning this trait, unlike the majority of T-ALL cell lines, MOLT4 cells resemble primary T-ALL samples at diagnosis as they rarely display deleterious mutations in the TP53 gene.44 Indeed, treatment of MOLT4 cells with DNA-damaging stimuli determines TP53 stabilization and strong induction of TP53specific target genes. This response is significantly downregulated when MOLT4 cells are engineered to loose WT1, as in knockdown or knockout experiments. Our data suggest that WT1 loss can influence both stabilization and transcriptional activity of TP53. On the other hand, we found that TP53-mutated T-ALL cell lines, such as PF382 and P12-Ichikawa, were not significantly affected by WT1 loss in response to DNA damage, suggesting a possible link between TP53 and WT1, that remains a subject for future studies. The effects on TP53 response mediated by WT1 loss in MOLT4 cells also resulted in deregulation of downstream effectors such as XIAP. In fact, we found elevated levels of XIAP following γ-radiation and etoposide treatment in WT1-deficient MOLT4 cells. Importantly, pre-treatment of WT1-deficient MOLT4 cells

References 1. Miller-Hodges E, Hohenstein P. WT1 in disease: shifting the epithelial-mesenchymal balance. J Pathol. 2012;226(2):229-240. 2. Toska E, Roberts SG. Mechanisms of transcriptional regulation by WT1 (Wilms' tumour 1). Biochem J. 2014;461(1):15-32. 3. Yang L, Han Y, Suarez Saiz F, Minden MD. A tumor suppressor and oncogene: the WT1 story. Leukemia. 2007;21(5):868-876. 4. Hastie ND. Life, sex, and WT1 isoforms-three amino acids can make all the difference. Cell. 2001;106(4):391-394. 5. Maheswaran S, Park S, Bernard A, et al. Physical and functional interaction between WT1 and p53 proteins. Proc Natl Acad Sci USA. 1993;90(11):5100-5104. 6. Rong Y, Cheng L, Ning H, et al. Wilms' tumor 1 and signal transducers and activators of transcription 3 synergistically promote cell proliferation: a possible mecha-

276

10.

11.

with embelin, a natural compound that specifically inhibits XIAP22,45 and direct inactivation of XIAP using specific shRNAs, rescued the resistance of WT1-deficient MOLT4 towards γ-radiation and etoposide. Notably, in TALL PDX samples, as for MOLT4 cells, XIAP stability was maintained following DNA damage in WT1-mutant samples but not in WT1 wild-type xenografts, confirming a prominent role of XIAP downstream of WT1 resistance in T-ALL. In addition, also WT1-mutant T-ALL xenografts were seen to be sensitized to γ-radiation and etoposide when the treatments were performed in the presence of the XIAP inhibitor embelin. XIAP acts by inhibiting caspase activation at an important point where intrinsic and extrinsic apoptotic pathways converge, making it an attractive therapeutic target. Pre-clinical studies concerning the use of embelin and small molecule XIAP inhibitors have demonstrated antitumor effects in xenograft and mouse models.46-49 Importantly, no specific toxicity was reported following XIAP inhibition in these studies.46 In conclusion, our study unveils an important role for the WT1 tumor suppressor gene in dampening TP53 DNA damage response and suggest that the combination of XIAP inhibitors with conventional chemotherapy or radiotherapy in T-ALL patients carrying WT1 mutations might be beneficial. Acknowledgments We are grateful to W. Pear (Perelman School of Medicine, University of Pennsylvania, Philadelphia) for providing MigR1 empty vector; E. Peta for performing some molecular analyses; S. Indraccolo for providing patient-derived xenografts (PDX); V. Agnusdei and M. Pinazza for maintaining PDX samples in immunodeficient mice; Cinzia Candiotto for TP53 mutational analysis in T-ALL cell lines; R. Grancara for computational assistance. Funding This work was supported by the Italian Association for Cancer Research (AIRC) grant to VT (MFAG #13053), to AA (IG#14032), to PZ (IG #14256) and to GB (IG#19186); Ministero dell’Istruzione, dell’Università e della Ricerca (MIUR) Ex 60% to E.P; Progetto di Ricerca di Ateneo (PRAT; #152403) to EP; Istituto Oncologico Veneto 5x1000 fund.

nism in sporadic Wilms' tumor. Cancer Res. 2006;66(16):8049-8057. Rampal R, Alkalin A, Madzo J, et al. DNA hydroxymethylation profiling reveals that WT1 mutations result in loss of TET2 function in acute myeloid leukemia. Cell Rep. 2014;9(5):1841-1855. Wang Y, Xiao M, Chen X, et al. WT1 recruits TET2 to regulate its target gene expression and suppress leukemia cell proliferation. Mol Cell. 2015;57(4):662-673. Call KM, Glaser T, Ito CY, et al. Isolation and characterization of a zinc finger polypeptide gene at the human chromosome 11 Wilms' tumor locus. Cell. 1990; 60(3):509-520. Gessler M, Poustka A, Cavenee W, Neve RL, Orkin SH, Bruns GA. Homozygous deletion in Wilms tumours of a zinc-finger gene identified by chromosome jumping. Nature. 1990;343(6260):774-778. King-Underwood L, Pritchard-Jones K.

12. 13.

14.

15.

Wilms' tumor (WT1) gene mutations occur mainly in acute myeloid leukemia and may confer drug resistance. Blood. 1998; 91(8):2961-2968. Tosello V, Mansour MR, Barnes K, et al. WT1 mutations in T-ALL. Blood. 2009; 114(5):1038-1045. Abbas S, Erpelinck-Verschueren CA, Goudswaard CS, Lowenberg B, Valk PJ. Mutant Wilms' tumor 1 (WT1) mRNA with premature termination codons in acute myeloid leukemia (AML) is sensitive to nonsense-mediated RNA decay (NMD). Leukemia. 2010;24(3):660-663. Oshima K, Khiabanian H, da Silva-Almeida AC, et al. Mutational landscape, clonal evolution patterns, and role of RAS mutations in relapsed acute lymphoblastic leukemia. Proc Natl Acad Sci USA. 2016; 113(40): 11306-11311. Heesch S, Goekbuget N, Stroux A, et al. Prognostic implications of mutations and

haematologica | 2018; 103(2)

WT1 role in DNA damage in T-cell leukemia

16.

17.

18.

19.

20.

21.

22.

23. 24.

25.

26.

27.

expression of the Wilms tumor 1 (WT1) gene in adult acute T-lymphoblastic leukemia. Haematologica. 2010;95(6):942949. Agnusdei V, Minuzzo S, Frasson C, et al. Therapeutic antibody targeting of Notch1 in T-acute lymphoblastic leukemia xenografts. Leukemia. 2014;28(2):278-288. Pinazza M, Borga C, Agnusdei V, et al. An immediate transcriptional signature associated with response to the histone deacetylase inhibitor Givinostat in T acute lymphoblastic leukemia xenografts. Cell Death Dis. 2016;6:e2047. den Dunnen JT, Dalgleish R, Maglott DR, et al. HGVS Recommendations for the Description of Sequence Variants: 2016 Update. Hum Mutat. 2016;37(6):564-569. Leroy B, Girard L, Hollestelle A, Minna JD, Gazdar AF, Soussi T. Analysis of TP53 mutation status in human cancer cell lines: a reassessment. Hum Mutat. 2014; 35(6):756-765. Ikediobi ON, Davies H, Bignell G, et al. Mutation analysis of 24 known cancer genes in the NCI-60 cell line set. Mol Cancer Ther. 2006;5(11):2606-2612. Riley T, Sontag E, Chen P, Levine A. Transcriptional control of human p53-regulated genes. Nat Rev Mol Cell Biol. 2008; 9(5):402-412. Nikolovska-Coleska Z, Xu L, Hu Z, et al. Discovery of embelin as a cell-permeable, small-molecular weight inhibitor of XIAP through structure-based computational screening of a traditional herbal medicine three-dimensional structure database. J Med Chem. 2004;47(10):2430-2440. Pui CH. Recent advances in acute lymphoblastic leukemia. Oncology. 2011;25(4):341, 346-347. Tzoneva G, Perez-Garcia A, Carpenter Z, et al. Activating mutations in the NT5C2 nucleotidase gene drive chemotherapy resistance in relapsed ALL. Nat Med. 2013;19(3):368-371. Mullighan CG, Phillips LA, Su X, et al. Genomic analysis of the clonal origins of relapsed acute lymphoblastic leukemia. Science. 2008;322(5906):1377-1380. Clappier E, Gerby B, Sigaux F, et al. Clonal selection in xenografted human T cell acute lymphoblastic leukemia recapitulates gain of malignancy at relapse. J Exp Med. 2011; 208(4):653-661. Renneville A, Kaltenbach S, Clappier E, et

haematologica | 2018; 103(2)

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

al. Wilms tumor 1 (WT1) gene mutations in pediatric T-cell malignancies. Leukemia. 2010;24(2):476-480. Van Vlierberghe P, Homminga I, Zuurbier L, et al. Cooperative genetic defects in TLX3 rearranged pediatric T-ALL. Leukemia. 2008;22(4):762-770. Zhang J, Ding L, Holmfeldt L, et al. The genetic basis of early T-cell precursor acute lymphoblastic leukaemia. Nature. 2012;481(7380):157-163. Bachas C, Schuurhuis GJ, Reinhardt D, et al. Clinical relevance of molecular aberrations in paediatric acute myeloid leukaemia at first relapse. Br J Haematol. 2014;166(6):902-910. Becker H, Marcucci G, Maharry K, et al. Mutations of the Wilms tumor 1 gene (WT1) in older patients with primary cytogenetically normal acute myeloid leukemia: a Cancer and Leukemia Group B study. Blood. 2010;116(5):788-792. Hollink IH, van den Heuvel-Eibrink MM, Zimmermann M, et al. Clinical relevance of Wilms tumor 1 gene mutations in childhood acute myeloid leukemia. Blood. 2009;113(23):5951-5960. Virappane P, Gale R, Hills R, et al. Mutation of the Wilms' tumor 1 gene is a poor prognostic factor associated with chemotherapy resistance in normal karyotype acute myeloid leukemia: the United Kingdom Medical Research Council Adult Leukaemia Working Party. J Clin Oncol. 2008;26(33):5429-5435. Paschka P, Marcucci G, Ruppert AS, et al. Wilms' tumor 1 gene mutations independently predict poor outcome in adults with cytogenetically normal acute myeloid leukemia: a cancer and leukemia group B study. J Clin Oncol. 2008;26(28):4595-4602. Krauth MT, Alpermann T, Bacher U, et al. WT1 mutations are secondary events in AML, show varying frequencies and impact on prognosis between genetic subgroups. Leukemia. 2015;29(3):660-667. Ho PA, Zeng R, Alonzo TA, et al. Prevalence and prognostic implications of WT1 mutations in pediatric acute myeloid leukemia (AML): a report from the Children's Oncology Group. Blood. 2010;116(5):702-710. Nyvold CG, Stentoft J, Braendstrup K, et al. Wilms' tumor 1 mutation accumulated during therapy in acute myeloid leukemia: Biological and clinical implications.

Leukemia. 2006;20(11):2051-2054. 38. Liu Y, Easton J, Shao Y, et al. The genomic landscape of pediatric and young adult Tlineage acute lymphoblastic leukemia. Nat Genet. 2017;49(8):1211-1218. 39. Coustan-Smith E, Mullighan CG, Onciu M, et al. Early T-cell precursor leukaemia: a subtype of very high-risk acute lymphoblastic leukaemia. Lancet Oncol. 2009; 10(2):147-156. 40. Jeffers JR, Parganas E, Lee Y, et al. Puma is an essential mediator of p53-dependent and -independent apoptotic pathways. Cancer Cell. 2003;4(4):321-328. 41. Kreidberg JA, Sariola H, Loring JM, et al. WT-1 is required for early kidney development. Cell. 1993;74(4):679-691. 42. Idelman G, Glaser T, Roberts CT Jr, Werner H. WT1-p53 interactions in insulin-like growth factor-I receptor gene regulation. J Biol Chem. 2003;278(5):3474-3482. 43. Menke AL, Clarke AR, Leitch A, et al. Genetic interactions between the Wilms' tumor 1 gene and the p53 gene. Cancer Res. 2002;62(22):6615-6620. 44. Hof J, Krentz S, van Schewick C, et al. Mutations and deletions of the TP53 gene predict nonresponse to treatment and poor outcome in first relapse of childhood acute lymphoblastic leukemia. J Clin Oncol. 2011;29(23):3185-3193. 45. Chen J, Nikolovska-Coleska Z, Wang G, Qiu S, Wang S. Design, synthesis, and characterization of new embelin derivatives as potent inhibitors of X-linked inhibitor of apoptosis protein. Bioorg Med Chem Lett. 2006;16(22):5805-5808. 46. Vogler M, Walczak H, Stadel D, et al. Small molecule XIAP inhibitors enhance TRAILinduced apoptosis and antitumor activity in preclinical models of pancreatic carcinoma. Cancer Res. 2009;69(6):2425-2434. 47. Dai Y, Desano J, Qu Y, et al. Natural IAP inhibitor Embelin enhances therapeutic efficacy of ionizing radiation in prostate cancer. Am J Cancer Res. 2011;1(2):128-143. 48. Lunardi A, Ala U, Epping MT, et al. A coclinical approach identifies mechanisms and potential therapies for androgen deprivation resistance in prostate cancer. Nat Genet. 2013;45(7):747-755. 49. Yang T, Lan J, Huang Q, et al. Embelin sensitizes acute myeloid leukemia cells to TRAIL through XIAP inhibition and NFkappaB inactivation. Cell Biochem Biophys. 2015;71(1):291-297.

277

ARTICLE

Non-Hodgkin Lymphoma

Ferrata Storti Foundation

Haematologica 2018 Volume 103(2):278-287

Epstein-Barr virus-associated primary nodal T/NK-cell lymphoma shows a distinct molecular signature and copy number changes Siok-Bian Ng,1,2,3* Tae-Hoon Chung,3* Seiichi Kato,4 Shigeo Nakamura,4 Emiko Takahashi,5 Young-Hyeh Ko,6 Joseph D. Khoury,7 C. Cameron Yin,7 Richie Soong,1,3 Anand D. Jeyasekharan,3 Michal Marek Hoppe,3 Viknesvaran Selvarajan,1 Soo-Yong Tan,1,2 Soon-Thye Lim,8 Choon-Kiat Ong,9 Maarja-Liisa NairismĂ¤gi,9 Priyanka Maheshwari,2 Shoa-Nian Choo,1 Shuangyi Fan,1 Chi-Kuen Lee,1 Shih-Sung Chuang10 and Wee-Joo Chng3,11

Department of Pathology, Yong Loo Lin School of Medicine, National University of Singapore; 2Department of Pathology, National University Hospital, National University Health System, Singapore; 3Cancer Science Institute of Singapore, National University of Singapore; 4Department of Pathology and Laboratory Medicine, Nagoya University Hospital, Nagoya, Japan; 5Department of Pathology, Aichi Medical University Hospital, Nagakute, Japan, 6Department of Pathology, Samsung Medical Center, Sungkyunkwan University, Seoul, Korea; 7Department of Hematopathology, The University of Texas MD Anderson Cancer Center, Houston, TX, USA; 8Lymphoma Genomic Translational Research Laboratory, National Cancer Centre Singapore, Division of Medical Oncology, National Cancer Center Singapore; 9Lymphoma Genomic Translational Research Laboratory, Division of Medical Oncology, National Cancer Centre Singapore; 10 Department of Pathology, Chi-Mei Medical Center, Tainan, Taiwan and 11Department of Haematology-Oncology, National University Cancer Institute of Singapore, National University Health System 1

*S-BN and T-HC contributed equally to this work.

ABSTRACT

Correspondence: patnsb@nus.edu.sg or mdccwj@nus.edu.sg.

Received: September 9, 2017. Accepted: October 27 2017. Pre-published: November 2, 2017. doi:10.3324/haematol.2017.180430 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/278 ÂŠ2018 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.

278

he molecular biology of primary nodal T- and NK-cell lymphoma and its relationship with extranodal NK/T-cell lymphoma, nasal type is poorly understood. In this study, we assessed the relationship between nodal and extranodal Epstein-Barr virus-positive T/NK-cell lymphomas using gene expression profiling and copy number aberration analyses. We performed gene expression profiling and copy number aberration analysis on 66 cases of Epstein-Barr virus-associated T/NKcell lymphoma from nodal and extranodal sites, and correlated the molecular signatures with clinicopathological features. Three distinct molecular clusters were identified with one enriched for nodal presentation and loss of 14q11.2 (TCRA loci). T/NK-cell lymphomas with a nodal presentation (nodal-group) were significantly associated with older age, lack of nasal involvement, and T-cell lineage compared to those with an extranodal presentation (extranodal-group). On multivariate analysis, nodal presentation was an independent factor associated with short survival. Comparing the molecular signatures of the nodal and extranodal groups it was seen that the former was characterized by upregulation of PD-L1 and T-cell-related genes, including CD2 and CD8, and downregulation of CD56, consistent with the CD8+/CD56immunophenotype. PD-L1 and CD2 protein expression levels were validated using multiplexed immunofluorescence. Interestingly, nodal group lymphomas were associated with 14q11.2 loss which correlated with loss of TCR loci and T-cell origin. Overall, our results suggest that T/NK-cell lymphoma with nodal presentation is distinct and deserves to be classified separately from T/NK-cell lymphoma with extranodal presentation. Upregulation of PD-L1 indicates that it may be possible to use anti-PD1 immunotherapy in this distinctive entity. In addition, loss of 14q11.2 may be a potentially useful diagnostic marker of T-cell lineage.

haematologica | 2018; 103(2)

Nodal EBV+ T/NK-cell lymphoma is molecularly distinct

Introduction Epstein-Barr virus (EBV)-associated T- and NK-cell lymphoproliferative diseases (TNKL) comprise a heterogeneous group of cytotoxic T/NK lymphoproliferative disorders, including extranodal NK/T-cell lymphoma, nasal type (ENKTL), and a group of cutaneous and systemic diseases in children such as systemic EBV-positive T-cell lymphoma, aggressive NK-cell leukemia, chronic active EBV infection of T/NK subtype, hydroa vacciniforme–like lymphoproliferative disorder, and mosquito bite hypersensitivity.1 ENKTL is the prototypic example of TNKL in adults.2 The majority of ENKTL are derived from NK cells but some show a cytotoxic T-cell phenotype. Although the nasal cavity is the most common site of involvement, ENKTL can involve a wide variety of extranodal sites including the skin, gastrointestinal tract and testes. Lymph node involvement is uncommon but may occur as a secondary event.3 In addition to ENKTL, there is a group of TNKL that occurs in adults and presents exclusively with lymph node disease but may involve a limited number of extranodal organs (nodal TNKL).4 These cases show significant overlap with ENKTL but a few reports have described clinicopathological features distinct from ENKTL, including the lack of nasal involvement, frequent T-cell origin, and CD8+/CD56- phenotype.5-8 The World Health Organization (WHO) recognizes that most of this group of primary nodal lymphomas are derived from T cells but that a minority have an NK-cell origin and might be different from the prototypic ENKTL, which rarely involve lymph nodes. Since it is currently unclear whether this group of nodal TNKL represents a distinct entity, the revised 2016 WHO lymphoma classification has recommended including primary nodal TNKL as an EBV-positive variant of peripheral T-cell lymphoma, not otherwise specified.2 The molecular biology of nodal TNKL is unknown and its relationship with ENKTL remains controversial. To date, mainly due to the rarity of such malignancies, efforts to resolve these controversies have been largely based on clinicopathological assessment of relatively few cases. In this study, we assessed the relationship between nodal and extranodal EBV-positive TNKL using gene expression profiling (GEP) and copy number aberration (CNA) analyses in a large cohort of adult patients. We correlated the molecular signature and copy number profile with lineage and with clinicopathological features in an attempt to understand whether cases with nodal disease at presentation are distinct from their extranodal counterparts. In addition, we also analyzed the GEP and CNA of nodal TNKL compared to control tissues in order to expand our knowledge on the tumor biology of this relatively unknown group of diseases.

Methods Study group The study group included 66 adult patients with no known immunodeficiency who had been diagnosed with EBV-positive TNKL involving extranodal and nodal sites. All the cases were positive for cCD3, EBV-encoded small RNA (EBER) in the majority of tumor cells (>50%), and at least one cytotoxic marker (TIA1 and/or granzyme B). Systemic and cutaneous EBV-positive T/NK haematologica | 2018; 103(2)

lymphoproliferative diseases occurring in children were excluded. Clinical data were obtained (Online Supplementary Table S1) and cases were then categorized based on: (i) disease presentation (extranodal versus nodal); (ii) nasal involvement (absence versus presence); and (iii) cell of origin (T cell versus NK cell). Details of lineage assignment are provided in Online Supplementary Figure S1 and the Online Supplementary Methods. This study was approved by NHG Domain Specific Review Board B (2009/212).

Gene expression profiling GEP was performed on 66 cases using the Ovation formalinfixed paraffin-embedded (FFPE) WTA System (NuGEN, San Carlos, CA, USA) and labeled products were hybridized onto Affymetrix Human Gene 1.0 ST arrays (Affymetrix, ThermoFisher Scientific, Waltham, MA, USA).

Copy number analysis The OncoScan® FFPE assay (Affymetrix) was performed on 41 TNKL cases and 15 control tissues according to the manufacturer's instructions as previously described.9,10 Copy number analysis was performed using OncoScan® Console (v1.3) software (ThermoFisher Scientific, Waltham, MA, USA).

Multiplex immunofluorescence and multispectral imaging Multiplex immunofluorescent staining was performed to assess CD3/PD-L1 and CD3/CD2 expression on 3 mm FFPE tissue sections using the Opal 7-color Flourophore TSA plus Fluorescence Kit (NEL 797001KT, PerkinElmer Inc., Waltham, MA, USA). Images were acquired and analyzed with the Vectra 2 multispectral automated imaging system (PerkinElmer Inc., Waltham, MA, USA) and inForm 2.0 image analysis software (PerkinElmer Inc., Waltham, MA, USA).

Gene expression profiling and copy number aberration analysis GEP data were processed with a frozen robust multiarray analysis algorithm, an extension of the renowned robust multiarray analysis algorithm that utilizes probe-specific effects and variances.11 Unsupervised hierarchical clustering was performed using the top 500 most variable genes (Online Supplementary Table S2). To account for the difference in inherent gene-by-gene signal intensity range, we divided signal intensities of a gene by its gross median and then applied the log2 function before variability assessment. CNA analysis was performed using segmentation results and probe-by-probe signal intensities from OncoScan® Console. Five CNA levels were determined as follows: -2 (CN ≤ 0.5), -1 (0.5 < CN ≤ 1.5), 0 (1.5 < CN < 2.5), +1 (2.5 ≤ CN < 3.5), +2 (3.5 ≤ CN). Cluster and nodal-group specific CNA, as well as CNA of four TCR loci (TCRA, TCRB, TCRG, TCRD) were derived. The genes differentially expressed between TNKL with a nodal presentation (N-group) and those with an extranodal presentation (EN-group) were identified by using the significance analysis of microarrays algorithm.12 The T-cell signatures of the two groups were analyzed using published genes CD27, CD3G, CD3D, ICOS, MAL, TCF7, PKIA13 according to the procedure outlined in a previous publication.14 Gene set enrichment analysis (GSEA)15 of the N-group compared to the EN-group and normal lymph nodes was performed using the gseapy package with MSigDB version 5.2 (Broad Institute, Cambridge, MA, USA). GEP (GSE90597) and CNA (GSE90777, GSE90783) raw data have been deposited in the Gene Expression Omnibus database repository. Full methods, including the fluorescence in situ hybridization for TCRA, are detailed in the Online Supplementary Methods. The overall study design is illustrated in Online Supplementary Figure S2. 279

S-B.Ng et al.

Results Clinicopathological features of the study group Overall, 47 patients presented with extranodal disease (EN-group) and 19 with nodal disease (N-group) (Table 1, Figure 1A). Twenty-nine cases were classified as having Tcell origin, 27 as NK-cell lineage and ten cases as being of indeterminate lineage (Online Supplementary Figure S1). For 56 cases with sufficient data, the most common CD8/CD56 phenotype was CD8-/CD56+ (25/56, 44.5%), followed by CD8-/CD56- (14/56, 25%), CD8+/CD56(10/56 cases, 18%), and CD8+/CD56+ (7/56, 12.5%) (Online Supplementary Table S3). As expected, all ten cases with the CD8+/CD56- phenotype were of T-cell origin and the majority (20/25, 80%) of cases with the CD8-/CD56+ phenotype were of NK-cell lineage (P=0.0001) (Figure 1B), consistent with CD56 being a marker of NK-cell origin. Interestingly, cases with the CD8+/CD56+ phenotype could be of T- or NK-cell derivation. This indicates that the presence of CD8 positivity, despite the presence of CD56, should prompt the need for clonality assessment to distinguish T-cell from NK-cell origin (Figure 1B). For the EN-group, three out of 45 cases (7%) were positive for TCRB and two out of 43 were positive for TCRG (5%). Thirty-nine of 44 cases (89%) were negative for both TCRB and TCRG. For the N-group, nine out of 17 cases

(53%) were positive for TCRB and none of the 11 cases analyzed was positive for TCRG. Six of 15 cases (40%) were silent for both TCRB and TCRG.

Table 1. Comparison of clinicopathological features, gene expression profiling clusters, PD-L1 expression, and loss of TCR loci in the nodal and extranodal group.

Nodal (n=19) Age (SD), years 60.95 (15.43) Sex (female/male) 4 / 15 Nasal involvement (yes/no) 0 / 17 COO1: T vs. NK vs. Indeterminate 16 / 2 / 1 Stage (1 & 2 / 3 & 4) 2 / 15 CD8 (positive/negative) 12 / 6 CD56 (positive/negative) 4 / 14 Cluster (1 vs. 2 & 3) 9 / 10 PD-L1 (tumor)2 1.12 (0.29) PD-L1 (normal)2 1.44 (0.47) TCR (loss/intact) 12 / 0 Median survival (months) 2.47

Extranodal (n=47)

P-value

48.68 (16.65) 14 / 33 20 / 15 13 / 25 / 9 23 / 12 6 / 33 33 / 13 10 / 37 0.86 (0.34) 1.15 (0.48) 7 / 22 25.7

0.0071(3) 0.554(4) 3.49 × 10-5(4) 0.000116(4) 0.00031(4) 0.00021(4) 0.00053(4) 0.069(4) 9.1 × 10-7(3) 0.00031(3) 6.38 × 10-6(4) 0.00112(5)

SD: standard deviation; 1COO: cell of origin; 2average optical density ratio (standard deviation); 3 t-test; 4Fisher exact test; 5log-rank test.

Figure 1. Clinicopathological features of the nodal and extranodal groups. (A) Composite map showing clinicopathological features and loss of TCR loci in the N- and EN-groups. The map highlights lack of nasal involvement, CD8+/CD56- phenotype, T-cell lineage and loss of TCR loci in the N-group compared to the EN-group. (B) Graph illustrating the association of CD8/CD56 phenotype with cell lineage. The CD8-/CD56+ phenotype is associated with NK-cell lineage while the CD8+/CD56- phenotype is associated with T-cell origin. (C) Survival curve between the N- and EN-groups. Patients in the N-group had significantly shorter overall survival (OS) compared to those in the EN-group.

280

haematologica | 2018; 103(2)

Nodal EBV+ T/NK-cell lymphoma is molecularly distinct

Molecular signature and copy number profile of the study group revealed distinct molecular subsets Unsupervised clustering of GEP performed on 66 TNKL cases revealed three distinct clusters (Figure 2). Cases with upregulation of top variable genes formed a distinct cluster in the middle (cluster 1), which showed enrichment for nodal presentation and nine of 13 cases (69%) tested showed loss of 14q11.2 (Figure 2). The second cluster (cluster 2, left) showed predominant downregulation of genes, loss of 13q14.3-q21.33 (3/7 cases, 43%), and lack of 14q11.2 loss. The third cluster (cluster 3, right) revealed a mixture of up- and downregulated genes. Clusters 2 and 3 showed gain of 1q32.1-q32.3 (8/28 cases, 29%) and loss of Xp22.33 (14/28, 50%), which were absent or uncommon in cluster 1. There was no correlation of the molecular clusters with tissue sites (P>0.05), suggesting that the clustering is not related to the signature of the tissue type or tumor microenvironment. Overall, our results showed that TNKL is molecularly heterogeneous with distinct clusters associated with specific CNA. Notably, cases with nodal presentation (Ngroup) clustered together (Figure 2) and showed frequent loss of 14q11.2, consistent with the proposal that EBV-positive T/NK lymphoma with nodal disease (nodal TNKL) is distinct from those with extranodal disease. We therefore proceeded to compare the clinical, pathological and molecular features of nodal versus extranodal TNKL.

Comparison of the clinicopathological features of the nodal and extranodal groups Patients in the N-group were older than those in the ENgroup (average age, 61 versus 48 years; P=0.0071) (Table 1). None of the patients had nasal involvement (Figure 1A). Regarding CD8 and CD56 expression, the most common phenotype in the N-group was CD8+/CD56- (9/17 cases, 52.9%) (Online Supplementary Table S3). The majority (14/16, 87%) were derived from T cells and only two were of NK-cell origin (Figure 1A). Interestingly, patients in the N-group more often had advanced stage disease (P=0.0003) (Table 1) and poorer overall survival (P=0.00112) (Figure 1C) compared to those in the EN-group. We also performed multivariate analysis to assess the effects of other covariates such as disease presentation, age, cell of origin, nasal involvement and stage of disease on overall survival. Our results demonstrated that only nodal presentation (HR=4.04, P=0.016) and stage of disease (HR=2.06,

Table 2A. Comparison of clinicopathological features, gene expresson profiling clusters, and loss of TCR loci in nodal group-T cases and extranodal group-NK cases.

Nodal-T a

Age (SD), years Sex (female/male) Nasal involvement (no/yes) CD8 (negative/positive) CD56 (negative/positive) TCR loss (negative/positive) Stage (1 & 2/3 & 4) Cluster (cluster 1 / 2 or 3) Median survival (months)

63.2 (15.83) 4/12 14/0 3/12 13/2 0/12 2/13 9/7 2.47

Extranodal-NK 49.3 (17.6) 8/17 7/11 20/3 2/23 14/0 14/4 4/21 24

Mean (standard deviation); bt-test; cFisher exact test; dlog-rank test.

haematologica | 2018; 103(2)

P=0.0015) were still significant after adjusting for other factors (Online Supplementary Table S4). We further compared the above clinicopathological features between T-cell cases in the N-group (N-T) and NK-cell cases in the EN-group (EN-NK) and between T-cell cases in the N-group (N-T) and the EN-group (EN-T) to understand whether the differences between the N-group and the ENgroup were simply related to lineage (Table 2A,B). Our results revealed that older age, lack of nasal involvement, CD8 expression and poor outcome remained significantly associated with N-T cases when compared to both EN-NK and EN-T cases, indicating that the N-group is distinct from the EN-group beyond cell of origin.

Molecular signature and copy number aberrations of the nodal-group compared to extranodal-group Gene expression profiling In order to determine whether the molecular features of cases with nodal presentation are distinct, we compared the molecular signatures and copy number changes between the N-group and the EN-group. From the list of differentially expressed genes (Online Supplementary Table S5), we observed upregulation of several T-cell-related genes in the N-group, such as CD2, CD8, CD3G, CD3D, TRAC and LEF1 (Figure 3, Online Supplementary Table S5B), and downregulation of CD56 (NCAM1) in the Ngroup compared to the EN-group (Online Supplementary Table S5A), consistent with the CD8+/CD56- phenotype of the N-group cases. In addition, PD-L1 (CD274) is upregulated in the N-group compared to the EN-group (Figure 3, Online Supplementary Table S5B). PD-L1 has been reported to be overexpressed in EBV-associated lymphomas, including ENKTL and patients with relapsed ENKTL have shown effective responses to anti-PD1 therapy.16-18 Hence, the upregulation of PD-L1 in the N-group may have potential therapeutic implications for anti-PD1 treatment. To confirm our data on differentially expressed genes, we selected CD2, a pan-T-cell marker, and PD-L1 for validation using multiplex immunofluorescence. Consistent with the GEP results, the expression of PD-L1 and CD2 was significantly higher in the N-group than in the ENgroup (Figure 4). Further analysis revealed higher expression of CD2 in the tumor cells (CD3+), but not in nontumor cells (CD3-), of N-group cases (Figure 4), corroborating the T-cell origin of the cases. For PD-L1 expression,

Table 2B. Comparison of clinicopathological features, gene expresson profiling clusters, and loss of TCR loci in nodal group-T cases and extranodal group-T cases.

P b

0.0131 0.734c 0.000389c 8.55×10-5 c 7.92×10-5 c 1.04×10-5 c 0.000361c 0.0144c 0.0316d

Nodal-T

Extranodal-T

63.2 (15.83) 4/12 14/0 3/12 13/2 0/12 2/13 9/7 2.47

50.5 (13.5) 6/7 6/6 9/2 7/6 3/7 5/7 6/7 25.7

0.0281b 0.27c 0.00401c 0.00431c 0.0957c 0.0779c 0.185c 0.715c 0.0253d

Mean (standard deviation); bt-test; cFisher exact test; dlog-rank test.

281

S-B.Ng et al.

both tumor and non-tumor cells of the N-group cases revealed higher expression compared to their counterparts in the EN-group (Figure 4). Based on morphology, we affirmed that most of the PD-L1+ non-tumor cells were macrophages indicating a higher number of macrophages in the tumor microenvironment of N-group cases. This is also in line with our gene expression results showing upregulation of CD68, a macrophage marker, in the Ngroup compared to the EN-group (Figure 3).

We further assessed the 9p24.1/PD-L1/PD-L2 locus for CNA since this was reported to be associated with upregulation of PD-L1 in the majority of cases of classical Hodgkin lymphoma.19 However, only two samples in the EN-group (NKTL24, NKTL27A) and none in the N-group showed gain of 9p24.1, indicating that the upregulation of PD-L1 in the N-group is likely due to mechanisms other than CNA. In order to understand whether PD-L1 overexpression may be driven by EBV,16,17 we also correlated the

Figure 2. Composite heatmap of gene expression profiling clusters with (top) dendrogram highlighting three clusters in blue, red and green, (upper-middle) disease presentation (nodal in red and extranodal in blue), (lower middle) matrix of the top 500 most highly variable genes, and (bottom) cluster-specific copy number alterations of 66 samples of T/NK-cell lymphoma. Each column represents a case. Genes (rows) in the GEP matrix are ordered based on clustering, but chromosomal segments (rows) in cluster-specific CNA were ordered simply based on their genomic location. Yellow in the GEP matrix indicates upregulation and blue represents downregulation. For the cluster-specific CNA, orange indicates copy number gain and blue represents copy number loss. The GEP heatmap shows three distinct clusters. Cluster 1 (middle, red) shows enrichment for nodal presentation, upregulation of genes and distinctive loss of 14q11.2. Cluster 2 (left, blue) reveals downregulation of genes and loss of 13q14.3-q21.33. Cluster 3 (right, green) shows a mixture of upregulation and downregulation.

282

haematologica | 2018; 103(2)

Nodal EBV+ T/NK-cell lymphoma is molecularly distinct

expression of EBER and PD-L1 in the tumor population but did not observe any significant association between the two (Online Supplementary Figure S3). This suggests that the upregulation of PD-L1 in this group of EBV-positive T/NK lymphomas may either be a consequence of indirect mechanisms driven by EBV or attributed to other novel pathways which will require additional functional studies for further validation. We performed GSEA and compared the N-group to the EN-group, as well as N-T cases to both EN-NK and EN-T cases. However, we did not identify any significant signaling pathways which are distinctly enriched in either group. This may be related to small sample size, and/or measurement variability from heterogeneity in sample processing and preparation.

Copy number analysis The copy number profile revealed similarities and differences between the two groups (Figure 5). The recurrent copy number losses present in both groups included loss of 3q26.1 and 22q11.23 (Online Supplementary Table S6). Interestingly, loss of 14q11.2 was present in the majority

of N-group cases but identified only in a minority of ENgroup cases (P<0.05). In contrast, the EN-group was characterized by recurrent loss of 6q22.3-q26 and 9p21.3 as well as gains in 1q22-q44, 2q22.2-q33.1, 6p21.3, 7q21.1q36.3, 11q24.3, 13q14.2, and 17q21.2-q25.3 (Online Supplementary Table S6B,C). In support of the validity of our copy number results, the recurrent CNA identified in the EN-group in this study have also been reported to be frequently observed in ENKTL in previous studies.20,21 Since the 14q11.2 locus contains the T-cell receptor alpha constant (TRAC or TCRA locus, Online Supplementary Table S6A), we hypothesized that the loss of 14q11.2 in the N-group might be an indication of T-cell lineage as focal loss within the TCR loci (commonly biallelic) can be a reflection of physiological rearrangement of the TCR loci via VDJ recombination. We correlated the loss of TCR loci (TCRA, TCRB, TCRG, TCRD) in our samples with cell of origin. Only cases with lineage assigned based on complete TCRB/TCRG immunohistochemical and clonality data were used for this analysis (cases denoted with an asterix in Online Supplementary Table S1 were excluded). Indeed 13 out of 16 (81%) TNKL

Figure 3. Heatmap of differentially expressed genes between nodal- and extranodal-groups (fold change > 1.5, P<0.05, false discovery rate < 0.15). Selected genes related to T cells are highlighted. Upregulation and downregulation in the heatmap are marked by yellow and blue, respectively.

haematologica | 2018; 103(2)

283

S-B.Ng et al.

of T-cell origin (as detected by clonality and TCRB/TCRG immunohistochemistry) showed deletion of at least two TCR loci (Figure 1A, Online Supplementary Figure S4, Online Supplementary Table S7). TCRA and TCRD were the most commonly deleted loci, possibly due to positional proximity. None of the five TNKL of NK-cell lineage displayed loss of TCR loci (Figure 1A). To verify the specificity of 14q11.2 loss as a marker for T-cell lineage, we performed OncoScan profiling of additional T-cell lymphomas, benign lymphoid tissues, and normal lymphocytes including T cells, B cells, and NK cells (Online Supplementary

Methods). The loss of one or more TCR loci was detected in five out of six (83%) peripheral T-cell lymphomas and two normal T-cell samples, but not observed in benign lymphoid tissues, normal B cells, or NK cells (Online Supplementary Figure S5; Online Supplementary Table S7). Taken together, 20 of 24 samples of T-cell origin showed loss of TCR loci but not any of the nine non-T-cell samples (P<0.0001) (Online Supplementary Table S8, Online Supplementary Figure S6). Our findings suggest that loss of TCR loci, especially the 14q11.2 region, detected via the OncoScan platform may be an indicator of T-cell origin.

Figure 4. Protein expression of CD2 and PD-L1 in the nodal and extranodal groups. (A). Plots of CD2 and PD-L1 expression in all cells, tumor and non-tumor cells. Each dot represents the median optical density (OD) ratio of a specific marker in each image analyzed. At least four images containing at least 10,000 cells were quantified per case. The expression of CD2 and PD-L1 across all cell types (both tumor and non-tumor cells) in the cases is illustrated in the left plots indicated â&#x20AC;&#x2DC;Overallâ&#x20AC;&#x2122;, while that for tumor and non-tumor cells is represented in the middle and right plots, respectively. Our results showed that CD2 expression is significantly upregulated in tumor cells of the N-group compared to the EN-group. Interestingly, PD-L1 expression is significantly higher in both tumor and non-tumor cells in the N-group than in the EN-group. (B). Expression of CD2 and PD-L1 in the N-group and the EN-group cases using multiplexed immunofluorescence (MIF, a-f and m-r) and corresponding multispectral analysis (g-l and s-x). CD3 stained cell membrane of tumor cells (magenta) in single unmixed images (a, d, m, and p). Single CD2 (b and n, cell membrane stain) and PD-L1 (e and q, cell membrane) expression are stained yellow. Images a-l represent a case in the N-group (TNK4) showing high CD2 and PD-L1 expression. CD3+ tumor cells accounted for 82.6% of cells (a and g, positive cells marked magenta and red, respectively) and 67.75% of cells (d and j, positive cells marked magenta and red, respectively). CD2 was present in 81.84% of cells (b and h, positive cells marked yellow and green, respectively). Composite CD3 and CD2 image (c) illustrated 75.9% of double positive cells (I, yellow cells), which represented tumor cells with positive CD2 expression. This case also showed high PD-L1 expression in 87.0% of cells (e and k, positive cells marked yellow and green, respectively). Composite CD3 and PD-L1 image (f) revealed 55.9% of tumor cells showing PD-L1 expression (l, yellow cells). Images m-o, s-u represent a case in the EN-group (TW9) with comparatively low expression of CD2. CD3+ tumor cells accounted for 66.22% of cells (m and s, positive cells marked magenta and red, respectively). Single CD2 expression quantified as 6.75% of cells (n and t, positive cells marked yellow and green, respectively) and the composite image (o) showed 6.51% of CD3+/CD2+ double positive cells (u, double positive cells marked yellow). Images p-r, v-x represent a case in the EN-group (NKTL43) with relatively low expression for PD-L1. CD3+ tumor cells accounted for 67.12% of cells (p and v, positive cells marked magenta and red, respectively). PD-L1 expression is low in 17.53% of cells (q and w, positive cells marked yellow and green, respectively). Composite CD3 and PD-L1 image (r) revealed 5.55% of CD3+/PD-L1+ cells (x, yellow cells).

284

haematologica | 2018; 103(2)

Nodal EBV+ T/NK-cell lymphoma is molecularly distinct

To understand the mechanism of loss of the TCRA locus in 14q11.2 better, we performed fluorescence in situ hybridization using breakapart probes labeling the TCRA locus to look for translocation of TCRA. Our results showed an intact TCRA locus, indicating that the 14q11.2 loss was not a result of translocation of the TCRA gene and, therefore, more likely a result of the V(D)J recombination that occurs during early stages of T-lymphocyte maturation when segments between the VDJ gene loci are deleted (Online Supplementary Figure S7; Online Supplementary Methods). The N-group also demonstrated significantly higher expression of a previously published T-cell signature13 compared to the EN-group, reinforcing our hypothesis that loss of TCR loci in the former is associated with T-cell lineage (P<0.0001, Online Supplementary Figure S8; Online Supplementary Methods).

Molecular signature and copy number aberrations of the nodal group compared to normal controls Since very little is known about the molecular biology of nodal TNKL as a disease entity, we generated the differentially expressed genes (Online Supplementary Table S9) of the 19 N-group cases compared to normal lymph node controls and performed GSEA. Our results revealed significant enrichment for hallmark gene sets such as MTORC1_SIGNALING, IL6_JAK_STAT3_SIGNALING as well as several gene sets related to the cell cycle and genomic instability,

including G2M_CHECKPOINT, E2F_TARGETS, MYC_TARGETS, and APOPTOSIS (Online Supplementary Table S10). Twelve out of 19 N-group cases were analyzed for CNA. The recurrent CNA occurring in at least 20% of samples include loss of chr14q11.2 (12/12, 100%), chr3q26.1 (8/12, 67%), and chr22q11.23 (4/12, 33%) (Online Supplementary Table S6A). Loss of chr14q11.2 was the most frequent CNA, consistent with this aberration as a marker of T-cell lineage. The genes residing in chr22q11.23 include LOC768328, GSTTP2, LOC391322, LOC100420769, GSTTP1 and GSTT1. No known genes reside in the chr3q26.1 locus. There were no recurrent gains observed in this group.

Discussion This is the first large-scale, high-throughput genomics study to interrogate the molecular heterogeneity of TNKL using an unbiased and unsupervised approach. As expected, TNKL is not a uniform entity and molecular subsets exist. Interestingly, we identified one molecular subtype to be associated with distinct CNA, including the loss of 14q11.2, and this group was enriched for nodal presentation and T-cell lineage. In line with recent reports, we verified that nodal TNKL is associated with older age, lack of nasal involvement and the CD8+/CD56- phenotype, in contrast to ENKTL. We also discovered that patients with

Figure 5. Penetrance plots showing the frequency of gains and losses of genomic regions of cases in the nodal and extranodal groups. Each chromosome is represented on the x-axis, and the y-axis indicates the proportion of gain or loss of the corresponding genomic region within the corresponding population. Gains are shown in red and losses in blue.

haematologica | 2018; 103(2)

285

S-B.Ng et al.

nodal TNKL often presented with an advanced stage of disease and had a much poorer survival than those with ENKTL. These differences remained significant even when we compared only the T-cell cases in both groups, indicating that there are underlying biological differences, other than cell of origin, between the two groups (Tables 1 and 2). GSEA failed to identify distinct signaling pathways in the two groups although this may be due to limited sample size and/or measurement variability. Alternatively, there may be mutational and epigenetic differences between the two groups which were not investigated in this study. The diagnosis and classification of TNKL is difficult because of considerable heterogeneity and overlap in morphology, phenotype, cell lineage, and clinical presentation due to involvement of multiple anatomic sites.22 Our data revealed that TNKL presenting with nodal disease is characterized by unique clinical and phenotypic features as well as molecular and copy number signatures, thereby suggesting that it has a distinct molecular biology and deserves a separate classification from ENKTL. Our results also highlight the importance of anatomic location in defining the disease and that localization of the disease to nodal sites at presentation can be a valuable clue to distinguish between nodal TNKL and ENKTL.22 This distinction is especially important considering the significantly different clinical outcome between the two entities and hence a pressing need to treat nodal TNKL with more effective therapy. However, whether nodal TNKL represents a distinctive entity or should be classified under the broad category of peripheral T-cell lymphoma, not otherwise specified, as recommended by the 2016 revised WHO classification of lymphoid neoplasms, requires further study.1,2,4 Interestingly, two cases in the N-group (TW23 and TNK5) were of NK-cell lineage with the CD8-/CD56+ phenotype and both clustered with the EN-group rather than with the N-group (Figure 2) despite their nodal presentation. Whether such cases are biologically akin to EN-group cases remains uncertain and warrants additional study with more numerous samples. The molecular biology of nodal TNKL is poorly understood and, to date, there is only one report describing the molecular signature of three cases of nodal TNKL.23 The study showed significant enrichment of immune response genes with upregulation of genes associated with cytotoxic activation and downregulation of genes associated with T- and B-cell activation. Pathways generally known to be involved in tumorigenesis, such as apoptosis, proliferation, or cell adhesion, were not significantly enriched. In this study, 19 cases of nodal TNKL were profiled for gene expression and 12 were subjected to copy number testing. Our study demonstrated enrichment of gene sets related

References 1. Attygalle AD, Cabecadas J, Gaulard P, et al. Peripheral T-cell and NK-cell lymphomas and their mimics; taking a step forward report on the lymphoma workshop of the XVIth meeting of the European Association for Haematopathology and the Society for Hematopathology. Histopathology. 2014;64(2):171-199.

286

to the JAK/STAT3_signaling pathway, similar to ENKTL in which activating mutations of JAK3 and phosphorylation of STAT3 result in activation of this oncogenic pathway. Interestingly, gene sets related to the cell cycle and genomic instability were also enriched in nodal TNKL. Genomic instability is an evolving hallmark of cancer and it can be a result of many different pathways.24 Understanding the mechanisms leading to genomic instability will lay the foundation for novel therapeutic approaches that exploit the selective vulnerability of cancers conferred by their unstable genomes.25,26 Interestingly, we have verified that loss of 14q11.2 correlated well with loss of TCR loci and T-cell lineage. Considering the importance of accurate diagnosis of the cell of origin in lymphoid malignancies, it is worth noting that detection of aberrations in TCR loci using the OncoScan platform, which works reliably on FFPE materials, may be a novel and potentially useful diagnostic tool for determining T-cell lineage. Successful treatment of patients with advanced cancer using antibodies against PD-1 and its ligand (PD-L1) has highlighted the critical importance of PD-1/PD-L1-mediated immune escape in cancer development. Targeting the PD-1/PD-L1 pathway has shown clinical efficacy not only in solid tumors but also in Hodgkin and nonHodgkin lymphoma.27 Overexpression of PD-L1 and/or PD-L2 has been shown to play a critical role in immune evasion by lymphoma cells.19,28 Overexpression of PD-L1 has been reported in ENKTL16,17 and anti-PD1 immunotherapy has been shown to be effective in patients with relapsed and refractory ENKTL.18 Our findings have shown that nodal TNKL is a much more aggressive disease than ENKTL with a median survival of less than 3 months, highlighting a pressing need for more effective treatment modalities. Importantly, the upregulation of PD-L1 in nodal TNKL elucidated in this study may have potential therapeutic implications and definitely warrants further study on the possibility of PD-L1 immunotherapy in this group of tumors. The mechanism of PD-L1 upregulation in TNKL remains unclear as it did not correlate with EBV expression nor was it a result of gain in the 9p24.1/PD-L1/PD-L2 locus. Finally, we observed high PD-L1 expression on the tumor cells as well as tumor-infiltrating macrophages in the N-group cases. The mechanisms responsible for the recruitment of macrophages to tumors and the control of PD-L1 upregulation on these cells also require further evaluation. Acknowledgments S.-B.N was supported by the Singapore Ministry of Healthâ&#x20AC;&#x2122;s National Medical Research Council Transition Award (NMRC/TA/0020/2013) and Translational & Clinical Research (TCR) Flagship Program (TCR12dec005).

2. Swerdlow SH, Campo E, Pileri SA, et al. The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood. 2016;127(20):2375-2390. 3. Chan JKC, Quintanilla-Martinez L, Ferry JA, Peh SC. Extranodal NK/T-cell lymphoma, nasal type. In: Swerdlow SH, Campo E, Harris NL, et al., eds. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. Lyon: IRAC Press; 2008:285-288.

4. Ko YH, Chan JKC, Quintanilla-Martinez L. Virally associated T-cell and NK-cell Neoplasms. In: Jaffe ES, Arber DA, Campo E, Harris NL, Quintanilla-Martinez L, eds. Hematopathology. Philadelphia: Elsevier; 2017:565-598. 5. Kato S, Asano N, Miyata-Takata T, et al. Tcell receptor (TCR) phenotype of nodal Epstein-Barr virus (EBV)-positive cytotoxic T-cell lymphoma (CTL): a clinicopathologic study of 39 cases. Am J Surg Pathol.

haematologica | 2018; 103(2)

Nodal EBV+ T/NK-cell lymphoma is molecularly distinct

2015;39(4):462-471. 6. Kato S, Takahashi E, Asano N, et al. Nodal cytotoxic molecule (CM)-positive EpsteinBarr virus (EBV)-associated peripheral T cell lymphoma (PTCL): a clinicopathological study of 26 cases. Histopathology. 2012;61(2):186-199. 7. Jeon YK, Kim JH, Sung JY, Han JH, Ko YH, Hematopathology Study Group of the Korean Society of P. Epstein-Barr virus-positive nodal T/NK-cell lymphoma: an analysis of 15 cases with distinct clinicopathological features. Hum Pathol. 2015;46(7):981-990. 8. Takahashi E, Asano N, Li C, et al. Nodal T/NK-cell lymphoma of nasal type: a clinicopathological study of six cases. Histopathology. 2008;52(5):585-596. 9. Lee CS, Bhaduri A, Mah A, et al. Recurrent point mutations in the kinetochore gene KNSTRN in cutaneous squamous cell carcinoma. Nat Genet. 2014;46(10):1060-1062. 10. Harms PW, Fullen DR, Patel RM, et al. Cutaneous basal cell carcinosarcomas: evidence of clonality and recurrent chromosomal losses. Hum Pathol. 2015;46(5):690697. 11. McCall MN, Bolstad BM, Irizarry RA. Frozen robust multiarray analysis (fRMA). Biostatistics. 2010;11(2):242-253. 12. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA. 2001;98(9):5116-5121. 13. Iqbal J, Weisenburger DD, Chowdhury A, et al. Natural killer cell lymphoma shares strikingly similar molecular features with a group of non-hepatosplenic gammadelta T-

haematologica | 2018; 103(2)

14.

15.

16.

17.

18.

19.

20.

cell lymphoma and is highly sensitive to a novel aurora kinase A inhibitor in vitro. Leukemia. 2011;25(2):348-358. Chng WJ, Chung TH, Kumar S, et al. Gene signature combinations improve prognostic stratification of multiple myeloma patients. Leukemia. 2016;30(5):1071-1078. 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 USA. 2005;102(43):15545-15550. Han L, Liu F, Li R, et al. Role of programmed death ligands in effective T-cell interactions in extranodal natural killer/Tcell lymphoma. Oncol Lett. 2014;8(4):14611469. Chen BJ, Chapuy B, Ouyang J, et al. PD-L1 expression is characteristic of a subset of aggressive B-cell lymphomas and virusassociated malignancies. Clin Cancer Res. 2013;19(13):3462-3473. Kwong YL, Chan TS, Tan D, et al. PD1 blockade with pembrolizumab is highly effective in relapsed or refractory NK/T-cell lymphoma failing L-asparaginase. Blood. 2017;129(17):2437-2442. Roemer MG, Advani RH, Ligon AH, et al. PD-L1 and PD-L2 genetic alterations define classical Hodgkin lymphoma and predict outcome. J Clin Oncol. 2016;34(23):26902697. Iqbal J, Kucuk C, Deleeuw RJ, et al. Genomic analyses reveal global functional alterations that promote tumor growth and novel tumor suppressor genes in natural killer-cell malignancies. Leukemia. 2009;23(6):1139-1151.

21. Huang Y, de Reynies A, de Leval L, et al. Gene expression profiling identifies emerging oncogenic pathways operating in extranodal NK/T-cell lymphoma, nasal type. Blood. 2010;115(6):1226-1237. 22. Swerdlow SH, Jaffe ES, Brousset P, et al. Cytotoxic T-cell and NK-cell lymphomas: current questions and controversies. Am J Surg Pathol. 2014;38(10):e60-71. 23. Ha SY, Sung J, Ju H, et al. Epstein-Barr virus-positive nodal peripheral T cell lymphomas: clinicopathologic and gene expression profiling study. Pathol Res Pract. 2013;209(7):448-454. 24. Negrini S, Gorgoulis VG, Halazonetis TD. Genomic instability--an evolving hallmark of cancer. Nat Rev Mol Cell Biol. 2010;11(3):220-228. 25. Ferguson LR, Chen H, Collins AR, et al. Genomic instability in human cancer: molecular insights and opportunities for therapeutic attack and prevention through diet and nutrition. Semin Cancer Biol. 2015;35(Suppl):S5-24. 26. Pawitan Y, Michiels S, Koscielny S, Gusnanto A, Ploner A. False discovery rate, sensitivity and sample size for microarray studies. Bioinformatics. 2005;21(13):3017-3024. 27. Bachy E, Coiffier B. Anti-PD1 antibody: a new approach to treatment of lymphomas. Lancet Oncol. 2014;15(1):7-8. 28. Green MR, Monti S, Rodig SJ, et al. Integrative analysis reveals selective 9p24.1 amplification, increased PD-1 ligand expression, and further induction via JAK2 in nodular sclerosing Hodgkin lymphoma and primary mediastinal large B-cell lymphoma. Blood. 2010;116(17):3268-3277.

287

ARTICLE

Non-Hodgkin Lymphoma

Ferrata Storti Foundation

A bioclinical prognostic model using MYC and BCL2 predicts outcome in relapsed/refractory diffuse large B-cell lymphoma

Mark Bosch,1 Ariz Akhter,2 Bingshu E. Chen,3 Adnan Mansoor,2 David Lebrun,4 David Good,3 Michael Crump,5 Lois Shepherd,3 David W. Scott6 and Douglas A. Stewart2 University of Saskatchewan, SK; 2University of Calgary, AB; 3Canadian Cancer Trials Group, Queen's University, Kingston, ON; 4Department of Pathology and Molecular Medicine, Queenâ&#x20AC;&#x2122;s University, Kingston, ON; 5University of Toronto, ON and 6BC Cancer Agency, Vancouver, BC, Canada

Haematologica 2018 Volume 103(2):288-296

ABSTRACT

Correspondence: douglas.stewart@albertahealthservices.ca

Received: August 21, 2017. Accepted: October 31, 2017. Pre-published: November 2, 2017. doi:10.3324/haematol.2017.179309 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/288 ÂŠ2018 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.

288

he objective of this study was to create a bioclinical model, based on clinical and molecular predictors of event-free and overall survival for relapsed/refractory diffuse large B-cell lymphoma patients treated on the Canadian Cancer Trials Group (CCTG) LY12 prospective study. In 91 cases, sufficient histologic material was available to create tissue microarrays and perform immunohistochemistry staining for CD10, BCL6, MUM1/IRF4, FOXP1, LMO2, BCL2, MYC, P53 and phosphoSTAT3 (pySTAT3) expression. Sixty-seven cases had material sufficient for fluorescent in situ hybridization (FISH) for MYC and BCL2. In addition, 97 formalin-fixed, paraffin-embedded tissue samples underwent digital gene expression profiling (GEP) to evaluate BCL2, MYC, P53, and STAT3 expression, and to determine cell-of-origin (COO) using the Lymph2Cx assay. No method of determining COO predicted event-free survival (EFS) or overall survival (OS). Factors independently associated with survival outcomes in multivariate analysis included primary refractory disease, elevated serum lactate dehydrogenase (LDH) at relapse, and MYC or BCL2 protein or gene expression. A bioclinical score using these four factors predicted outcome with 3-year EFS for cases with 0-1 vs. 2-4 factors of 55% vs. 16% (P<0.0001), respectively, assessing MYC and BCL2 by immunohistochemistry, 46% vs. 5% (P<0.0001) assessing MYC and BCL2 messenger ribonucleic acid (mRNA) by digital gene expression, and 42% vs. 21% (P=0.079) assessing MYC and BCL2 by FISH. This proposed bioclinical model should be further studied and validated in other datasets, but may discriminate relapsed/refractory diffuse large B-cell lymphoma (DLBCL) patients who could benefit from conventional salvage therapy from others who require novel approaches. The LY12 study; clinicaltrials.gov Identifier: 00078949. Introduction DLBCL has considerable biologic and clinical heterogeneity.1 Following standard chemoimmunotherapy, 10-65% of patients will relapse depending upon their presenting International Prognostic Index (IPI) score.2 Salvage chemotherapy +/- rituximab followed by high dose chemotherapy/autologous stem-cell transplantation (HDCT/ASCT) is the standard treatment for relapsed/refractory DLBCL (rrDLBCL),3-6 although fewer than half of patients are cured with this approach.6-9 To date, research assessing predictive and prognostic biomarkers for rrDLBCL has been limited,10 but with the dawn of multiple novel agents, such research will be critical to help stratify patients into personalized treatment strategies. The two molecular COO subtypes of DLBCL recognized by the 2016 revised WHO classification are the germinal center B-cell (GCB) and activated B-cell (ABC). Hans,11 Choi,12 and the tally13 algorithms use immunohistochemistry (IHC) to assign COO, however, the clinical significance of COO subtyping using IHC remains controversial, especially for relapsed disease.14-18 Current data suggest that COO is more strongly associated with prognosis of DLBCL if assessed by GEP rather than by IHC.19-21 In addition, it is known that genetic aberrations or protein haematologica | 2018; 103(2)

Bioclinical prognostic model for rrDLBCL

expression of BCL2 and MYC identified by FISH or IHC, respectively, are associated with a poor prognosis in newly diagnosed DLBCL.22-28 The objective of this study was to create a bioclinical model, comprising clinical and molecular features, for EFS and OS of rrDLBCL patients utilizing materials and data from the prospective LY12 study conducted by the CCTG.

R) (gemcitabine, dexamethasone and cisplatin +/- rituximab) to DHAP(+/- R) (dexamethasone, high-dose cytarabine, and cisplatin +/- rituximab) followed by HDCT/ASCT for patients with relapsed/refractory aggressive histology lymphoma.3 All patients gave written informed consent to participate and to provide tissue material for biologic studies. This correlative science study was approved by the Health Research Ethics Board of Alberta.

Morphology, IHC, Digital GEP, FISH Methods In the study herein, we evaluated a subset of the 619 patients enrolled on the LY12 trial that compared the efficacy of GDP(+/-

Among the 619 patients in LY12, 130 had rrDLBCL and sufficient formalin-fixed paraffin-embedded (FFPE) tissue samples available to create tissue microarrays (TMAs). TMAs were constructed, using triplicate, 1.0 mm cores from each donor paraffin

Table 1. Characteristics of B-cell lymphoma patients. Total Number Patients Treatment Arm: DHAP(+/- R) GDP(+/- R) Sex Male Female Age, years Median Range ECOG 0-1 2-3 Relapse Stage I-II III-IV Serum LDH Elevated â&#x20AC;&#x153;Bâ&#x20AC;? symptoms Present Extranodal sites >1 Bone marrow Involved Relapse aaIPI 0-1 2-3 Duration of Initial Response > 12 months < 12 months No Response (SD/PD) Prior rituximab Yes Rituximab with Salvage Yes Transformed Lymphoma Yes

All LY12

IHC

NanoString GEP

FISH

554

91 P=0.09 53 (58.2) 38 (41.8) P=0.55 52 (57.1) 39 (42.9) P=0.69 54.7 28-66 P=0.91 79 (86.8) 12 (13.2) P=0.56 27 (29.7) 64 (70.3) P=0.95 50 (54.9) P=0.05 39 (42.9) P=0.51 21 (23.1) P=0.53 5 ( 5.5) P=0.78 33 (36.3) 58 (63.7) P=0.70 27 (29.7) 40 ( 44.0) 24 ( 26.4) P=0.16 63 (69.2) P=0.04 68 (74.7) P=0.29 18 (19.8)

97 P=0.06 57 (58.8) 40 (41.2) P=0.24 53 (54.6) 44 (45.4) P=0.65 55.3 28-66 P=0.89 85 (87.6) 12 (12.4) P=0.94 31 (32.0) 66 (68.0) P=0.99 56 (57.7) P=0.09 40 (41.2) P=0.60 23 (23.7) P=0.42 5 ( 5.2) P=0.58 34 (35.1) 63 (64.9) P=0.61 28 (28.9) 44 (45.4) 25 (25.8) P=0.32 69 (71.1) P=0.03 73 (75.3) P=0.18 20 (20.6)

67 P=0.15 39 (58.2) 28 (41.8) P=0.62 42 (62.7) 25 (62.7) P=0.66 55.4 29-66 P=0.58 57 (85.1) 10 (14.9) P=0.46 19 (28.4) 48 (71.6) P=0.78 38 (56.7) P=0.08 29 (43.3) P=0.50 15 (22.4) P=0.17 2 (3.0) P=0.26 21 (31.3) 46 (68.7) P=0.68 20 (29.9) 30 (44.8) 17 (25.4) P=0.32 47 (70.1) P=0.03 52 (77.6) P=0.06 16 (23.9)

277 (50.0) 277 (50.0) 332 (59.9) 222 (40.1) 55 19-74 483 (87.2) 71 (12.8) 179 (32.3) 375 (67.7) 316 (57.0) 185 (33.4) 143 (25.8) 39 (7.0) 208 (37.5) 346 (62.5) 152 (27.4) 236 (42.6) 166 (30.0) 416 (75.1) 363 (65.5) 89 (16.1)

P-values indicate comparisons of study groups with overall LY12 population. FISH: fluorescence in situ hybridization; IHC: immunohistochemistry; LDH: lactate dehydrogenase; DHAP: dexamethasone, high-dose cytarabine and cisplatin; GDP: gemcitabine, dexamethasone and cisplatin; R: rituximab; SD/PD: stable disease/progressive disease; ECOG: The Eastern Cooperative Oncology Group; IPI: The International Prognostic Index.

haematologica | 2018; 103(2)

289

M. Bosch et al.

block. IHC staining was performed using a Ventana automated immunostainer (Tucson, AZ, USA) for the following proteins: CD10, BCL6, MUM1/IRF4, FOXP1, LMO2, GCET, BCL2, CMYC, P53, and pySTAT3. Inclusion criteria for the IHC study was a diagnosis of rrDLBCL at LY12 trial entry as determined by central hematopathology review and a minimum of two out of three histo spots containing at least 200 tumor cells/histo spot. Only 91 out of the 130 (70%) samples on the TMA met these inclusion criteria for IHC staining. Protein expression was recorded in 10% increments of positive cells and cases were dichotomized by previously reported criteria: MYC (>40%), BCL2 (>70%), p53 (>30%), pySTAT3 (>50%), and COO assigned by the Hans, Choi, and tally algorithms.11-13, 24,25,29 In total, 97 cases had a sufficient quantity of FFPE tissue samples to successfully extract at least 500ng RNA and perform GEP to assess BCL2, MYC, TP53, and STAT3 expression. Raw counts were normalized using nSolver Analysis Software v3.0. Background subtraction was performed for each sample by subtracting the mean of eight negative controls from all data points. Raw counts were further normalized to the geometric mean of nine housekeeping genes, namely ACTB, PRL19, GAPDH (high expressers), PGK1, CLTC, HPRT1 (intermediate expressers) and TBP, GUSB, ABCF1 (low expressers) to adjust possible variations in RNA quantity subjected to hybridization, between samples. At the time we performed this exploratory analysis, no clearly defined cut-off for high expression existed for digital GEP data. We arbitrarily specified the cut-off for both MYC and BCL2 digital GEP to be 1.5x median prior to data analysis, as per previous reports evaluating GEP.30-33 COO was assigned using the Lymph2Cx assay.20,29 Sixty-seven cases had adequate tissue for FISH. The FISH testing was completed with the use of Vysis (Abbott Park, IL, USA) break-apart probes for MYC and BCL2.

Statistical Analysis Research personnel scoring IHC, FISH and GEP testing were blind to all clinical data. Kaplan-Meier (KM) estimate and the logrank test were used to compare EFS and OS among different groups.34 Multivariate Cox proportional hazards models35 were constructed to evaluate potentially independent clinical and molecular predictors of EFS and OS. The analyses were conducted using SAS software package version 9.2,36 and P-values of <0.05 were considered statistically significant.

Results Patient Characteristics and Outcome The 91 patients included in the IHC study, the 97 patients in the digital GEP study, and the 67 patients in the FISH study, were representative of the 554 transformed or de novo DLBCL patients in the LY12 trial (Table 1). Initial diagnostic tissue biopsies were used in all cases except for patients where DLBCL was a result of transformation of follicular lymphoma transformation. The median time from sample collection to study randomization was 0.93 years (interquartile range: 0.47-2.18). Figure 1 demonstrates the relationship between patients analyzed by IHC, digital GEP and FISH in a Venn diagram.

Cell-of-origin COO was determined to be GCB in 44% of patients by Hans, 50.5% by Choi, and 50.5% by Tally IHC algorithms, and in 75% by Lymph2Cx GEP. There was a 73.2% concordance between the Hans IHC algorithm and GEP Lymph2Cx assay in determining COO, with both identifying GCB in 47.8% and non-GCB/ ABC in

Figure 1. Relationship between IHC, digital GEP and FISH testing, and results for overlapping cases. FISH: fluorescence in situ hybridization; IHC: immunohistochemistry; GEP: gene expression profiling.

290

haematologica | 2018; 103(2)

Bioclinical prognostic model for rrDLBCL

25.4%; while 1.5% were GCB by Hans and ABC by Lymph2Cx, and 25.3% were non-GCB by Hans and GCB by Lymph2Cx. COO was not associated with either EFS or OS, whether determined by any IHC-based algorithm (eg., Hans P=0.90 for EFS and P=0.98 for OS) or by the digital GEP Lymph2Cx assay (EFS P=0.63, OS P=0.25).

Biomarker Analysis All patients By IHC, pySTAT3 was positive in 7.7%, p53 in 19.8%, MYC in 36.3%, and BCL2 in 63.7% of cases. The overall response rate (ORR) for the 22 patients with dual protein expresser (DPE) lymphomas (MYC+/BCL2+) vs. single expressers (n=33) vs. neither MYC or BCL2 expresser (n=69) were 32.8%, 32.9% and 45.5%, respectively (P=0.51). MYC+ IHC was associated with lower 3-year EFS rates (10% vs. 42%, P=0.007) and OS rates (29% vs. 56%, P=0.002) compared to MYC- cases. Similarly, 3-year EFS rates were 25% vs. 41% (P=0.03) and OS rates were 37% vs. 63% (P=0.02) for BCL2+ vs. BCL2- cases, respectively. The 22 patients with DPE lymphomas had significantly worse 3-year EFS (0% vs. 40%, log-rank P=0.001)

and OS rates (20% vs. 54%, log-rank P=0.004) relative to the other 69 patients. In addition, p53+ vs. p53- lymphomas had 3-year EFS rates of 11% vs. 36% (P=0.03), and 3-year OS rates of 39% vs. 48% (P=0.18). pySTAT3 was not associated with EFS (P=0.25) or OS (P=0.80). There was no interaction between treatment regimen (GDP(+/-R) or DHAP(+/-R)) and COO or MYC/BCL2 expression related to OS or EFS. By GEP, the 1.5x median cut-off for MYC was 922.1 total mRNA counts (giving 24% positive cases) and for BCL2 was 2906.6 counts (giving 25% positive cases). ORR to salvage therapy for double mRNA expressers vs. single expressers vs. neither MYC or BCL2 expression were 22.2%, 21.4% and 43.3%, respectively (P=0.09). The nine patients with dual MYC/BCL2 mRNA expressing lymphomas had significantly worse 3-year EFS (0% vs. 32%, log-rank P=0.007) and OS rates (0% vs. 45%, logrank P=0.002) relative to the 88 other patients. Because we arbitrarily chose the 70% cut-off for BCL2 expression by IHC, and the 1.5x median cut-off for digital GEP analysis, we also ran sensitivity analyses using other cut-offs. We analyzed our data using a 50% cut-point for BCL2 expression by IHC, and found that the 61 (67%)

Table 2. Univariate analysis for EFS and OS.

Factors Clinical Factors GDP(+/- R) vs. DHAP(+/-R) Age > 60 years Female vs. male DLBCL vs. Transformed Extranodal Sites >1 vs. 0-1 SD/PD to initial therapy ECOG 2-3 vs. 0-1 B Symptoms Elevated LDH Stage 3-4 vs. 1-2 Digital GEP Lymph2Cx GCB MYC BCL2 PD1 PDL1 IHC Hans COO GCB BCL2 > 70% MYC > 40% pySTAT3 > 50% P53 > 50% FISH MYC (All) MYC (Double hits removed) BCL2

OS 95%CI

EFS 95%CI

0.868 1.091 0.933 1.848 1.466 2.985 1.197 1.938 2.309 1.468

0.501, 1.506 0.635, 1.877 0.561, 1.550 0.876, 3.899 0.837, 2.567 1.751, 5.102 0.568, 2.520 1.154, 3.255 1.325, 4.032 0.839, 2.571

0.62 0.75 0.79 0.11 0.18 <0.0001 0.64 0.01 0.003 0.18

1.075 1.044 0.969 1.600 0.711 2.809 0.969 1.762 1.608 1.277

0.664, 1.742 0.638, 1.710 0.610, 1.538 0.860, 2.978 0.425, 1.189 4.695, 1,689 0.481, 1.952 1.101, 2.819 0.996, 2.597 0.775, 2.104

0.77 0.86 0.89 0.14 0.19 <0.0001 0.93 0.02 0.05 0.3

1.411 2.908 2.994 2.214 1.395

0.782, 2.458 1.675, 5.045 1.749, 5.128 1.236, 3.968 0.768, 2.532

0.25 0.0001 <0.0001 0.005 0.26

1.151 2.033 2.486 1.689 1.242

0.651, 2.036 1.207, 3.425 1.507, 4.100 1.025, 2.785 0.737, 2.091

0.63 0.008 0.0004 0.03 0.41

1.007 2.036 2.278 0.879 1.409

0.578, 1.754 1.103, 3.757 1.319, 3.934 0.317, 2.438 0.725, 2.737

0.98 0.02 0.003 0.85 0.31

1.033 1.764 1.950 0.559 1.738

0.630, 1.694 1.051, 2.960 1.185, 3.208 0.203, 1.539 0.958, 3.152

0.90 0.03 0.009 0.26 0.07

1.840 2.698 1.162

0.753, 4.494 1.026, 7.095 0.604, 2.237

0.18 0.04 0.65

1.243 1.681 1.141

0.555, 2.782 0.657, 4.300 0.645, 2.018

0.60 0.28 0.65

CI: confidence interval; EFS: event-free survival; FISH: fluorescence in situ hybridization; GEP: gene expression profiling; HR: hazard ratio; IHC: immunohistochemistry; LDH: lactate dehydrogenase; OS: overall survival; ECOG: The Eastern Cooperative Oncology Group; DHAP: dexamethasone, high-dose cytarabine and cisplatin; GDP: gemcitabine, dexamethasone and cisplatin; R: rituximab; DLCBL: diffuse large B-cell lymphoma; GCB: germinal center B-cell; COO: cell-of-origin; pySTAT3: phosphoSTAT3.

haematologica | 2018; 103(2)

291

M. Bosch et al.

patients who expressed BCL2>50% had inferior EFS (hazard ratio [HR] 1.721, 95% confidence interval [CI] 1.0112.933, P=0.04) and OS (HR 1.984, 95%CI 1.058-3.717, P=0.03) compared to the 30 (33%) patients whose expression of BCL2 was <50%. This is a very similar result to that which we have reported using the 70% cut-point (Table 2), which demonstrated that the 58 (63.7%) patients who expressed BCL2>70% had inferior EFS (HR 1.764, 95%CI 1.051-2.960, P=0.03) and OS (HR 2.036, 95%CI 1.103-3.757, P=0.02) compared to the 33 (36.3%) patients whose expression of BCL2 was <70%. Regarding digital GEP, Figure 2 plots hazard ratios vs. possible thresholds and shows that the HR is predominantly over 1, suggesting that the association is robust to changes in threshold. For OS, the optimal cut-off for MYC was 504.0 counts (giving 29.9% positive), HR = 3.43 (95%CI 2.23, 5.86, P<0.0001), and for BCL2 the optimal cut-off was 2887.5 counts (giving 26.8% positive), HR = 2.91 (95%CI 1.72, 4.95, P<0.0001). For EFS, the optimal cut-off for MYC was 803.1 counts (giving 62.1% positive), HR = 2.23 (95%CI 1.35, 3.68, P=0.002) and for BCL2 the optimal cut-off was 2861.4 counts (giving 27.8% positive), HR = 2.14 (95%CI 1.32, 3.50, P=0.004). By FISH, 9/63 (14.3%) patients had MYC rearrangement and 29/64 (45.3%) had BCL2 rearrangement. There were three double hit lymphoma (DHL) patients by FISH analysis. All three proceeded to ASCT, with two patients relapsing at five and 27 months post-ASCT while the third is alive and relapse-free.

Transplanted patients There were no significant differences in ORR, transplantation rate, EFS or OS between patients treated with GDP(+/-R) or DHAP(+/-R), whereas survival was associated with MYC and BCL2 protein and mRNA expression. The 3-year OS rates were 56% vs. 92% for patients who were transplanted for IHC-determined MYC+ vs. MYC- lymphomas, respectively (log-rank P=0.0005). The 3-year OS post-ASCT was 55% for patients who underwent ASCT for DPE lymphomas compared to 88% for patients without the DPE phenotype (log-rank P=0.001). Moreover, all patients with both MYC and BCL2 overexpression by digital GEP relapsed.

Univariate analysis By univariate analysis (Table 2), the only clinical factors that were significantly associated with OS and EFS were primary refractory lymphoma (no response or progressive disease to initial chemotherapy), B symptoms, and elevated LDH. MYC and BCL2 expression by IHC or digital GEP were associated with EFS and OS. However, neither MYC nor BCL2 rearrangement by FISH were significantly associated with EFS or OS. As there were only three DHL patients, it was not feasible to determine whether DHL was associated with inferior EFS or OS.

Multivariate analysis In multivariate analyses, four factors were adversely associated with EFS and OS: primary refractory DLBCL,

Figure 2. Hazard ratios by different thresholds of MYC and BCL2 GEP for EFS and OS. Vertical dash line indicates pre-specified 1.5x median threshold used in the analysis. EFS: event-free survival; OS: overall survival; mRNA: messenger ribonucleic acid.

292

haematologica | 2018; 103(2)

Bioclinical prognostic model for rrDLBCL

elevated serum LDH at relapse, MYC expression and BCL2 expression (assessed by either IHC or digital GEP; Table 3). These four factors were associated with relatively similar HR for EFS and OS, and were therefore combined with equal weighting to create a bioclinical score that predicted ORR, EFS and OS from the initiation of salvage chemotherapy. Patients with a bioclinical score of 01 were considered low-risk, while those with a score or 24 were considered high-risk. This grouping allowed adequate numbers of patients to be analyzed within each group. Using IHC to assess MYC and BCL2, ORR was 73.5% vs. 45.3% (P=0.01), CR 41.2% vs. 30.2% (P=0.36), 3-year EFS was 55% vs. 16% (log-rank P<0.0001), and 3year OS rate was 76% vs. 26% (log-rank P<0.0001) for the 34 patients with 0-1 factor vs. the 53 with 2-4 factors (see Figure 3). Similarly, using digital GEP to assess MYC and BCL2 expression, ORR was 74.1% vs. 28.2% (P<0.0001), CR 43.1% vs. 23.1% (P=0.05), 3-year EFS rate was 46% vs. 5% (log-rank P<0.0001), and 3-year OS rate was 66% vs. 4% (log-rank P<0.0001) for the 58 patients with 0-1 factor vs. the 39 with 2-4 factors. The same four factor model predicted EFS for the 54 patients who received ASCT. Specifically, the post-transplant 3-year EFS was 68% vs. 34% for 0-1 vs. 2-4 factors, respectively (log-rank P=0.013) assessing MYC and BCL2 by IHC, and 53% vs. 18% for 0=1 vs. 2-4 factors, respec-

tively (log-rank P=0.008) when assessing MYC and BCL2 by digital GEP.

Discussion In the study herein, by using tissue samples and clinical data from patients with rrDLBCL enrolled in a prospective trial of salvage therapy prior to HDCT/ASCT, we were able to derive a clinical predictor of both response to salvage chemotherapyâ&#x20AC;&#x201D;important in the decision to proceed to transplantâ&#x20AC;&#x201D;and EFS and OS. Factors which were independently associated with EFS and OS in multivariate analysis included: primary refractory disease, elevated serum LDH at relapse, and MYC and BCL2 expression, assessed either by IHC or digital GEP. A bioclinical score using these four factors predicted EFS and OS. Cell-of-origin was not associated with EFS or OS regardless of whether it was assessed by IHC algorithms or the Lymph2Cx digital GEP assay. These results are consistent with previous publications that have reported adverse ASCT outcomes for DLBCL patients with primary refractory disease, elevated LDH (either alone or as part of the IPI score), as well as MYC and BCL2 expression.37-39 A recent retrospective study involving 331 rrDLBCL patients, of whom 132 eventually

EFS

Figure 3. Outcome of rrDLBCL according to bioclinical model score comparing 0-1 factors (low-risk) vs. 2-4 factors (high-risk), where factors include: primary refractory disease, elevated LDH, MYC expression, and BCL2 expression. Bioclinical model assessing MYC and BCL2 by IHC (Figure 3A EFS, Figure 3B OS) or by GEP (Figure 3C EFS, Figure 3D OS). EFS: event-free survival; OS: overall survival; GEP: gene expression profiling; IHC: immunohistochemistry.

haematologica | 2018; 103(2)

293

M. Bosch et al. Table 3. Multivariate analyses for OS and EFS within IHC, GEP and FISH subgroups.

Factors OS BCL2 Expression MYC Expression SD/PD to Initial Therapy Elevated LDH at Salvage Therapy EFS BCL2 Expression MYC Expression SD/PD to Initial Therapy Elevated LDH at Salvage Therapy

IHC 95%CI

1.935 2.636 3.195 3.484

1.016, 3.685 1.469, 4.730 1.730, 5.882 1.818, 6.667

0.046 0.001 0.0002 0.0002

3.526 2.755 2.899 2.545

1.872 2.081 2.519 1.900

1.085, 3.231 1.232, 3.517 1.416, 4.484 1.133, 3.175

0.024 0.006 0.002 0.015

3.336 1.763 2.299 1.976

Digital GEP 95%CI

FISH 95%CI

1.945, 6.392 <0.0001 1.487, 5.104 0.001 1.605, 5.236 0.0004 1.441, 4.505 0.001

1.090 2.364 2.604 2.786

0.529, 2.243 0.856, 6.528 1.176, 5.747 1.256, 6.173

0.82 0.10 0.02 0.01

1.878, 5.925 <0.0001 1.008, 3.086 0.047 1.330, 3.984 0.003 1.163, 3.356 0.012

1.065 1.710 1.802 1.724

0.565, 2.006 0.699, 4.182 0.883, 3.676 0.932, 3.247

0.85 0.24 0.11 0.08

CI: confidence interval; EFS; event-free survival; FISH; fluorescence in situ hybridization; GEP: gene expression profiling; HR: hazard ratio; IHC: immunohistochemistry; LDH: lactate dehydrogenase; OS: overall survival; SD/PD: stable disease/progressive disease.

received ASCT, reported that primary progression during rituximab, cyclophosphamide, doxorubicin, vincristine and prednisone (R-CHOP) treatment, MYC translocation by FISH and IPI=3-5 at time of salvage therapy negatively affected survival, whereas COO did not.38 Another retrospective study evaluated ASCT outcomes in 117 patients with chemotherapy-sensitive rrDLBCL, and reported inferior outcomes for those with dual protein expression (44% of patients) as well as DHL(10%).39 Novel approaches are required for these patients.24,38-40 Our study had several limitations, including the small sample size of less than 100 cases with adequate tissue available for biomarker evaluation. Although this was a relatively small proportion of patients who enrolled in the LY12 clinical trial, the baseline characteristics, treatments, and outcomes of the biomarker study population were similar to the overall LY12 population of patients with aggressive B-cell lymphoma. We included transformed disease to maximize sample size as the outcomes of transformed cases were similar to DLBCL in univariate and multivariate analyses for both our IHC- and GEP-defined study populations, as well as in the overall LY12 population of 87 transformed cases and 429 rrDLBCL cases.41 Another limitation was the fact that the biomarkers were assessed retrospectively. These assessments, however, were made blinded to all clinical factors and outcomes, and patients were treated in a prospective randomized clinical trial. A third limitation relates to the thresholds chosen for positivity and negativity of IHC and digital GEP. Although our thresholds for IHC protein expression were consistent with published literature, several different cutpoints have been reported for BCL2 positivity.42-44 Despite our choice of the >70% cut-off for BCL2 positivity by IHC, approximately 2/3 of our cases were considered positive for BCL2 protein expression, with similar expression rates between ABC and GCB COO subtypes (P=0.79). This rate of BCL2 positivity using the >70% cut-off is similar to some prior studies.42,43 Our sensitivity analysis demonstrated similar results using a >50% IHC cut-off for BCL2. Although MYC expression by IHC and GEP was fairly similar in our study, approximately 40% of our cases were positive for BCL2 by IHC but considered negative by digital GEP. This finding may be a result of our relatively high pre-specified 1.5x median threshold for GEP positiv294

ity. In order to reduce the risk of overfitting, we did not want to analyze our data according to post hoc optimization of multiple cut-offs for each biomarker and the endpoint of interest (eg., rates of response, ASCT, EFS or OS).45,46 Instead we analyzed the data by a pre-specified, albeit arbitrary threshold for GEP positivity. Sensitivity analyses showed that the results using other cut-offs, including optimal cut-offs, were consistent with our prespecified cut-off, giving HR in the same direction and the 95%CI in a similar range for EFS and OS. Our choice of 1.5x median cut-off for digital GEP worked well in predicting EFS and OS in univariate and multivariate analyses, as well as the bioclinical score, but requires confirmation in a validation cohort. A final limitation of our study is the lack of a validation cohort, due to the modest sample size available. However, there have already been extensive publications concerning the prognostic impact of COO as well as MYC and BCL2 expression in DLBCL that generally support our findings, as discussed below. In addition, we plan to evaluate the proposed bioclinical model in our ongoing clinical trial evaluating the addition of novel agents to GDP salvage therapy for rrDLBCL. A unique aspect of this study was the evaluation of MYC and BCL2 abnormalities by IHC, GEP, and FISH in the same patient samples. Much of the literature reporting the poor prognosis of DHLs relates to newly diagnosed DLBCL.22,24,27,28 Analysis of our FISH cohort, however, contained only three DHL patients (4.5%), of whom two relapsed post-ASCT. In contrast, IHC identified 22 (24.2%) patients with DPE lymphoma and digital GEP identified 9 (9.3%) patients with MYC/BCL2 dual gene expression, of whom all relapsed and died after salvage therapy, while patients without dual MYC/BCL2 expression had an excellent 88% OS after transplant. If further validated, this finding may help clinicians determine who would or would not benefit from transplant. We were unable to determine any association between COO and EFS or OS using accepted IHC algorithms or the NanoString GEP Lymph2Cx assay. It is possible that this might relate to our small sample size and low power to detect an association. Although we assessed diagnostic tissue samples for their relationship to outcome of second-line therapy, it has previously been demonstrated that biopsies haematologica | 2018; 103(2)

Bioclinical prognostic model for rrDLBCL

at diagnosis and relapse have similar COO phenotypes.19 It is possible that the acquisition of other mutations or activation of pathways that are integral to drug resistance may make COO less relevant in the relapse setting. Of note, other groups have also failed to demonstrate a significant association between COO and ASCT outcomes for DLBCL.47-49 Although the bio-CORAL study suggested that COO may be associated with the outcome of salvage therapy for rrDLBCL patients, it was only the specific interaction between rituximab - dexamethasone, cytarabine and cisplatin (R-DHAP) salvage therapy and GCB-like DLBCL (based on the Hans algorithm) that was associated with better progression-free survival (PFS), but no such association was found for the rituximab - ifosfamide, carboplatin and etoposide (R-ICE) regimen.19 The relative greater prognostic importance of MYC and BCL2 expression over COO is also supported by the recently reported German High Grade Lymphoma Study Group (DSHNHL) retrospective study which evaluated the Nanostring Lymph2Cx COO assay in patient samples from the prospective front-line RICOVER-60 (n=326) and R-MegaCHOEP (n=88) clinical trials, and found that COO profiling failed to identify prognostic subgroups, whereas MYC/BCL2 double expression by IHC was highly predictive of poor survival.50

References 1. Sujobert P, Salles G, Bachy E. Molecular classification of diffuse large B-cell lymphoma: what Is clinically relevant? Hematol Oncol Clin North Am. 2016; 30(6):1163-1177. 2. Zhou Z, Sehn LH, Rademaker AW, et al. An enhanced International Prognostic Index (NCCN-IPI) for patients with diffuse large B-cell lymphoma treated in the rituximab era. Blood. 2014;123(6):837-842. 3. Crump M, Kuruvilla J, Couban S, et al. Randomized comparison of gemcitabine, dexamethasone, and cisplatin versus dexamethasone, cytarabine, and cisplatin chemotherapy before autologous stem-cell transplantation for relapsed and refractory aggressive lymphomas: NCIC-CTG LY.12. J Clin Oncol. 2014;32(31):3490-3496. 4. Gisselbrecht C, Glass B, Mounier N, et al. Salvage regimens with autologous transplantation for relapsed large B-cell lymphoma in the rituximab era. J Clin Oncol. 2010;28(27):4184-4190. 5. Martin A, Conde E, Arnan M, et al. RESHAP as salvage therapy for patients with relapsed or refractory diffuse large B-cell lymphoma: the influence of prior exposure to rituximab on outcome. A GEL/TAMO study. Haematologica. 2008;93(12):18291836 6. Hernandez-Ilizaliturri FJ, Czuczman MS. Therapeutic options in relapsed or refractory diffuse large B-cell lymphoma. Part 1. current treatment approaches. Oncology (Williston Park). 2009;23(6):546-553. 7. Elstrom RL, Martin P, Ostrow K, et al. Response to second-line therapy defines the potential for cure in patients with recurrent diffuse large B-cell lymphoma: implications for the development of novel therapeutic strategies. Clin Lymphoma Myeloma Leuk. 2010;10(3):192-196.

haematologica | 2018; 103(2)

In conclusion, we combined important prognostic clinical factors with molecular biomarkers to create a novel bioclinical score that predicts outcome of salvage therapy for rrDLBCL. The model described herein identified a group of patients with 2-4 factors who had very poor 3year EFS, whether MYC and BCL2 were assessed by IHC or by GEP. Future research is warranted to validate these findings in another dataset, and to evaluate novel agents and treatment approaches for patients who have poor prognosis rrDLBCL with dual MYC/BCL2 expression or a high-risk bioclinical score. Acknowledgments The authors wish to thank Susana Ben-Neriah and Brett Collinge who performed the FISH analyses, as well as Marina Djurnfeldt who managed the LY12 clinical database, and Shakeel Virk who retrieved and managed the tissue samples in the CCTG tissue bank. Funding Canadian Cancer Society Research Institute grants 021039 and 704970. Alberta Cancer Foundation (ACF) grants 25987 and 26070 (Dr. Anthony Fields Fellowship Award for Mark Bosch).

8. Oliansky DM, Czuczman M, Fisher RI, et al. The role of cytotoxic therapy with hematopoietic stem cell transplantation in the treatment of diffuse large B cell lymphoma: update of the 2001 evidence-based review. Biol Blood Marrow Transplant. 2011;17(1):20-47. 9. Rigacci L, Fabbri A, Puccini B, et al. Oxaliplatin-based chemotherapy (dexamethasone, high-dose cytarabine, and oxaliplatin) Âą rituximab is an effective salvage regimen in patients with relapsed or refractory lymphoma. Cancer. 2010;116(19): 4573â&#x20AC;&#x201C;4579. 10. Morin RD, Assouline S, Alcaide M, et al. Genetic landscapes of relapsed and refractory diffuse large B-cell lymphomas. Clin Cancer Res. 2016;22(9):2290-2300. 11. Hans CP, Weisenburger DD, Greiner TC, et al. Confirmation of the molecular classification of diffuse large B-cell lymphoma by immunohistochemistry using a tissue microarray. Blood. 2004;103(1):275-282. 12. Choi WW, Weisenburger DD, Greiner TC, et al. A new immunostain algorithm classifies diffuse large B-cell lymphoma into molecular subtypes with high accuracy. Clin Cancer Res. 2009;15(17):5494-5502. 13. Meyer PN, Fu K, Greiner TC, et al. Immunohistochemical methods for predicting cell of origin and survival in patients with diffuse large B-cell lymphoma treated with rituximab. J Clin Oncol. 2011;29(2): 200-207. 14. Winter JN. Prognostic markers in diffuse large B-cell lymphoma: Keys to the underlying biology. Curr Hematol Malig Rep. 2007;2(4):235-241. 15. Mounier N, Briere J, Gisselbrecht C, et al. Rituximab plus CHOP (R-CHOP) overcomes bcl-2--associated resistance to chemotherapy in elderly patients with diffuse large B-cell lymphoma (DLBCL). Blood. 2003;101(11):4279-4284. 16. de Jong D, Rosenwald A, Chhanabhai M,

17.

18.

19.

20.

21.

22.

23.

et al. Immunohistochemical prognostic markers in diffuse large B-cell lymphoma: validation of tissue microarray as a prerequisite for broad clinical applications--a study from the Lunenburg Lymphoma Biomarker Consortium. J Clin Oncol. 2007;25(7):805-812. de Jong D, Xie W, Rosenwald A, et al. Immunohistochemical prognostic markers in diffuse large B-cell lymphoma: validation of tissue microarray as a prerequisite for broad clinical applications (a study from the Lunenburg Lymphoma Biomarker Consortium). J Clin Pathol. 2009;62(2):128138. Fu K, Weisenburger DD, Choi WW, et al. Addition of rituximab to standard chemotherapy improves the survival of both the germinal center B-cell-like and non-germinal center B-cell-like subtypes of diffuse large B-cell lymphoma. J Clin Oncol. 2008;26(28):4587-4594. Thieblemont C, Briere J, Mounier N, et al. The germinal center/activated B-cell subclassification has a prognostic impact for response to salvage therapy in relapsed/refractory diffuse large B-cell lymphoma: a bio-CORAL study. J Clin Oncol. 2011;29(31):4079-4087. Scott DW, Wright GW, Williams PM, et al. Determining cell-of-origin subtypes of diffuse large B-cell lymphoma using gene expression in formalin-fixed paraffinembedded tissue. Blood. 2014;123(8):12141217. Scott DW. Cell-of-origin in diffuse large Bcell lymphoma: are the assays ready for the clinic? Am Soc Clin Oncol Educ Book. 2015:e458-466. Sesques P, Johnson NA. Approach to the diagnosis and treatment of high-grade Bcell lymphomas with MYC and BCL2 and/or BCL6 rearrangements. Blood. 2017;129(3):280-288. Perry AM, Alvarado-Bernal Y, Laurini JA, et

295

M. Bosch et al.

24.

25.

26.

27.

28.

29.

30.

31.

296

al. MYC and BCL2 protein expression predicts survival in patients with diffuse large B-cell lymphoma treated with rituximab. Br J Haematol. 2014;165(3):382-391. Hu S, Xu-Monette ZY, Tzankov A, et al. MYC/BCL2 protein coexpression contributes to the inferior survival of activated B-cell subtype of diffuse large B-cell lymphoma and demonstrates high-risk gene expression signatures: a report from The International DLBCL Rituximab-CHOP Consortium Program. Blood. 2013;121(20):4021-4031. Green TM, Young KH, Visco C, et al. Immunohistochemical double-hit score is a strong predictor of outcome in patients with diffuse large B-cell lymphoma treated with rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone. J Clin Oncol. 2012;30(28):3460-3467. Agarwal R, Lade S, Liew D, et al. Role of immunohistochemistry in the era of genetic testing in MYC-positive aggressive B-cell lymphomas: a study of 209 cases. J Clin Pathol. 2016;69(3):266-270. Horn H, Ziepert M, Becher C, et al. MYC status in concert with BCL2 and BCL6 expression predicts outcome in diffuse large B-cell lymphoma. Blood. 2013;121(12):2253-2263. Ennishi D, Mottok A, Ben-Neriah S, et al. Genetic profiling of MYC and BCL2 in diffuse large B-cell lymphoma determines cellof-origin-specific clinical impact. Blood. 2017;129(20):2760-2770. Scott DW, Mottok A, Ennishi D, et al. Prognostic significance of diffuse large Bcell lymphoma cell of origin determined by digital gene expression in formalin-fixed paraffin-embedded tissue biopsies. J Clin Oncol. 2015;33(26):2848-2856. Cortez MA, Scrideli CA, Yunes JA, et al. mRNA expression profile of multidrug resistance genes in childhood acute lymphoblastic leukemia. Low expression levels associated with a higher risk of toxic death. Pediatr Blood Cancer. 2009;53(6):996-1004. Zhou K, Yi S, Yu Z, et al. MicroRNA-223 expression is uniformly down-regulated in B cell lymphoproliferative disorders and is associated with poor survival in patients with chronic lymphocytic leukemia. Leuk Lymphoma. 2012;53(6):1155-1161.

32. Sauerbrey A, Voigt A, Wittig S, Hafer R, Zintl F. Messenger RNA analysis of the multidrug resistance related protein (MRP1) and the lung resistance protein (LRP) in de novo and relapsed childhood acute lymphoblastic leukemia. Leuk Lymphoma. 2002;43(4):875-879. 33. Lin SC, Gan ZH, Yao Y, Min da L. The prognostic value of forkhead box P3 expression in operable breast cancer: a large-scale meta-analysis. PLoS One. 2015;10(8):e0136374. 34. Kaplan ELM, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc. 1958;53(282):457â&#x20AC;&#x201C;481. 35. Cox DR. Regression Models and LifeTables. J Royal Stat Soc. Series B (Methodological). 1972;34(2):187â&#x20AC;&#x201C;220. 36. SAS Institute Inc. C, NC, USA. SAS software package version 9.2; 2013. 37. Vardhana SA, Sauter CS, Matasar MJ, et al. Outcomes of primary refractory diffuse large B-cell lymphoma (DLBCL) treated with salvage chemotherapy and intention to transplant in the rituximab era. Br J Haematol. 2017;176(4):591-599. 38. Costa LJ, Maddocks K, Epperla N, et al. Diffuse large B-cell lymphoma with primary treatment failure: Ultra-high risk features and benchmarking for experimental therapies. Am J Hematol. 2017;92(2):161170. 39. Herrera AF, Mei M, Low L, et al. Relapsed or refractory double-expressor and doublehit lymphomas have inferior progressionfree survival after autologous stem-cell transplantation. J Clin Oncol. 2017; 35(1):24-31. 40. Smith SM. Impact of double-hit and doubleexpressor phenotypes in relapsed aggressive B-cell lymphomas treated with autologous hematopoietic stem cell transplantation. J Clin Oncol. 2017;35(1):1-3. 41. Kuruvilla J, MacDonald DA, Kouroukis CT, et al. Salvage chemotherapy and autologous stem cell transplantation for transformed indolent lymphoma: a subset analysis of NCIC CTG LY12. Blood. 2015;126(6):733-738. 42. Xie Y, Bulbul MA, Ji L, et al. p53 expression is a strong marker of inferior survival in de novo diffuse large B-cell lymphoma and may have enhanced negative effect with

43.

44.

45.

46.

47.

48.

49.

50.

MYC coexpression: a single institutional clinicopathologic study. Am J Clin Pathol. 2014;141(4):593-604. Perry AM, Alvarado-Bernal Y, Laurini JA, et al. MYC and BCL2 protein expression predicts survival in patients with diffuse large B-cell lymphoma treated with rituximab. Br J Haematol. 2014;165(3):382-391. Johnson NA, Slack GW, Savage KJ, et al. Concurrent expression of MYC and BCL2 in diffuse large B-cell lymphoma treated with rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone. J Clin Oncol. 2012;30(28):3452-3459. Chen BE, Jiang W, Tu D. (2014). A hierarchical Bayes model for biomarker subset effects in clinical trials. Computational Statistics and Data Analysis. 2014;71:324334. Fang T, Mackillop W, Jiang W, Hildesheim A, Wacholder S, Chen BE. A Bayesian method for risk window estimatin with application to HPV vaccine trial. Computational Statistics and Data Analysis. 2017;112:53-62. Costa LJ, Feldman AL, Micallef IN, et al. Germinal center B (GCB) and non-GCB cell-like diffuse large B cell lymphomas have similar outcomes following autologous haematopoietic stem cell transplantation. Br J Haematol. 2008;142(3):404412. Gu K, Weisenburger DD, Fu K, et al. Cell of origin fails to predict survival in patients with diffuse large B-cell lymphoma treated with autologous hematopoietic stem cell transplantation. Hematol Oncol. 2012; 30(3):143-149. Nyman H, Jantunen E, Juvonen E, et al. Impact of germinal center and non-germinal center phenotypes on overall and failure-free survival after high-dose chemotherapy and auto-SCT in primary diffuse large B-cell lymphoma. Bone Marrow Transplant. 2008;42(2):93-98. Staiger AM, Ziepert M, Horn H, et al. Clinical impact of the cell-of-origin classification and the MYC/ BCL2 dual expresser status in diffuse large B-cell lymphoma treated within prospective clinical trials of the German High-Grade Non-Hodgkin's Lymphoma Study Group. J Clin Oncol. 2017;35(22):2515-2526.

haematologica | 2018; 103(2)

ARTICLE

Non-Hodgkin Lymphoma

Outcomes among North American patients with diffuse large B-cell lymphoma are independent of tumor Epstein-Barr virus positivity or immunosuppression Sean I. Tracy,1 Thomas M. Habermann,1 Andrew L. Feldman,1 Matthew J. Maurer,1 Ahmet Dogan,2 Usha S. Perepu,3 Sergei Syrbu,3 Stephen M. Ansell,1 Carrie A. Thompson,1 George J. Weiner,3 Grzegorz S. Nowakowski,1 Cristine Allmer,1 Susan L. Slager,1 Thomas E. Witzig,1 James R. Cerhan1 and Brian K. Link3

Mayo Clinic, Rochester, MN; 2Memorial Sloan Kettering Cancer Center, New York, NY and 3University of Iowa, Iowa City, IA, USA 1

Ferrata Storti Foundation

Haematologica 2018 Volume 103(2):297-303

ABSTRACT

he prevalence, presenting clinical and pathological characteristics, and outcomes for patients with diffuse large B-cell lymphoma that is Epstein-Barr virus positive remain uncertain as does the impact of congenital or iatrogenic immunosuppression. Patients with newly diagnosed diffuse large B-cell lymphoma with available tissue arrays were identified from the University of Iowa/Mayo Clinic Molecular Epidemiology Resource. Patients infected with human immunodeficiency virus or who had undergone a prior organ transplant were excluded. Epstein-Barr virus-associated ribonucleic acid testing was performed on all tissue arrays. A history of significant congenital or iatrogenic immunosuppression was determined for all patients. At enrollment, 16 of the 362 (4.4%) biopsies were positive for Epstein-Barr virus. Thirtynine (10.8%) patients had a significant history of immunosuppression. Patients with Epstein-Barr-positive diffuse large B-cell lymphoma had no unique clinical characteristics but on pathology exhibited a higher frequency of CD30 positivity (25.0% versus 8.1%, respectively; P<0.01), and non-germinal-center subtype (62.5% versus 34.1%, respectively; P<0.01). No baseline clinical characteristics were associated with a history of immunosuppression. With a median follow up of 59 months, and after adjustment for International Prognostic Index, there was no association of Epstein-Barr virus positivity or immunosuppression with eventfree survival at 24 months (odds ratio=0.49; 95% confidence interval: 0.13-1.84 and odds ratio=0.81; 95% confidence interval: 0.37-1.77) or overall survival (hazard ratio=0.86; 95% confidence interval: 0.38-1.97 and hazard ratio=1.00; 95% confidence interval: 0.57-1.74). In contrast to non-Western populations, our North American population had a low prevalence of Epstein-Barr virus-positive diffuse large B-cell lymphoma that did not convey an adverse prognosis. A history of immunosuppression, while known to be a risk factor for the development of diffuse large B-cell lymphoma, did not affect subsequent prognosis.

Presented in abstract form at the 58th annual meeting of the American Society of Hematology, San Diego, CA December 3rd 2016

Correspondence: brian-link@uiowa.edu

Received: July 14, 2017. Accepted: November 22, 2017. Pre-published: November 23, 2017. doi:10.3324/haematol.2017.176511 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/297 ÂŠ2018 Ferrata Storti Foundation

Introduction Diffuse large B-cell lymphoma (DLBCL) displays substantial clinical and pathological heterogeneity. Epstein-Barr virus (EBV)-positive DLBCL of the elderly (EDLBCLe) was a provisional entity in the 2008 World Health Organization (WHO) classification of tumors of hematopoietic and lymphoid tissues1 defined as an EBV+ clonal B-cell lymphoid proliferation that occurs in patients older than 50 years, without any known immunodeficiency or prior lymphoma. EDLBCLe has been found to account for 2-16% of all DLBCL in Asia or Latin countries, with most studiesâ&#x20AC;&#x2122; estimates at the higher end of this range.2-7 Such reports generally haematologica | 2018; 103(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.

297

S.I. Tracy et al.

describe an association of EBV positivity with advanced stage, extranodal disease, constitutional symptoms and low rates of response to chemotherapy, with resultant poor survival.4-6,8 In contrast, studies focused on patients in Western countries have generally found both lower prevalence rates and weaker associations of EBV with aggressive clinical features or inferior outcomes.9-11 Direct comparison of these studies is limited by their inclusion of selected cases rather than prospective cohorts. Furthermore, multiple studies from both Western and non-Western populations have now also identified EBV+ DLBCL in immunocompetent younger patients, in addition to elderly ones.4,8 The 2016 WHO classification was revised to recognize the wide age spectrum of patients affected by this condition, and “EBV+ DLBCL of the elderly” has been replaced by “EBV+ DLBCL, not otherwise specified” (EDLBCL-NOS).12,13 A further area of uncertainty concerns the relative impact of EBV positivity versus immunosuppression itself on adverse outcomes among patients with DLBCL. EBV+ lymphomas arise more frequently in patients with compromised immune systems, as exemplified in EBV+ posttransplant lymphoproliferative disease. However, immunosuppression due to iatrogenic agents or congenital immunodeficiency has also been shown to be a risk factor for EBV– DLBCL.14,15 More broadly, it is unknown whether immunosuppression is also associated with poor outcomes among DLBCL patients in general. In light of these considerable uncertainties, we sought to define the prevalence, clinical correlations, and prognosis of EBV+ DLBCL among a prospectively assembled cohort of patients from the upper Midwestern area of the USA. To delineate the effects of immune dysfunction from those of EBV, standardized definitions were developed for clinically significant immunosuppression, and a subgroup analysis of immunocompetent and immunosuppressed patients was performed. This study represents a large systematic predominantly prospective evaluation of EDLBCL-NOS in the USA, as well as the first independent examination of the effects of immune suppression on outcomes among patients with DLBCL.

Methods Study population This study was approved by institutional review boards at the University of Iowa and Mayo Clinic. Written informed consent was obtained from all participants. This study utilized the Molecular Epidemiology Resource of the University of Iowa/Mayo Clinic Lymphoma Specialized Program of Research Excellence which has been previously described.16,17 Briefly, starting from September 2002, we offered enrollment to consecutive, newly diagnosed adult patients with lymphoma evaluated at the University of Iowa or Mayo Clinic Rochester. All diagnoses were confirmed by study hematopathologists. Diagnostic tissue blocks from DLBCL cases were collected and tissue microarrays were constructed using three 0.6 mm cores from all cases with sufficient tissue to be used in subsequent bulk analyses. Baseline clinical, laboratory, and treatment data were collected. All participants were systematically contacted for follow-up every 6 months for the first 3 years, and then annually thereafter. Disease progression, retreatment, and death were verified through review of medical records. Inclusion criteria for this analysis were initial diagnosis of de novo or transformed DLBCL enrolled from 2002-2012, with tis298

sue microarrays. Patients with a primary central nervous system lymphoma, primary cutaneous lymphoma, or primary mediastinal large B-cell lymphoma were excluded, as were patients with a history of organ transplant or known infection with human immunodeficiency virus.

Immunohistochemistry and in situ hybridization Immunohistochemistry for CD30, CD10, bcl-6, and MUM1 was centrally scored on tissue microarrays. Cell of origin was determined according to the Hans criteria.18 The cutoff for CD30 positivity was 20% of neoplastic cells.19 EBV testing was performed by in situ hybridization for EBER with a threshold of 30% of neoplastic cells, and scored by an expert hematopathologist (AD, AF).20

Definition of immunosuppression or immunodeficiency The prospectively collected information was augmented for this analysis with a retrospective chart review focused on evidence of immunosuppression for each patient included. One patient with a history of congenital immunodeficiency was identified. Patients with a documented history of prior treatment with methotrexate, cyclophosphamide, azathioprine, hydroxychloroquine, antiepileptic agents, or biologic agents including anti-tumor necrosis factor monoclonal antibodies were considered to have received iatrogenic immunosuppression, as were patients who had received a lifetime exposure to corticosteroids equal to or greater than 6 months of daily prednisone at a dose of 20 mg/day. Patients with a seizure history treated with antiepileptic drugs were also considered immunodeficient based on papers describing quantitative and qualitative defects in circulating lymphocytes associated with antiepileptic therapy.21-23

Statistical analysis EBV and immunosuppression status were correlated with clinical features using Wilcoxon signed-rank and chi-square testing, where appropriate. Events were defined as documented DLBCL progression, subsequent anti-lymphoma therapy, or death. Overall survival was defined as time from diagnosis to death. EFS24 was defined as event-free survival status at 24 months after diagnosis.24 Survival estimates were obtained with the Kaplan– Meier method. Survival comparisons between and within positive, negative, and indeterminate groupings as well as International Prognostic Index (IPI) groupings were performed with log-rank tests. The associations of EBV and immunosuppression history with event-free and overall survival were estimated using hazard ratios (HR) and 95% confidence intervals (CI) for Cox models, adjusted for IPI; logistic regression and odds ratios (OR) were used for associations with EFS24. All tests were twosided and assessed for significance at the 5% level.

Results Patients’ characteristics, prevalence estimates and clinical correlates From 2002 through 2012, 1,081 patients with DLBCL were enrolled into the Molecular Epidemiology Resource, of whom 362 had diagnostic tissue on an available tissue microarray, met the study selection criteria, and form the population analyzed for this report (Figure 1). As compared to patients enrolled in the Molecular Epidemiology Resource who did not have sufficient tissue available for tissue microarray construction, study cohort patients had lower Ann Arbor stages, lactate dehydrogenase values, IPI scores, longer follow-up, and more often achieved EFS24 haematologica | 2018; 103(2)

DLBCL with EBV or immunosuppression

Figure 1. CONSORT diagram of patients included in the study.

(Online Supplementary Table S1). Treatment regimens utilized for the DLBCL patients were largely anthracyclinebased immunochemotherapy as described in more detail elsewhere.17 The median age at enrollment of the cohort analyzed was 63 years (range, 20-89) and 59% were male. EBV testing was positive in 16 (4.4%) of the cases. The baseline characteristics by tissue EBV positivity are detailed in Table 1. Fifteen of the 16 EBV+ patients were treated initially with rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) and one with etoposide and procarbazine substituting for doxorubicin and vincristine (R-CEPP). The age and gender distributions, performance status, Ann Arbor Stage, IPI scores, numbers of affected extranodal sites, and lactate dehydrogenase levels of the EBV– cases were similar to those of the EBV+ cases. Bone marrow involvement with large-cell lymphoma was more common among patients with EBV+ disease (43.8% versus 18.5%; P=0.03). Non-germinal center B-cell subtype disease was more frequent among EBV+ DLBCL than among EBV– DLBCL (62.5% versus 34.1%; P<0.01). CD30 positivity was also more frequent among EBV+ cases (25.0% versus 8.1%; P<0.01). Including all patients regardless of EBV tumor status, 39 (10.8%) of the analysis cohort were considered immunosuppressed following our systematic review. Of these 39 patients 38 had received significant iatrogenic immunosuppression – though none with rituximab, while one patient had a history of common variable immunodeficiency; Online Supplementary Table S2 details the types of immunosuppressive agents administered, as well as their indications. There were no differences in baseline charachaematologica | 2018; 103(2)

teristics between immunocompetent versus immunosuppressed patients (Table 1). The management of immunosuppressive agents after the diagnosis of DLBCL was highly variable. Seven of the 362 DLBCL cases met the 2008 WHO criteria for EDLBCLe (EBV+, age >50 years at the time of diagnosis, and immunocompetent), a prevalence rate of 1.9%. Twelve of the 362 cases met the 2016 WHO criteria for EBV+ DLBCL-NOS (EBV+, immunocompetent), a prevalence rate of 4.4%.

Prognosis During a median follow-up of 59 months (range, 0.7155.2), there were 150 events including 119 deaths. The median overall survival for all 362 patients was 11.7 years. Analyzing the influence of EBV status, it was seen that the median overall survival was similar in patients with EBV+ or EBV– DLBCL (129 versus 143 months, respectively; P=0.97). EBV+ status was not associated with adverse overall survival (HR adjusted for IPI=0.86; 95% CI: 0.381.97) (Figure 2A). The median event-free survival was also similar in EBV+ and EBV– DLBCL patients (129 versus 138 months, P=0.51; HR adjusted for IPI =0.64; 95% CI; 0.281.46) (Figure 2D). Kaplan-Meier estimates for the percentage of patients reaching EFS24 were 81% for EBV+ patients, and 71% for EBV– patients (P=0.52; OR adjusted for IPI=0.49; 95% CI: 0.13-1.84). In order to evaluate the prognostic importance of immunosuppression, outcomes were compared between the 323 immunocompetent patients and 39 patients with a history of immunosuppression. Outcomes were not affected by a history of immunosuppression with the 299

S.I. Tracy et al.

median overall survival being 129 months for patients with a history of immunosuppression versus 143 months for immunocompetent patients (P=0.95; HR adjusted for IPI=1.00; 95% CI: 0.57-1.74) and the median event-free survival being 103 months versus 138 months, respectively (HR adjusted for IPI=1.09; 95% CI: 0.67-1.76) (Figure 3). Kaplan-Meier estimates for the percentage of patients reaching EFS24 were 71% for immunocompetent patients versus 74% for patients with a history of immunosuppression (P=0.53; OR adjusted for IPI= 0.81; 95% CI: 0.371.77).

Discussion We describe one of the largest systematic evaluations of EBV+ DLBCL in a North American human immunodeficiency virus-negative population in the modern immunochemotherapy era, and the first using 2016 updated WHO definitions of EBV+ DLBCL-NOS. As a means of distinguishing the relative prognostic importance of immunosuppression from that of EBV positivity, we additionally report a survival analysis of patients with DLBCL occurring on a background of significant iatrogenic

immune suppression or congenital immunodeficiency. Overall, EBV+ DLBCL occurred infrequently within the studied population. EBV+ DLBCL was found in patients of all age groups. In contrast to multiple reports focused on Asian or Latin patients, we did not observe an association of EBV positivity with aggressive presentation or adverse prognosis. Finally, while immunosuppression is a known risk factor for the development of both EBV+ and EBV– DLBCL, there was no evidence that immunosuppression was independently associated with adverse outcomes. The prevalence of EBV+ DLBCL varies across geographical regions. Estimates of the prevalence of EDLBCLe among patients from Asian or Latin countries range from 2-12%. In contrast, a previous screening study in the USA identified five selected patients with EBV+ DLBCL, but found no further cases in screening of 90 unselected older patients with DLBCL.25 Ok et al., in a recent study of 732 patients of all age groups with DLBCL from developed Western countries, identified 28 (4.0%) as EBV+.26 Our prevalence estimates of 4.4% for EBV+ DLBCL-NOS, and 1.9% for EDLBCLe, are generally in agreement with prior reports from Western countries. Previous series have described distinct clinical and pathological characteristics associated with EBV positivi-

Table 1. Patients’ demographics and clinical characteristics.

Characteristic

Total (n = 362)

Age at diagnosis, years Median 63 Range 20-89 Age ≤50 years 68 (19%) >50 years 294 (81%) Gender Female 147 (42%) Male 215 (59%) Residence Upper Midwest 298 (82%) Other MW States 64 (18%) Performance Status Missing 1 (0.3%) <2 291 (80%) ≥2 70 (19%) Ann Arbor Stage Missing 3 (0.8%) I-II 144 (39.8%) III-IV 215 (59.4%) Number of extranodal sites ≤1 297 (82%) >1 65 (18%) Lactate dehydrogenase Missing 40 (11%) ≤ Normal 163 (45%) > Normal 159 (44%) IPI internation prognostic index score 0-1 130 (36%) 2 109 (30%) 3 86 (24%) 4-5 37 (10%)

EBV+ (n = 16)

EBV(n = 346)

58 25-78

63 20-89

5 (31%) 11 (69%)

63 (18%) 283(82%)

6 (38%) 10 (62%)

141 (41%) 202 (58%)

11 (69%) 5 (31%)

287 (83%) 59 (17%)

0 (0%) 12 (75%) 4 (25%)

1 (0.3%) 279 (81%) 66 (19%)

0 (0%) 4 (25%) 12 (75%)

3 (0.9%) 140 (40%) 203 (59%)

13 (81%) 3 (19%)

284 (82%) 62 (18%)

1 (6%) 8 (50%) 8 (50%)

39 (11%) 155 (45%) 152 (44%)

4 (25%) 6 (38%) 5 (31%) 1 (6%)

126 (36%) 103 (30%) 81 (23%) 36 (10%)

P-value

Immunosuppressed (n = 39)

Immunocompetent (n =323)

64 42-79

63 20-89

4 (10%) 35 (90%)

64 (20%) 259 (80%)

20 (51%) 19 (49%)

127 (39%) 196 (61%)

29 (74%) 10 (26%)

269 (83%) 54 (17%)

0 (0%) 33 (85%) 6 (15%)

1 (0.3%) 258 (80%) 64 (20%)

0 (0%) 14 (36%) 25 (64%

3 (1%) 130 (40%) 190 (59%)

33 (85%) 6 (15%)

264 (82%) 59 (18%)

4 (10%) 14 (36%) 21 (54%) 0.68 11 (28%) 16 (41%) 9 (23%) 3 (8%)

36 (11%) 149 (46%) 138 (43%)

0.18

0.64

0.19

0.15

0.80

0.15

P-value

0.15

0.82

0.17 0.75

0.42

0.71

0.93

0.66

0.80

0.40

0.43 119 (37%) 93 (29%) 77 (24%) 34 (10%) continued on the next page

300

haematologica | 2018; 103(2)

DLBCL with EBV or immunosuppression

ty.5-7,10-14 In agreement with prior reports, we found an association of EBV positivity with the non-germinal center B-cell subtype of DLBCL, as well as with CD30 positivity.26 However, we did not find evidence of an association of EBV positivity with aggressive clinical features such as B symptoms, multiple sites of extranodal involvement, or older age. Several reports focused on Asian and Latin patients have suggested an association of EBV positivity with an overall worse clinical prognosis. Older large Korean and Peruvian series demonstrated inferior outcomes for EBV+ DLBCL patients.5,27 In the immunochemotherapy era, at least two subsequent studies continued to demonstrate inferior outcomes among Asian patients treated with R-CHOP.28,29 Our findings are more consistent with those of a previous study of patients from Western populations, which found no evidence of an association of EBV positivity with adverse outcomes.26 The numbers of EBV+ DLBCL patients and events included in our study were low and, accordingly, the results should be considered more descriptive than conclusive, but comparable to those of other studies which demonstrated significant differences in outcomes, suggesting adequate power.8,28 To date, no consensus definition has been adopted by the WHO to define significant clinical immunosuppression, leading to the variable inclusion of patients with a history of

immunocompromised states in previous studies of EBV+ DLBCL. This suggested a possible confounding of immunosuppression with EBV status in prior reports associating EBV positivity with poor outcomes. To explore this hypothesis, we developed standard definitions of clinically significant immunosuppression, and used them to identify such patients in our cohort. Direct comparison of immunosuppressed and immunocompetent DLBCL patients revealed similar outcomes, and subgroup analysis of immunocompetent EBV+ versus EBVâ&#x20AC;&#x201C; DLBCL patients also found no difference. We conclude that among predominantly white North American patients, a history of immunosuppression is unlikely to confer an adverse prognosis. These results require verification in subsequent cohorts, using standardized definitions of immunosuppression. Strengths of this study include the prospective cohort design of consecutively enrolled, newly diagnosed lymphoma patients; central pathology review; systematically collected clinical data; virtually complete follow-up of the cohort for disease progression and death; and medical record validation of these events. Our series of 362 systematically studied patients with biopsy material is one of the largest cohorts from the Western hemisphere published to date, and all patients were managed in the current immunochemotherapy era. Limitations include the modest and imperfectly representative availability of

continued from the previous page

Characteristic Age-adjusted IPI score 0 1 2 3 Bulky disease Missing Yes No GCB/non-GCB Missing Non-GCB GCB Composite No Yes CD30 Missing <20 20+ Immuno-chemotherapy No Yes Radiation therapy No Yes Follow-up, months Median Range

Total (n = 362)

EBV+ (n = 16)

EBV(n = 346)

103 (28%) 114 (32%) 105 (29%) 40 (11%)

2 (12%) 7 (44%) 5 (31%) 2 (12%)

101 (29%) 107 (31%) 100 (29%) 38 (11%)

4 (1%) 29 (8%) 329 (91%)

1 (6%) 0 (0%) 15 (94%)

3 (1%) 29 (8%) 314 (91%)

75 (21%) 128 (35%) 159 (44%)

5 (31%) 10 (62%) 1 (6%)

70 (20%) 118 (34%) 158 (46%)

296 (82%) 66 (18%)

14 (88%) 2 (12%)

282 (82%) 64 (18%)

89 (25%) 241 (67%) 32 (9%)

7 (44%) 5 (31%) 4 (25%)

82 (24%) 236 (68%) 28 (8%)

38 (10%) 324 (90%)

2 (12%) 14 (88%)

36 (10%) 310 (90%)

289 (80%) 73 (20%)

13 (81%) 3 (19%)

276 (80%) 70 (20%)

59 (1-155)

83 (1-143)

59 (1-155)

P-value

Immunosuppressed (n = 39)

Immunocompetent (n =323)

7 (18%) 15 (38%) 14 (36%) 3 (8%)

96 (30%) 99 (31%) 91 (28%) 37 (12%)

0 (0%) 1 (3%) 38 (97%)

4 (1%) 28 (9%) 291 (90%)

4 (10%) 19 (49%) 16 (41%)

71 (22%) 109 (34%) 143 (44%)

34 (87%) 5 (13%)

262 (81%) 61 (19%)

7 (18%) 27 (69%) 5 (13%)

82 (25%) 214 (66%) 27 (8%)

5 (13%) 34 (87%)

33 (10%) 290 (90%)

33 (85%) 6 (15%)

256 (79%) 67 (21%)

73 (3-143)

59 (1-155)5

0.50

P-value 0.32

0.07

0.32

0.007

0.10

0.54

0.35

0.005

0.45

0.79

0.62

0.88

0.43

0.14

0.1

EBV: Epstein-Barr virus; MW: Mid-Western; IPI: International Prognostic Index; GCB: germinal-center B cell; Bold values indicate statistical significance.

haematologica | 2018; 103(2)

301

S.I. Tracy et al. A

Figure 2. Association of Epstein-Barr virus tumor positivity with clinical endpoints. (A) Tumor EBV status and overall survival. (B) Tumor EBV status and event-free survival.

Figure 3. Association of immunosuppression with clinical endpoints. (A) Immune status and overall survival. (B) Immune status and event-free survival.

biopsy material sufficient for tissue microarray production (362 of 1,081), and the modest number of EBV+ patients which limits the power of any subset analyses. Although prospectively assembled, some analyses were retrospective in nature with all incumbent limitations. Finally, we were unable to quantify the degree of immunocompetence satisfactorily. Clearly a remote history of a brief course of immunosuppression is very different from ongoing aggressive immunosuppression, but given the subjectivity of comparing one immunosuppressing regimen to another and the complexity of factoring host issues, we chose a dichotomous “any” versus “none” variable. Over 85% of those identified as receiving immunosuppressive therapy were on such at the time of diagnosis. Attempts at quantifying immunosuppression could be the target of future research.

In summary, we found that EBV+ DLBCL represents a small fraction of DLBCL. Outcomes among EBV+ DLBCL patients are not significantly worse than those among their EBV– counterparts. The lack of difference in clinical outcomes between the studied subsets should suggest to practicing hematologists that prognosis is independent of EBV status or a history of immunosuppression, among North American patients.

References 1. Campo E, Swerdlow SH, Harris NL, Pileri S, Stein H, Jaffe ES. The 2008 WHO classification of lymphoid neoplasms and beyond: evolving concepts and practical applications. Blood. 2011;117(19):5019-5032. 2. Ahn JS, Yang DH, Duk Choi Y, et al. Clinical outcome of elderly patients with Epstein-Barr virus positive diffuse large B-

302

Acknowledgments We thank Julianne Lunde, Laura Jacobus, Ashley McCarthy and the study staff at Mayo Clinic Rochester and University of Iowa. This work was supported by grants from the National Institutes of Health (Lymphoma SPORE (P50 CA CA97274), Lymphoma Epidemiology of Outcomes (U01 CA195568) and the Predolin Foundation.

cell lymphoma treated with a combination of rituximab and CHOP chemotherapy. Am J Hematol. 2013;88(9):774-779. 3. Beltran BE, Castillo JJ, Morales D, et al. EBV-positive diffuse large B-cell lymphoma of the elderly: a case series from Peru. Am J Hematol. 2011;86(8):663-667. 4. Hong JY, Yoon DH, Suh C, et al. EBV-positive diffuse large B-cell lymphoma in young adults: is this a distinct disease entity? Ann Oncol. 2015;26(3):548-555.

5. Park S, Lee J, Ko YH, et al. The impact of Epstein-Barr virus status on clinical outcome in diffuse large B-cell lymphoma. Blood. 2007;110(3):972-978. 6. Sato A, Nakamura N, Kojima M, et al. Clinical outcome of Epstein-Barr virus-positive diffuse large B-cell lymphoma of the elderly in the rituximab era. Cancer Sci. 2014;105(9):1170-1175. 7. Wada N, Ikeda J, Hori Y, et al. Epstein-Barr virus in diffuse large B-cell lymphoma in

haematologica | 2018; 103(2)

DLBCL with EBV or immunosuppression

10.

11.

12.

13.

14.

15.

immunocompetent patients in Japan is as low as in Western countries. J Med Virol. 2011;83(2):317-321. Lu TX, Liang JH, Miao Y, et al. Epstein-Barr virus positive diffuse large B-cell lymphoma predict poor outcome, regardless of the age. Sci Rep. 2015;5:1216-1218. Cohen M, Narbaitz M, Metrebian F, De Matteo E, Preciado MV, Chabay PA. Epstein-Barr virus-positive diffuse large Bcell lymphoma association is not only restricted to elderly patients. Int J Cancer. 2014;135(12):2816-2824. Hoeller S, Tzankov A, Pileri SA, Went P, Dirnhofer S. Epstein-Barr virus-positive diffuse large B-cell lymphoma in elderly patients is rare in Western populations. Hum Pathol. 2010;41(3):352-357. Ok CY, Li L, Xu-Monette ZY, et al. Prevalence and clinical implications of epstein-barr virus infection in de novo diffuse large B-cell lymphoma in Western countries. Clin Can Res. 2014;20(9):23382349. Castillo JJ, Beltran BE, Miranda RN, Young KH, Chavez JC, Sotomayor EM. EBV-positive diffuse large B-cell lymphoma of the elderly: 2016 update on diagnosis, riskstratification, and management. Am J Hematol. 2016;91(5):529-537. Swerdlow SH, Campo E, Pileri SA, et al. The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood. 2016;127(20):2375-2390. Kersey JH, Shapiro RS, Filipovich AH. Relationship of immunodeficiency to lymphoid malignancy. Pediatr Infect Dis J. 1988;7(5 Suppl):S10-12. Grulich AE, Vajdic CM, Cozen W. Altered

haematologica | 2018; 103(2)

16.

17.

18.

19.

20.

21.

22.

immunity as a risk factor for non-Hodgkin lymphoma. Cancer Epidemiol Biomarkers Prev. 2007;16(3):405-408. Drake MT, Maurer MJ, Link BK, et al. Vitamin D insufficiency and prognosis in non-Hodgkin's lymphoma. J Clin Oncol. 2010;28(27):4191-4198. Cerhan JR, Link BK, Habermann TM, et al. Cohort Profile: The Lymphoma Specialized Program of Research Excellence (SPORE) Molecular Epidemiology Resource (MER) cohort study. Int J Epidemiol. 2017;46(6): 1753-1754. Hans CP, Weisenburger DD, Greiner TC, et al. Confirmation of the molecular classification of diffuse large B-cell lymphoma by immunohistochemistry using a tissue microarray. Blood. 2004;103(1):275-282. Hu S, Xu-Monette ZY, Balasubramanyam A, et al. CD30 expression defines a novel subgroup of diffuse large B-cell lymphoma with favorable prognosis and distinct gene expression signature: a report from the International DLBCL Rituximab-CHOP Consortium Program Study. Blood. 2013;121(14):2715-2724. Stuhlmann-Laeisz C, Szczepanowski M, Borchert A, Bruggemann M, Klapper W. Epstein-Barr virus-negative diffuse large Bcell lymphoma hosts intra- and peritumoral B-cells with activated Epstein-Barr virus. Virchows Arch. 2015;466(1):85-92. Dinopoulos A, Attilakos A, Paschalidou M, et al. Short-term effect of levetiracetam monotherapy on haematological parameters in children with epilepsy: a prospective study. Epilepsy Res. 2014;108(4):820-823. Guenther S, Bauer S, Hagge M, et al. Chronic valproate or levetiracetam treat-

23.

24.

25.

26.

27.

28.

29.

ment does not influence cytokine levels in humans. Seizure. 2014;23(8):666-669. Wei M, Li L, Meng R, et al. Suppressive effect of diazepam on IFN-gamma production by human T cells. Int Immunopharmacol. 2010;10(3):267-271. Maurer MJ, Ghesquieres H, Jais JP, et al. Event-free survival at 24 months is a robust end point for disease-related outcome in diffuse large B-cell lymphoma treated with immunochemotherapy. J Clin Oncol. 2014;32(10):1066-1073. Gibson SE, Hsi ED. Epstein-Barr virus-positive B-cell lymphoma of the elderly at a United States tertiary medical center: an uncommon aggressive lymphoma with a nongerminal center B-cell phenotype. Hum Pathol. 2009;40(5):653-661. Ok CY, Li L, Xu-Monette ZY, et al. Prevalence and clinical implications of Epstein-Barr virus infection in de novo diffuse large B-cell lymphoma in Western countries. Clin Cancer Res. 2014;20(9):2338-2349. Beltran BE, Castillo JJ, Morales D, et al. EBV-positive diffuse large B-cell lymphoma of the elderly: a case series from Peru. Am J Hematol. 2011;86(8):663-667. Sato A, Nakamura N, Kojima M, et al. Clinical outcome of Epstein-Barr virus-positive diffuse large B-cell lymphoma of the elderly in the rituximab era. Cancer Sci. 2014;105(9):1170-1175. Ahn JS, Yang DH, Duk Choi Y, et al. Clinical outcome of elderly patients with Epstein-Barr virus positive diffuse large Bcell lymphoma treated with a combination of rituximab and CHOP chemotherapy. Am J Hematol. 2013;88(9):774-779.

303

ARTICLE

Lymphoproliferative Disorders

Ferrata Storti Foundation

Somatic STAT3 mutations in Felty syndrome: an implication for a common pathogenesis with large granular lymphocyte leukemia Paula Savola,1,2 Oscar Brück,1,2 Thomas Olson,3 Tiina Kelkka,1,2 Markku J. Kauppi,4,5 Panu E. Kovanen,6 Soili Kytölä,7 Tuulikki Sokka-Isler,8 Thomas P. Loughran,3 Marjatta Leirisalo-Repo9 and Satu Mustjoki1,2

Haematologica 2018 Volume 103(2):304-312

Hematology Research Unit Helsinki, University of Helsinki and Department of Hematology, Helsinki University Hospital Comprehensive Cancer Center, Finland; Department of Clinical Chemistry and Hematology, University of Helsinki, Finland; 3 University of Virginia Cancer Center; University of Virginia, Charlottesville, VA, USA; 4 Päijät-Häme Central Hospital, Lahti, Finland; 5Faculty of Medicine, Tampere University, Finland; 6Department of Pathology, University of Helsinki and HUSLAB, Helsinki University Hospital, Finland; 7Laboratory of Genetics, HUSLAB, Helsinki University Hospital, Finland; 8Rheumatology/Medicine, Jyväskylä Central Hospital, Finland and 9 Rheumatology, University of Helsinki and Helsinki University Hospital, Finland 1 2

ABSTRACT

Correspondence: satu.mustjoki@helsinki.fi

Received: July 13, 2017. Accepted: December 6, 2017. Pre-published: December 7, 2017. doi:10.3324/haematol.2017.175729 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/304

elty syndrome is a rare disease defined by neutropenia, splenomegaly, and rheumatoid arthritis. Sometimes the differential diagnosis between Felty syndrome and large granular lymphocyte leukemia is problematic. Recently, somatic STAT3 and STAT5B mutations were discovered in 30-40% of patients with large granular lymphocyte leukemia. Herein, we aimed to study whether these mutations can also be detected in Felty syndrome, which would imply the existence of a common pathogenic mechanism between these two disease entities. We collected samples and clinical information from 14 Felty syndrome patients who were monitored at the rheumatology outpatient clinic for Felty syndrome. Somatic STAT3 mutations were discovered in 43% (6/14) of Felty syndrome patients with deep amplicon sequencing targeting all STAT3 exons. Mutations were located in the SH2 domain of STAT3, which is a known mutational hotspot. No STAT5B mutations were found. In blood smears, overrepresentation of large granular lymphocytes was observed, and in the majority of cases the CD8+ T-cell receptor repertoire was skewed when analyzed by flow cytometry. In bone marrow biopsies, an increased amount of phospho-STAT3 positive cells was discovered. Plasma cytokine profiling showed that ten of the 92 assayed cytokines were elevated both in Felty syndrome and large granular lymphocyte leukemia, and three of these cytokines were also increased in patients with uncomplicated rheumatoid arthritis. In conclusion, somatic STAT3 mutations and STAT3 activation are as frequent in Felty syndrome as they are in large granular lymphocyte leukemia. Considering that the symptoms and treatment modalities are also similar, a unified reclassification of these two syndromes is warranted. Introduction

©2018 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.

304

Felty syndrome (FS) is a disease defined by neutropenia, splenomegaly, and rheumatoid arthritis (RA). It affects less than 1% of RA patients.1 A classical FS patient has long-standing RA, unexplained neutropenia (causes such as antirheumatic drugs must be ruled out), and splenomegaly. Splenomegaly is not an absolute criterion for diagnosis.1 The disease has much in common with T-cell large granular lymphocyte (T-LGL) leukemia, an indolent chronic hematological malignancy in which patients have persistent monoclonal LGL lymphocytosis >0.5x109/l, which in the majority of cases consists of CD8+ T cells. Concomitant autoimmune manifestations such as neutropenia (70-80% of cases), RA (11-36%), and splenomegaly (20-60%) are also observed.2 Recent studies have unveiled the molecular pathogenesis of LGL leukemia; somatic STAT3 mutations in lymphocytes leading to constitutive STAT3 activation occur in 30-40% of LGL leukemia patients.3-6 Similar mutations have also been dishaematologica | 2018; 103(2)

Somatic STAT3 mutations in Felty syndrome

covered in the STAT5B gene in 2% of cases.7 STAT3 mutation status is closely linked to RA in LGL leukemia; patients with multiple STAT3 mutations have RA (43%) more frequently than patients without mutations (6%).6 Due to the similarity of clinical and laboratory findings, the differential diagnosis of LGL leukemia with RA and FS may sometimes be difficult. Past studies have shown a resemblance in patient phenotypes, responses to treatment, high HLA-DR4 prevalence, and analogous findings in splenectomy samples.2 Furthermore, both diseases are treated with immunosuppressive agents.2 Some authors have suggested that these two diseases represent the same entity,2,8,9 but this hypothesis currently lacks molecular evidence. Herein, we utilized deep next-generation sequencing (NGS) of STAT3 and STAT5B genes and cytokine profiling in order to examine if these two diseases are part of the same disease continuum.

Methods All methods are described in more detail in Online Supplementary Methods.

drome (n=20) and LGL leukemia patients (n=9) were used. The ethical boards of our institutions approved the study, and the Declaration of Helsinki guidelines were followed. Patients gave written informed consent.

Sample preparation DNA samples from different cell types were obtained (Online Supplementary Table S1 and Online Supplementary Figure S1). For six FS patients, only archived samples were available (Online Supplementary Table S1). Eight FS patients gave fresh peripheral blood samples. Mononuclear cells were extracted via Ficoll gradient centrifugation (Ficoll-Paque PLUS, GE Healthcare) from fresh blood samples and CD4+ and CD8+ cells with magnetic bead selection (Miltenyi Biotech). DNA was extracted according to manufacturer’s instructions using the NucleoSpin Tissue DNA extraction kit (Macherey-Nagel). T-cell clones were investigated with flow cytometry using a panel of antibodies against the different variable β regions of the T-cell receptor (IO Test Vβ Mark Kit, Beckman Coulter), accompanied by anti-CD3 (SK7), anti-CD4 (SK3), and anti-CD8 (SK-1) (Becton Dickinson) antibodies. The IO Test Vβ mark kit covers approximately 70% of the Vβ T-cell repertoire. Sorting of expanded CD8+ T-cell populations was performed using the same antibodies.

Patient recruitment The diagnosis of FS is defined by RA, neutropenia, and splenomegaly. We included patients with an established Felty syndrome diagnosis (n=14) stated in patient records. In addition, samples from healthy controls (n=8), RA patients without Felty syn-

STAT3 and STAT5B sequencing Amplicon sequencing of 23 exons of the STAT3 gene, STAT5B exon 16, and STAT5A exon 17 was performed on the DNA samples. Amplicon sequencing is a polymerase chain reaction (PCR)-

Figure 1. Felty syndrome patients harbor STAT3 mutations. (A) STAT3 mutations (presented as protein level amino acid changes) identified in study patients. Mutation variant-allele frequencies (VAFs) in the sample with the highest VAF are shown in the table. All samples and mutations detected are shown in Online Supplementary Table S1. (B) Schematic of sequencing results of all six patients with STAT3 mutations. Each patient is presented separately, and the presence of a mutation is indicated by -/+. Presented in smaller font are the data on the VAF and the sample type that was available for sequencing from the time-points. DNA sample types in A-B: Bone marrow (BM) cells, cultured cells from the BM (originally obtained for chromosome analysis); BM MNC: bone marrow mononuclear cells; PWB: peripheral whole blood; PB CD8+: peripheral blood CD8+ cells.

haematologica | 2018; 103(2)

305

P. Savola et al.

ery rate <0.1 (Benjamini-Hochberg method) were reported, and further post hoc testing was performed using Dunn’s multiple comparison tests. Statistical analyses for the clinical data were performed with GraphPad Prism software. Statistical tests included the Kruskal-Wallis omnibus test, Dunn’s multiple comparison test, Mann-Whitney test and Fisher’s exact test.

based NGS method that enables very high sequencing depths (up to over 100,000x). It detects variants with low variant-allele frequencies (VAF), locally reaching a sensitivity of 0.5%.6 Sequencing was performed using the Illumina MiSeq System, and the data analysis pipeline has been described in more detail previously.6 Herein, we considered the mutation to be true if it occurred in the sequenced sample with a VAF >1% and absent in control samples. All primers are listed in Online Supplementary Table S2.

Results

Cytokine profiling

Patient characteristics

Ethylenediaminetetraacetic acid (EDTA) plasma samples from eight healthy controls, nine RA patients, seven FS patients, and nine LGL leukemia patients were analyzed with the Proseek Multiplex Inflammation I (Olink Biosciences) immunoassay which detects 92 different biomarkers simultaneously in one sample via oligonucleotide-labeled antibodies. The target-antibody complexes are detected with real-time PCR. A list of the analyzed biomarkers is presented in Online Supplementary Table S3. The assay does not report protein concentrations, and the results are reported as normalized protein expression units (NPX).

The clinical characteristics of the 14 FS patients included herein are presented in Table 1 and Table 2, and Online Supplementary Table S4 and Online Supplementary Table S5. None of the patients represented classical LGL leukemia with marked LGL lymphocytosis and T-cell receptor (TCR) γ gene rearrangement. Thirteen patients had longstanding RA prior to Felty syndrome diagnosis, and they had been treated with immunosuppressive drugs at the time of sample collection. The first sample of patient 9 was taken before initiating immunosuppressive treatment. As LGL lymphocytosis is the hallmark of LGL leukemia, peripheral blood smears were re-examined for this study. FS patients showed an overrepresentation of LGL lymphocytes (Table 2; normal range up to 25% of lymphocytes), but only two patients (patients 3 and 6) had absolute LGL lymphocytosis. Patients’ bone marrow exams did not show dysplasia, but they often showed a slight left shift in granulopoiesis, and in some cases, overrepresentation of lymphocytes. Large CD8+ T-cell expansions are characteristic for

Immunohistochemistry To study phosphorylation of STAT3, we stained bone marrow biopsy samples from seven FS patients with anti-phospho-STAT3 (Tyr705) and anti-CD57 antibodies. We also used 14 control bone marrow samples, which had been taken to examine abnormal blood counts, but no specific diagnosis had been achieved.

Statistics Multiplex cytokine data was analyzed as NPX values with Qlucore Omics Explorer (Qlucore). Cytokines with a false discov-

Table 1. Basic clinical characteristics of Felty syndrome patients.

Patient ID

Sex

Age at sample collection

RA duration

1 2 3 4 5

M F F F F

76 49 71 63 72

15 17 10 10 23

6 5 0 0 2

3 0 1 0 5

pos pos pos pos pos

no yes yes yes yes

no no no no (1 s) no

yes 15cm yes yes 16cm yes 13cm no 11cm

6 7 8 9 10

F F M F F

71 68 71 65 64

24 54 21 0 10

0 22 21 0 0

0 1 0 1 5

pos neg* pos pos pos

no yes yes no yes

yes (1s, 1t) yes (1s, 1t) no yes (>5s, >2t) Yes >3s

no 10cm yes 17cm yes 13cm yes 17cm yes

pos unknown pos pos 92% pos, 8% neg

yes no yes yes 71% yes, 29% no

no no no yes 36% yes, 64% no

no no yes 14cm no 64% yes, 36% no

11 12 13 14 Summary

Felty Infections Serostatus Erosions Active arthritis Splenomegaly dg at sample collection

F 57 24 0 1 M 40 3 0 0 F 70 42 7 10 F 38 5 0 0 79% F, 66.5 16.0 0 1 21% M (55.0-71.0) (8.8-24.0) (0.0-6.3) (0.0-3.5)

Other extra-articular symptoms rheumatoid nodules no weight loss rheumatoid nodules weight loss, rheumatoid nodules no leg ulcer no fever, chills, cough weight loss, maculopapular rash fever persistent rash leg ulcer, weight loss no

Clinical characteristics of the patient cohort. Summaries of the parameters are shown on the bottom row. Age, RA duration, Felty diagnosis (dg), and infections are summarized as medians with interquartile ranges in parenthesis. The patients’ ages are shown as ages at the time of the first sample collection. RA duration is shown as years of RA prior to collecting the first sample for the study. Felty dg is shown as years, starting from FS diagnosis to the collection of the first sample. The number of infections requiring hospitalization is shown. Serostatus was defined as either elevated rheumatoid factor and/or anti-citrullinated protein antibodies (ACPAs). Erosions were defined via hand and feet x-rays. The number of swollen (s) and tender (t) joints at sample collection is shown in parenthesis. Spleen size was measured with ultrasonography. *Rheumatoid factor normal, however, ACPAs were not determined. ID: identity; ND, not determined; RA: rheumatoid arthritis.

306

haematologica | 2018; 103(2)

Somatic STAT3 mutations in Felty syndrome

T-LGL leukemia.4 In FS patients, flow cytometry screening revealed skewed CD8+ TCR Vβ usage (Table 2 and Online Supplementary Figure S2). In 40% (4/10) of the examined cases, CD8+ expansions comprised over 20% of all CD8+ T cells. Most of our patients’ clinical features were well in accordance with the FS criteria (Table 1 and Table 2). Only three patients did not show the classical phenotype. No neutropenia was recorded in the clinical registry of patient 8. Patient 9 had cytopenias during tuberculosis therapy (although neutropenia continued years after the completion of the therapy). In addition, patient 3 showed lymphocytosis, but her bone marrow and peripheral blood TCRγ rearrangement tests were negative, and flow cytometry did not show any excess of natural killer (NK) cells.

Somatic STAT3 mutations were identified in 43% of patients For many patients, several sample types and samples from multiple time-points were available for sequencing (Online Supplementary Table S1). STAT3 mutations occurred in 43% (6/14) of FS patients (Figure 1A,B). One patient had two different STAT3 mutations (N647I and Y640F). In all assessable cases, STAT3 mutations occurred in CD8+ cells but not in CD4+ cells. In 4/6 STAT3-mutated cases, the mutations were detected in multiple sample types, whereas in 2/6 patients the mutation was detected only in one sample (Figure 1B and Online Supplementary Table S1). Five STAT3-mutated patients had follow-up samples available,

and in three cases the mutations were also detected during the follow-up (Figure 1B). None of the patients had STAT5B mutations. When patients with STAT3 mutations were compared with patients with no STAT3 mutations, no differences emerged in age, sex, LGL counts, CD8+ Vβ clone size, neutropenia duration (adjusted for laboratory follow-up time), lymphopenia duration (adjusted for laboratory follow-up time), highest and lowest lymphocyte counts, duration of RA, or the number of infections (compared with MannWhitney test). No statistically significant differences occurred in TCRγ rearrangement status, existence of erosions, splenomegaly, or active arthritis status (compared with Fisher’s test). Sorting of expanded CD8+ T cells with flow cytometry using Vβ antibodies showed that patient 6 harbored the Y640F mutation in the major expanded CD8+ T-cell population (Vβ3) (Figure 2), but no STAT3 mutations were found in CD4+ cells or other CD8+ cells lacking Vβ3. A similar analysis was also performed with the sample of patient 4, but the mutation was not located in the largest CD8+ T-cell expansion (Vβ22) which was detectable by flow cytometry (Figure 2). The sum of all of the detected CD8+ T-cell expansions made up only 38% of all CD8+ T cells, which is clearly less than the anticipated level of 70% of the TCR repertoire (Online Supplementary Figure S3). This suggests that the sensitivity of flow cytometry is insufficient to detect all clones. An 'undetected' clone is likely to harbor the identified mutation, since the VAF of

Table 2. Laboratory findings of Felty syndrome patients.

Patient ID

LGL count

1 2 3 4 5 6 7 8 9 10 11 12 13 14 Summary

33%/0.20 ND 62%/0.86 30%/0.42 28%/0.34 69%/0.57 ND ND 28%/0.29 ND ND ND 36%/0.26 ND 33%/0.34 (28-62)/ (0.26-0.57)

TCRγ STAT3 CD8+ Lab Neutropenia Lowest Lymphopenia Lowest Highest Thrombocytopenia assay mutation expansion follow-up duration neutrophil duration lymphocyte lymphocyte duration positive? (% of CD8+) count count count no no no yes: PB no yes: BM ND ND no yes equivocal no ND ND 33% yes 67% no

yes ND 163 94 yes 7.6 210 73 yes 13.9 156 14 yes 10.6 115 3 yes ND 52 18 yes 21.7 189 103 no ND 207 32 no 25.5 180 ND no ND 144 33 no 52.4 4 4 no 17.2 11 11 no 6.3 5 3 no 27.8 119 33 no 8.8 62 16 43% yes, 15.6 131.5 18 57% no (8.5-26.1) (41.8-182.3) (7.5-53)

0.17 0.4 0 <0.05 <0.05 0.66 1.22 1.58 0.08 1.14 1 1.1 0 0.6 0.5 (0-1.1)

135 191 0.07 4 38 63 194 173 61 4 2 3 88 57 59 (3.8-144.5)

0.28 0.5 1.37 1.03 0.41 0.82 0.34 0.7 0.79 0.75 0.9 0.8 0.4 0.4 0.73 (0.4-0.84)

1.52 1.74 4.96 2.93 0.46 2.42 0.64 1.33 1.82 1.25 1.6 1.5 1.82 1.3 1.6 (1.3-2.0)

18 0 0.2* 2.8 2 0 204 394** 0 4 11 2 0 0 2 (0-9.3)

The table shows the laboratory findings of Felty syndrome patients. The bottom row shows summary statistics for all patients (when applicable): median and interquartile range in parenthesis. All blood cell counts are shown as 109/l. The patients’ peripheral blood (PB) smears were, where possible, re-examined for this study, and LGL cells were counted as percentage of lymphocytes (counting 300 lymphocytes). Absolute LGL counts (109/l) were calculated from lymphocyte counts. The sizes of CD8+ T-cell expansions were studied via flow cytometry. Laboratory follow-up time in months (Lab follow-up) was calculated. The cumulative durations of neutropenia (<1.5x109/l), lymphopenia (<1.3x109/l), and thrombocytopenia (<100x109/l) were calculated as months from laboratory records. Of note, patients 10 and 12 have laboratory follow-up only for the period before hematology referral, but oral history indicates a history of years (patient 10) and decades (patient 12) of leukopenia. *HIT: heparin-induced thrombopenia, not included in summary statistic calculations. **Thrombocytes 100-150, not included in summary statistic calculations. ND: not determined; LGL: large granular lymphocyte; TCRγ: T-cell receptor γ; BM: bone marrow.

haematologica | 2018; 103(2)

307

P. Savola et al.

the mutation was high in the CD8+ Vβ22-negative fraction (18.8%). Previously, we have studied the presence of STAT3 mutations in 82 newly diagnosed RA patients, and no mutations were detected.10 Herein, we sequenced the STAT3 hotspot exon 21 from RA patients who had been treated for several years with no evidence of FS (n=14). As in newly diagnosed RA, no STAT3 mutations were detected in this patient cohort. The patients are described in more detail in Online Supplementary Table S6.

FS patients have increased phosphorylation of STAT3 in bone marrow samples We had access to seven bone marrow biopsy samples from FS patients 1-6 and 9. Patients 1-6 harbored somatic STAT3 mutations (Figure 1). Immunohistochemical staining of the samples showed increased amounts of phospho-STAT3 positive cells in FS patients when compared to the controls (Figure 3). The degree of phosphorylation was not associated with mutation VAFs: patient 3 did not have mutations in bone marrow samples, nonetheless the

Figure 2. Sequencing of sorted, expanded CD8+ T-cell populations. The expanded CD8+ T-cell populations that were detected by the initial Vβ flow cytometry screen were sorted via flow cytometry and sequenced with Amplicon sequencing (n=2; patients 4 and 6). (A) Gating strategy for Vβ analysis. The subpopulation of interest was included in subsequent gates and other populations were excluded. (B) The expanded CD8+ clones in patients 6 and 4. The expanded clones expressing the Vβ in question, as well as all other CD8+ cells were collected and sequenced. (C) The summarized sequencing results of sorted cells. The small number of mutationpositive cells present in the CD8+Vβ 3neg cells of patient 6 are likely due to sorting impurities. APC: allophycocyanin; FSC: forward scatter; FITC: fluorescein isothiocyanate; VAF: variant-allele frequencies.

308

haematologica | 2018; 103(2)

Somatic STAT3 mutations in Felty syndrome

bone marrow biopsy showed marked STAT3 phosphorylation. Additionally, patient 9 showed STAT3 phosphorylation in the bone marrow despite the absence of STAT3 mutations. Overall, CD57+ cells were less abundant than phospho-STAT3+ cells (Figure 3).

Felty syndrome patients shared similar plasma cytokine profiles with LGL leukemia and RA patients Previous reports have shown that LGL leukemia patients have aberrant levels of several cytokines in their plasma.11-17 Analysis via a multiplex cytokine panel

showed significant differences between healthy controls and patients with immune-mediated disease (RA, Felty syndrome, LGL leukemia). Only minor differences emerged between individual disease conditions (Figure 4). Many plasma cytokines (10/92; colony stimulating factor 1 (CSF1), C-X-C motif chemokine 10 (CXCL10), interleukin (IL)-15RA, macrophage inflammatory protein 1-Îą (MIP-1-Îą), oncostatin-M (OSM), tumor necrosis factor receptor superfamily member 9 (TNFRSF9), programmed cell death 1 ligand 1 (PD-L1), CUB domain-containing protein 1 (CDCP1), IL-6, and hepatocyte growth factor

Figure 3. Felty syndrome patients show STAT3 phosphorylation in bone marrow. Representative immunohistochemical analysis of phospho-Tyr705-STAT3 (red) and CD57 (green) in bone marrow tissue samples of Felty syndrome (n=7) and control samples (n=14) counterstained with hematoxylin. All images presented in the figure are 40x magnified unless otherwise specified. Felty syndrome patients show more phosphorylated STAT3 than controls, but the amount of phosphorylation was not related to the STAT3 mutation VAFs. Patient numbers and mutation status are presented in the figure.

haematologica | 2018; 103(2)

309

P. Savola et al.

(HGF)) were elevated in both FS and LGL leukemia (Online Supplementary Figure S4). CDCP1, IL-6, and HGF were elevated in all three diseases. RA patients shared some characteristics with FS and LGL leukemia patients, but seven cytokines were statistically significantly elevated only in RA patients and not in other patients (Online Supplementary Figure S4). Clinical information on the LGL leukemia patients, RA patients, and healthy controls used in this assay is shown in Online Supplementary Table S7 and Online Supplementary Table S8. When STAT3-mutated patients were compared with patients without STAT3 mutations, only CXCL1 was statistically significantly elevated in plasma samples of STAT3-mutated patients (Online Supplementary Figure S5). However, there was no significant difference when compared to healthy controls.

Discussion Previous studies have shown that Felty syndrome and LGL leukemia share many clinical features.2 The study herein confirms that in addition to similar clinical aspects,

these two disease entities also share analogous pathogenetic molecular markers. In a cohort of 14 FS patients, STAT3 mutations were discovered in 43% of the cases. This is comparable with the STAT3 mutation frequency in LGL leukemia.4-6 The diagnostic criteria for FS are not specific. Therefore, our study patients are a heterogeneous group, which may be subject to criticism. However, these patients represent the clinical spectrum of the FS patients who are monitored at rheumatology outpatient clinics. Further, most of our patients’ clinical features were in accordance with the FS diagnostic criteria. All detected mutations occurred in the hotspot exon 21 in the SH2 domain of STAT3. Likewise, as in LGL leukemia, Y640F and D661 mutations were the most common.4-6 Further, STAT3 mutations occurred in CD8+ cells and not in CD4+ cells in the cases in which this could be assessed. Interestingly, we also observed increased STAT3 activation in the bone marrow samples of FS patients. The presence of phospho-STAT3+ cells was not related to STAT3 mutation status. Similarly, it has been shown in LGL leukemia that STAT3 is activated in the majority of patients,5,18,19 although STAT3 mutations only occur in 30-40% of patients.4-6 In

Figure 4. Felty syndrome patients have a similar cytokine profile as that of LGL leukemia. The heatmap compares the plasma cytokine profiles of healthy controls, RA patients, Felty syndrome patients, and LGL leukemia patients. Data was analyzed as NPX units, and were normalized to allow for coloring on the same scale for the heatmap. Bright blue represents smaller protein concentration, while bright yellow represents higher protein concentration. The heatmap shows all cytokines for which the Q-value (false discovery rate) was less than 0.1. RA: rheumatoid arthritis; LGL: large granular lymphocyte.

310

haematologica | 2018; 103(2)

Somatic STAT3 mutations in Felty syndrome

addition to LGL leukemia, somatic STAT3 mutations have been described in other disease conditions, such as aplastic anemia, myelodysplastic syndrome, T-cell lymphomas, and inflammatory hepatocellular adenoma.20 In addition, one case report has shown STAT3 mutations in Felty syndrome.21 Of note, we did not detect any STAT3 mutations in patients with chronic RA who had no evidence of FS. This is concordant with our previous findings showing that in a cohort of 82 newly diagnosed RA patients, no STAT3 mutations were detected.10 LGL lymphocytosis is a hallmark of LGL leukemia. Peripheral blood smear re-examinations revealed overrepresentation of LGL cells in FS patients. Despite the high percentages of LGL cells, only two of our patients had LGL lymphocytosis (>0.5*109/l; patients 3 and 6). Absolute lymphocyte counts were low, excepting one patient. A previous study, which defined LGL cells with flow cytometry, reported increased percentages of LGL cells in FS.22 This is concordant with our data. Our patients’ skewed TCR Vβ usage also indicates that large T-cell clones often exist in FS, but the expansions were smaller when compared with results obtained with the same antibody panel in T-LGL leukemia.4 More detailed clonality analysis would require TCR deep sequencing. The smaller T-cell clone sizes in FS patients when compared to LGL leukemia patients could also explain why the STAT3 mutation VAFs were generally lower in FS patients. The VAF is highly dependent on the proportion of cells harboring mutations in the sequenced sample, and smaller clone sizes result in smaller detected VAFs. We had CD8+ cells available from 4/6 patients with STAT3 mutations, and even these cell fractions had low VAFs (1-1.4%), barring one case (8.8%). However, some LGL leukemia patients also harbor clones with small (0.8%) VAFs.6 Thus, our results do not differ from LGL leukemia cases with oligoclonal TCR expansions. Therefore, we suggest that LGL leukemia and FS form a disease continuum, which could explain the differences in VAFs and clone sizes. Many plasma cytokines (10/92) were elevated both in FS and LGL leukemia when compared to healthy controls (such as IL-15RA, CXCL10 and PD-L1). Importantly, only three of them were also elevated in RA patients (CDCP1, IL-6, and HGF). Furthermore, our results confirm the previous findings of increased IL-15RA, IL-8, C-C motif chemokine ligand 4 (CCL4=MIP1-β), and CXCL10 in LGL leukemia.11,12,14 Of these cytokines, IL-15RA and CXCL10 were also elevated in FS. Although some differences emerged between the diseases, we discovered that the three immune-mediated diseases are not strikingly different in terms of cytokine profiles. This is not unexpected; many LGL leukemia patients have RA, comparative to the other patient groups. In this small study, patients with STAT3 mutations did not differ from patients without STAT3 mutations in terms of clinical or biochemical characteristics. In LGL leukemia, neutropenia and RA are more common in patients with

References 1. Balint GP, Balint PV. Felty's syndrome. Best Pract Res Clin Rheumatol. 2004;18(5):631645.

haematologica | 2018; 103(2)

mutations.4,5 Similarly, germline gain-of-function STAT3 mutations lead to early-onset multiorgan autoimmunity, including cytopenias.23-25 However, the exact mechanism of neutropenia in LGL leukemia and FS is not clear. Based on the current hypothesis, STAT3 mutations lead to clonal outgrowth.26 Depending on the antigen target, the lymphocyte clones can attack bone marrow and joints. In addition, disease manifestations may also result from the production of proinflammatory cytokines mediated by hyperactive STAT3 signaling.27 Thus, the activation of STAT3 is a likely causative factor for autoimmunity, but future research needs to address the exact mechanisms in more detail. Of note, four of the FS patients had celiac disease, despite the fact that RA (without FS) does not have an established connection with celiac disease. Although the observed overrepresentation may occur by chance, celiac disease could also be a result of chronic systemic immune dysregulation. This is strengthened by recent findings showing that activating STAT3 mutations can be detected in refractory celiac disease patients’ duodenal biopsyderived intraepithelial lymphocyte cell lines.28 The constitutive STAT3 activation of immune cells may contribute to the persistent autoimmune inflammation that occurs in these patients. A similar mechanism is possible in FS patients, but this issue requires further investigation. In conclusion, our results supplement the evidence that FS and LGL leukemia belong to the same disease continuum and diagnostic group. These diseases share similar molecular pathogenetic features and phenotypes, and are treated similarly with immunosuppressive agents. Thus, unified re-classification of these two diseases should be considered. Acknowledgments The authors would like to thank the personnel at the Hematology Research Unit Helsinki for their expert technical assistance. Amplicon sequencing and PCR were performed by the sequencing unit at the Technology Centre of Institute of Molecular Medicine Finland (FIMM), and Pekka Ellonen and Sonja Lagström are acknowledged for their expertise and assistance with the sequencing. Tom Pettersson (MD, PhD) is acknowledged for his valuable scientific insights regarding the project. Biomedicum Functional Genomics Unit (FuGU) is acknowledged for performing the Olink Inflammation I panel assay. Funding This work was supported by the European Research Council (M-IMM project), Academy of Finland, Finnish special governmental subsidy for health sciences, research and training, the Sigrid Juselius Foundation, the Instrumentarium Science foundation, the Finnish Cancer Societies, Biomedicum Helsinki Foundation and the Finnish Cancer Institute. TO and TL were supported by R01CA178393 awarded to TL, and TO was supported for part of this period by T32CA009109 (National Cancer Institute of the National Institutes of Health).

2. Liu X, Loughran TP, Jr. The spectrum of large granular lymphocyte leukemia and Felty's syndrome. Curr Opin Hematol. 2011;18(4):254-259. 3. Andersson E, Kuusanmaki H, Bortoluzzi S, et al. Activating somatic mutations outside

the SH2-domain of STAT3 in LGL leukemia. Leukemia. 2016;30(5):1204-1208. 4. Jerez A, Clemente MJ, Makishima H, et al. STAT3 mutations unify the pathogenesis of chronic lymphoproliferative disorders of NK cells and T-cell large granular lympho-

311

P. Savola et al.

9. 10.

11.

12.

312

cyte leukemia. Blood. 2012;120(15):30483057. 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. Rajala HL, Olson T, Clemente MJ, et al. The analysis of clonal diversity and therapy responses using STAT3 mutations as a molecular marker in large granular lymphocytic leukemia. Haematologica. 2015; 100(1):91-99. Rajala HL, Eldfors S, Kuusanmaki H, et al. Discovery of somatic STAT5b mutations in large granular lymphocytic leukemia. Blood. 2013;121(22):4541-4550. Bowman SJ, Sivakumaran M, Snowden N, et al. The large granular lymphocyte syndrome with rheumatoid arthritis. Immunogenetic evidence for a broader definition of Felty's syndrome. Arthritis Rheum. 1994;37(9):1326-1330. Starkebaum G. Leukemia of large granular lymphocytes and rheumatoid arthritis. Am J Med. 2000;108(9):744-745. 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. Chen J, Petrus M, Bamford R, et al. Increased serum soluble IL-15Ralpha levels in T-cell large granular lymphocyte leukemia. Blood. 2012;119(1):137-143. Kothapalli R, Nyland SB, Kusmartseva I, Bailey RD, McKeown TM, Loughran TP, Jr. Constitutive production of proinflammatory cytokines RANTES, MIP-1beta and IL-18 characterizes LGL leukemia. Int J Oncol.

2005;26(2):529-535. 13. Liu JH, Wei S, Lamy T, et al. Blockade of Fas-dependent apoptosis by soluble Fas in LGL leukemia. Blood. 2002;100(4):14491453. 14. Momose K, Makishima H, Ito T, et al. Close resemblance between chemokine receptor expression profiles of lymphoproliferative disease of granular lymphocytes and their normal counterparts in association with elevated serum concentrations of IP-10 and MIG. Int J Hematol. 2007; 86(2):174-179. 15. Saitoh T, Matsushima T, Kaneko Y, et al. T cell large granular lymphocyte (LGL) leukemia associated with Behcet's disease: high expression of sFasL and IL-18 of CD8 LGL. Ann Hematol. 2008;87(7):585-586. 16. Tanaka M, Suda T, Haze K, et al. Fas ligand in human serum. Nat Med. 1996;2(3):317322. 17. Zhang R, Shah MV, Yang J, et al. Network model of survival signaling in large granular lymphocyte leukemia. Proc Natl Acad Sci USA. 2008;105(42):16308-16313. 18. Andersson EI, Rajala HL, Eldfors S, et al. Novel somatic mutations in large granular lymphocytic leukemia affecting the STATpathway and T-cell activation. Blood Cancer J. 2013;3:e168. 19. Epling-Burnette PK, Liu JH, Catlett-Falcone R, et al. Inhibition of STAT3 signaling leads to apoptosis of leukemic large granular lymphocytes and decreased Mcl-1 expression. J Clin Invest. 2001;107(3):351-362. 20. Rajala HL, Porkka K, Maciejewski JP, Loughran TP, Jr., Mustjoki S. Uncovering the pathogenesis of large granular lymphocytic leukemia-novel STAT3 and STAT5b

mutations. Ann Med. 2014;46(3):114-122. 21. Schrenk KG, Krokowski M, Feller AC, et al. Clonal T-LGL population mimicking leukemia in Felty's syndrome--part of a continuous spectrum of T-LGL proliferations? Ann Hematol. 2013;92(7):985-987. 22. Bowman SJ, Bhavnani M, Geddes GC, et al. Large granular lymphocyte expansions in patients with Felty's syndrome: analysis using anti-T cell receptor V beta-specific monoclonal antibodies. Clin Exp Immunol. 1995;101(1):18-24. 23. 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. 24. Haapaniemi EM, Kaustio M, Rajala HL, et al. Autoimmunity, hypogammaglobulinemia, lymphoproliferation, and mycobacterial disease in patients with activating mutations in STAT3. Blood. 2015;125(4):639-648. 25. Milner JD, Vogel TP, Forbes L, et al. Earlyonset lymphoproliferation and autoimmunity caused by germline STAT3 gain-offunction mutations. Blood. 2015; 125(4):591-599. 26. Lamy T, Moignet A, Loughran TP, Jr. LGL leukemia: from pathogenesis to treatment. Blood. 2017;129(9):1082-1094. 27. Burks EJ, Loughran TP, Jr. Pathogenesis of neutropenia in large granular lymphocyte leukemia and Felty syndrome. Blood Rev. 2006;20(5):245-266. 28. Ettersperger J, Montcuquet N, Malamut G, et al. Interleukin 15-dependent T-cell-like innate intraepithelial lymphocytes develop in the intestine and transform into lymphomas in celiac disease. Immunity. 2016; 45(3):610-625.

haematologica | 2018; 103(2)

ARTICLE

Chronic Lymphocytic Leukemia

Expression of COBLL1 encoding novel ROR1 binding partner is robust predictor of survival in chronic lymphocytic leukemia

Ferrata Storti Foundation

Hana Plešingerová,1,2 Pavlína Janovská,3,4 Archana Mishra,3 Lucie Smyčková,3 Lucie Poppová,1,2 Antonín Libra,5 Karla Plevová,1,2 Petra Ovesná,6 Lenka Radová,2 Michael Doubek,1,2 Šárka Pavlová,1,2 Šárka Pospíšilová,1,2 and Vítězslav Bryja3,4

Center of Molecular Biology and Gene Therapy, Department of Internal Medicine– Hematology and Oncology, University Hospital Brno and Medical Faculty, Masaryk University, Brno; 2Central European Institute of Technology, Masaryk University, Brno; 3 Institute of Experimental Biology, Faculty of Science, Masaryk University, Brno; 4 Department of Cytokinetics, Institute of Biophysics, Academy of Sciences of the Czech Republic, Brno; 5Generi Biotech, s.r.o., Hradec Králové and 6Institute of Biostatistics and Analyses, Masaryk University, Brno, Czech Republic 1

Haematologica 2018 Volume 103(2):313-324

ABSTRACT

hronic lymphocytic leukemia is a disease with up-regulated expression of the transmembrane tyrosine-protein kinase ROR1, a member of the Wnt/planar cell polarity pathway. In this study, we identified COBLL1 as a novel interaction partner of ROR1. COBLL1 shows clear bimodal expression with high levels in chronic lymphocytic leukemia patients with mutated IGHV and approximately 30% of chronic lymphocytic leukemia patients with unmutated IGHV. In the remaining 70% of chronic lymphocytic leukemia patients with unmutated IGHV, COBLL1 expression is low. Importantly, chronic lymphocytic leukemia patients with unmutated IGHV and high COBLL1 have an unfavorable disease course with short overall survival and time to second treatment. COBLL1 serves as an independent molecular marker for overall survival in chronic lymphocytic leukemia patients with unmutated IGHV. In addition, chronic lymphocytic leukemia patients with unmutated IGHV and high COBLL1 show impaired motility and chemotaxis towards CCL19 and CXCL12 as well as enhanced B-cell receptor signaling pathway activation demonstrated by increased PLCγ2 and SYK phosphorylation after IgM stimulation. COBLL1 expression also changes during B-cell maturation in non-malignant secondary lymphoid tissue with a higher expression in germinal center B cells than naïve and memory B cells. Our data thus suggest COBLL1 involvement not only in chronic lymphocytic leukemia but also in B-cell development. In summary, we show that expression of COBLL1, encoding novel ROR1-binding partner, defines chronic lymphocytic leukemia subgroups with a distinct response to microenvironmental stimuli, and independently predicts survival of chronic lymphocytic leukemia with unmutated IGHV.

Correspondence: bryja@sci.muni.cz

Received: August 18, 2017. Accepted: November 3, 2017. Pre-published: November 9, 2017. doi:10.3324/haematol.2017.178699 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/313

Introduction Upregulation of transmembrane receptor tyrosine kinase-like orphan receptor 1 (ROR1) in chronic lymphocytic leukemia (CLL) cells was revealed as one of the most stable CLL markers.1,2 ROR1 is expressed on the cell surface of patients with mutated (M-CLL) as well as unmutated (U-CLL) IGHV. ROR1 is highly expressed during embryonal development but largely undetectable in the adult organism.3,4 Negligible ROR1 expression on healthy B cells4,5 makes it a suitable candidate for monitoring CLL remission6 and a candidate target for therapy with monoclonal antibodies7 or T cells with ROR1-specific chimeric antigen receptor.8,9 Although ROR1 is up-regulated in CLL patients, its activity may vary depending on its posttranslational modification10 and on the availability of its dedicated ligands.11 ROR1 is a member of the Wnt/PCP (planar cell polarity) signaling pathway,4 which regulates various processes during embryonic development, mainly linked to cell polarity, survival and migration. We have previously reported, in accordance haematologica | 2018; 103(2)

313

H. Plešingerová et al.

with others, that Wnt/PCP components are expressed differently in CLL subgroups defined by IGHV mutational status and CLL aggressiveness.11-13 It has also been well described that deregulated ROR1 and Wnt/PCP pathway affect CLL cell migration, survival and chemotaxis.11,13,14 Despite the generally accepted importance of ROR1/PCP signaling in CLL, surprisingly little is known about downstream effectors and links to other signaling pathways critical for CLL pathogenesis. In this study, we focused on the analysis of ROR1 downstream signaling in CLL. We took advantage of the proteomic approach and analyzed the protein composition of endogenous ROR1 complexes from primary CLL cells. This unbiased approach allowed us to identify a poorly known protein, Cordon-blue protein-like 1 (COBLL1), as a novel ROR1 binding partner. Examining COBLL1 expression in CLL cells showed that COBLL1 expression can serve as an independent molecular marker in U-CLL: U-CLL COBLL1-high patients had a deregulated response to microenvironmental stimuli and significantly worse prognosis, resulting in shorter overall survival (OS) and time to second treatment (TTST). These data further pinpoint the importance of the ROR1/PCP signaling axis in CLL and identify COBLL1 as an important and clinically relevant regulator of this process.

RNA was extracted with TriReagent (Molecular Research Center). For information on how IGHV mutation status15 and genetic aberrations16-20 were determined and HEK293 and MAVER-1 cells cultured, see the Online Supplementary Appendix.

Mass spectrometry, transfection, immunoprecipitation, immunofluorescence and western blotting To identify and confirm potential ROR1 binding partners, immunoprecipitation of ROR1 from primary CLL cells coupled to mass-spectrometry,21,22 transfection of HEK293 cells,10,23 immunoprecipitation of MAVER-1 and transfected HEK293 cells, 11 immunofluorescence of transfected HEK293 cells24 and western blotting25 were performed as previously described. For details, see the Online Supplementary Appendix.

Gene expression analysis COBLL1 and ROR1 mRNA expression was assessed using qRTPCR. Three COBLL1 expression datasets were obtained; for details see the Online Supplementary Appendix. Since all datasets showed a similar bimodal distribution (Online Supplementary Figure S1), dCt values (dCt = CtCOBLL1 – Ctmean of reference genes) were normalized using the mean expression and standard deviation of the U-CLL samples and subsequently merged into one dataset. ROR1 mRNA expression was examined as previously described.13 The expression was further calculated from dCt values (ROR1) and normalized dCt values (dCtN, COBLL1) by the 2-dCt x100% and 2dCtN x100% method, respectively.

Methods Transwell assay Patients and samples All samples were taken after informed consent in accordance with the Declaration of Helsinki, under protocols approved by the Ethical Committee of the University Hospital Brno, Czech Republic. Peripheral blood (PB) B cells from CLL patients or healthy volunteers and non-malignant tonsillar tissue were separated by nonB-cell depletion (RosetteSep CD3+ Cell Depletion Cocktail, RosetteSep B Cell Enrichment Cocktail, StemCell Technologies or magnetic B-cell isolation kit II, Miltenyi Biotec). Isolated B-cell purity was assessed by flow cytometry and exceeded 98%. Tonsillar B cells were stained and sorted as described previously.11

Cell migration in RPMI supplemented with 1% FBS and antibiotics towards chemokines CCL19 or CXCL12 (200 ng/mL; 350NS-010, 361-MI-025, R&D Systems) or chemokine-free media was analyzed as described previously.13 Migrated cells were counted using Accuri C6 flow cytometer (BD Biosciences).

BCR stimulation The protocol previously described by Palomba et al. was adopted.26 For response quantification, phosphorylation increase was assessed and calculated as a ratio of positive cells in a stimulated and unstimulated sample. For details and western blot analysis, see the Online Supplementary Appendix.

Table 1. Results of the proteomic analysis of proteins co-immunoprecipitated with ROR1 from chronic lymphocytic leukemia samples.

Accession n.

Protein name

ROR1_HUMAN

Tyrosine protein kinase transmembrane receptor ROR1 OS Homo sapiens GN ROR1 PE 2 SV 2 HOXA 9A OS Homo sapiens GN HOXA 9 PE 2 SV 1 Centromere associated protein E OS Homo sapiens GN CENPE PE 1 SV 2 Cordon bleu protein like 1 OS Homo sapiens GN COBLL1 PE 1 SV 2 Putative small intestine sodium dependent phosphate transport protein OS Homo sapiens GN SLC17A4 PE Adenomatous polyposis coli protein 2 OS Homo sapiens GN APC2 PE 1 SV 1 Kinesin 1 heavy chain OS Homo sapiens GN KIF5B PE 1 SV 1

O75805_HUMAN CENPE_HUMAN COBL1_HUMAN S17A4_HUMAN APC2_HUMAN KINH_HUMAN

314

x x

x x haematologica | 2018; 103(2)

COBLL1 in CLL pathogenesisâ&#x20AC;&#x2030;

Statistical analysis and data visualization For statistical analysis, GraphPad Prism 5 (GraphPad Software), Statistica 10 (StatSoft) and R v.3.1.2.27 supplemented with a KEGG profile package28 were used. Genomic aberrations were visualized as Circos plots.29 COBLL1-linked signaling pathways were analyzed using CLLE-ES dataset (www.icgc.org).30 The cut off dividing patients into COBLL1-low and COBLL1-high subgroups was determined according to their OS. Kaplan-Meier curve dichotomization was accessed for each dCtNCOBLL1 and the value with the strongest difference was further used as cut off. For details, see the Online Supplementary Appendix.

Results COBLL1 is a novel binding partner of ROR1 In order to investigate how ROR1 modulates CLL biology and pathogenesis, we decided to apply a proteomic approach and looked for novel ROR1 protein interaction partners. We immunoprecipitated endogenous ROR1 molecular complexes from the primary CLL cells of 5 CLL patients using anti-ROR1 specific antibody and analyzed the proteins pulled down with mass spectrometry. The hits that were identified in the ROR1 pulldown in at least

Figure 1. COBLL1 is an ROR1-interaction partner. (A) (Left) COBLL1-ROR1 complex was efficiently immunoprecipitated in HEK293 cells transfected with plasmids encoding FLAG-tagged COBLL1 and V5-tagged ROR1. (Right). Endogenous COBLL1 was pulled down with endogenous ROR1 in MAVER-1 cells; unspecific IgG was used as a negative control. Immunoprecipitation input is loaded on the right. Protein levels were determined using western blotting and anti-FLAG, anti-V5, anti-ROR1 and anti-COBLL1 antibodies. IP: immunoprecipitation. (B-D) Immunofluorescence of HEK293 cells transfected with plasmids encoding FLAG-COBLL1 (B-D) and V5ROR1 (C and D). COBLL1 over-expressed in HEK293 cells shows mostly cytoplasmic localization (B), but co-localizes with ROR1 in the membrane when ROR1 is coexpressed (C). The most efficient ROR1 and COBLL1 co-localization is observed in filopodia formed as a consequence of ROR1 overexpression (D, indicated by arrows). Protein expression was visualized using anti-FLAG, anti-V5 and corresponding secondary fluorescein-conjugated antibodies. Nuclei were visualized using DAPI staining.

haematologica | 2018; 103(2)

315

H. Plešingerová et al.

2 patients are shown in Table 1. We compared this list of putative ROR1 interaction partners with the microarraybased dataset of genes differentially expressed in M-CLL versus U-CLL samples.31 This comparison pointed out the cordon blue protein-like 1 (COBLL1) protein as one of the most promising targets. In the next step, we focused on the validation and functional characterization of COBLL1, encoded by the COBLL1 gene. First, we aimed to independently confirm that COBLL1 can indeed physically interact with ROR1. We transfected

HEK293 cells with plasmids encoding FLAG-tagged COBLL1 and V5-tagged ROR1 plasmids, and immunoprecipitated COBLL1 using anti-FLAG specific antibody. As shown in Figure 1A left, ROR1 can be efficiently coimmunoprecipitated with COBLL1. In order to confirm the interaction on an endogenous level in lymphoid cells, we also co-immunoprecipitated COBLL1 in endogenous ROR1 pulldown using anti-ROR1 antibody from protein lysates of MAVER-1 cells,32 a mantle cell lymphoma cell line expressing both COBLL1 and ROR1 (Figure 1A right).

Figure 2. COBLL1 expression in chronic lymphocytic leukemia (CLL) and non-malignant B cells. (A) COBLL1 mRNA expression in 86 mutated CLL (M-CLL), 92 unmutated CLL (U-CLL), and healthy B cells isolated from 4 tonsils and 5 peripheral blood (PB) samples. Individual dots represent individual patients. Full lines indicate median. dCtN - dCt value normalized for three independent datasets (see Methods); ***P<0.0001, Mann-Whitney test. (B) COBLL1 expression histogram follows a bimodal distribution pattern in U-CLL. (C) COBLL1 protein levels correspond very well with COBLL1 mRNA both in M-CLL and U-CLL cells. COBLL1 protein levels in CLL cells were determined using western blotting and anti-COBLL1 antibody. Actin was used as a loading control. Patient samples are ordered according to their IGHV mutation status and COBLL1 mRNA expression [in the ascending order; numbers indicate patients’ relative COBLL1 expression determined by qRT-PCR (see A)]. (D) COBLL1 mRNA expression does not change with time or treatment. COBLL1 expression was analyzed in each patient at two time points (T1 and T2, connected by line) with (left; 6 M-CLL, 6 U-CLL) or without (right; 1 M-CLL, 13 U-CLL) therapy in the interim. Patients were administrated mainly fludarabine-cyclophosphamide-rituximab (FCR) regimen. ● M-CLL, ● U-CLL: open circle; FCR; full circle: other therapy. Wilcoxon signed rank test. ns: not significant. (E) COBLL1 expression does not change with time or treatment. Protein expression was detected in 2 U-CLL COBLL1-low patients (1 and 2) and 2 U-CLL COBLL1-high patients (3 and 4) at two time points with or without therapy in the interim. Western blotting and anti-COBLL1 and anti-actin (as a loading control) antibody was used.

316

haematologica | 2018; 103(2)

COBLL1 in CLL pathogenesis

These co-immunoprecipitation experiments confirmed that ROR1 can indeed interact with COBLL1 both at the exogenous as well as endogenous level. ROR1 is a transmembrane receptor, which in most cell types has the capability to induce filopodia formation.33 Immunofluorescence staining of transfected HEK293 cells showed that, when solely COBLL1 is expressed, it was localized mainly in the cytoplasm (Figure 1B). However, when ROR1 was co-expressed, the COBLL1 signal was detected predominantly in the plasma membrane (Figure

1C) where it co-localized with ROR1. The co-localization was most prominent in the filopodia, which formed as a consequence of ROR1 overexpression (Figure 1D). These data demonstrate that COBLL1 is a true ROR1 binding partner which is recruited to the ROR1 signaling complexes in the membrane.

COBLL1 expression levels vary dramatically among CLL To evaluate COBLL1 relevance in CLL, we analyzed its expression in the cohort of 178 CLL untreated patients (86

Figure 3. Unmutated chronic lymphocytic leukemia (U-CLL) COBLL1-high patients show significantly shorter survival and progress more often compared to U-CLL COBLL1-low patients. (A) U-CLL COBLL1-high patients show shorter overall survival (left) and time to second treatment (right). Survival data are presented using Kaplan-Meier plots and tested by Gehan-Breslow-Wilcoxon test. (B) U-CLL COBLL1-high patients progress more often than patients in other groups. Progression (left y-axis, gray columns) categorized as 1 - no treatment and no/slow progression (clinical stage Rai 0/I at both diagnosis and sampling); 2 - no treatment but rapid progression (clinical stage Rai 0/I at diagnosis and II/III/IV at sampling); 3 - treatment or CLL-related death (various clinical stages at diagnosis and sampling). Patients are grouped based on their IGHV mutation/COBLL1 expression status, and ordered according to germline IGHV (in the ascending order, x-axis) and COBLL1 expression (descending order, full line, right y-axis). (A left and B). N: 86 mutated CLL (M-CLL), 58 U-CLL COBLL1-low, 34 U-CLL COBLL1-high. (A right). N: 28 M-CLL, 48 UCLL COBLL1-low, 32 U-CLL COBLL1-high. *P≤0.05, **P≤0.01, ***P≤0.001. CLL progression was tested by Fisher’s exact test (U-CLL COBLL1-high vs. U-CLL COBLL1low).

haematologica | 2018; 103(2)

317

H. Plešingerová et al. Table 2. Multivariate Cox analysis in unmutated chronic lymphocytic leukemia overall survival.

HR COBLL1 - high Age at diagnosis Risk according to Rai stage at diagnosis* Intermediate High CD38 - positive Cytogenetic hierarchical model** del(17p) del(11q) trisomy 12 del(13q)

95%CI for HR Lower

Upper

2.924 1.025

1.372 0.982

6.232 1.070

0.005 0.261

1.514 6.029 3.086

0.668 2.003 1.087

3.427 18.145 8.757

0.320 0.001 0.034

5.049 2.503 0.816 1.232

1.691 0.853 0.192 0.399

15.076 7.344 3.468 3.811

0.004 0.095 0.783 0.717

*Compared to low risk. **Compared to normal karyotype. HR: hazard ratio; CI: confidence interval. Statistically significant P-values are highlighted in bold.

M-CLL, 92 U-CLL) and compared it with non-malignant B cells from PB (5 samples) and tonsillar tissue (4 samples). COBLL1 was highly expressed in normal PB and tonsillar tissue (Figure 2A). The expression in individual tonsillar Bcell subpopulations varied; COBLL1 expression in centroblasts and centrocytes was increased compared to naïve and memory B cells. The expression in CLL cells differed significantly according to the IGHV mutation status (P<0.0001, MannWhitney test). COBLL1 levels were higher in M-CLL patients with an expression comparable to that of healthy tonsillar and PB B cells. On the contrary, the COBLL1 expression in U-CLL showed bimodal distribution (Figure 2B). A subgroup of U-CLL patients expressed COBLL1 at a level comparable with M-CLL patients, but in the majority of U-CLL samples COBLL1 expression was much lower. Since the COBLL1 expression had such a clearly bimodal distribution in all three independently analyzed datasets (see Methods section and Online Supplementary Figure S1), we set a cut off to distinguish COBLL1-high and COBLL1-low patients (for details see Methods section). The cut off is set close to the local distribution minimum (Figure 2B). Following this approach, all but one MCLL patient was classified as COBLL1-high. The majority of U-CLL patients (n=58; 63%) were classified as COBLL1-low, whereas the remaining U-CLL patients (n=34; 37%) were classified as COBLL1-high. Different expression in both cohorts was also confirmed at protein level (Figure 2C). To analyze the changes in COBLL1 expression over time and after treatment, we examined 26 patients at two time points (7 M-CLL, 19 U-CLL) (Figure 2D). A part of the cohort was not treated in the interim (6 M-CLL, 6 UCLL; median 37 months), whereas the remaining patients (1 M-CLL, 13 U-CLL; median 35 months) were administrated a fludarabine-cyclophosphamide-rituximab regimen or another chemoimmunotherapy. The COBLL1 expression category did not change, with one exception: COBLL1 expression was slightly increased after treatment in one borderline U-CLL COBLL1-low patient. We also examined the changes in expression at protein level in 4 U-CLL patients and obtained similar data (2 U-CLL COBLL1-low, 2 U-CLL COBLL1-high; 2 with treatment in 318

the interim, 2 without treatment in the interim) (Figure 2E). This suggests that COBLL1 expression at mRNA as well as protein level does not dramatically change with time or treatment. Recently, CLL patients with high ROR1 expression were found to suffer from a more aggressive disease.34 Since COBLL1 and ROR1 form a protein complex, we correlated COBLL1 and ROR1 expression (protein levels of COBLL1 and ROR111 correspond well with mRNA levels) (Figure 2C) but did not find any correlation (Online Supplementary Figure S2A). We were also unable to detect any obvious changes in COBLL1 levels or phosphorylation (detected as phosphorylation-dependent mobility shift) upon activation of ROR1 by its ligand Wnt-5a (Online Supplementary Figure S2B). This suggests that COBLL1 rather represents an independently-regulated ROR1 signaling modulator than a bona fide component of ROR1 signaling pathway.

High COBLL1 expression identifies a subgroup of U-CLL patients with inferior prognosis independent of other prognostic markers To explore the possible COBLL1 association with CLL disease course, we analyzed the survival of M-CLL, UCLL COBLL1-low and U-CLL COBLL1-high patients. As expected, M-CLL patients showed the best prognosis according to OS and time to second treatment (median OS and TTST not reached; M-CLL vs. U-CLL COBLL1low POS=0.0389, PTTST=0.0104; M-CLL vs. U-CLL COBLL1-high POS<0.0001, PTTST=0.0004, Gehan-BreslowWilcoxon test) (Figure 3A). The survival of U-CLL patients differed according to COBLL1 expression. U-CLL COBLL1-high patients showed a more aggressive disease course (median OS 65 months, TTST 17 months), whereas the U-CLL COBLL1-low patients progressed more slowly (median OS 123 months, TTST 37 months; U-CLL COBLL1-high vs. U-CLL COBLL1-low POS=0.0086, PTTST=0.0116). There was no significant difference in time to first treatment (TTFT) between U-CLL COBLL1-high and low (Online Supplementary Figure S3). To get further insight into the role of COBLL1, we categorized M-CLL, U-CLL COBLL1-low and U-CLL COBLL1-high patients, based on the aggressiveness of the haematologica | 2018; 103(2)

COBLL1 in CLL pathogenesis

disease. This parameter was defined based on the disease behavior between diagnosis and sampling (median time between diagnosis and sampling: 35 months in M-CLL, 11 months in U-CLL COBLL1-high, 19 months in U-CLL COBLL1-low). Patients were categorized into three groups: 1) no treatment and no/slow progression (clinical stage Rai 0/I at sampling); 2) no treatment but rapid progression (progression into clinical stage Rai II/III/IV at sampling); 3) treatment or CLL-related death. U-CLL

COBLL1-high progressed more often than U-CLL COBLL1-low (P=0.0297, Fisher’s exact test) (Figure 3B). UCLL COBLL1-high progressed in almost all cases; only 3% did not progress versus 17% in U-CLL COBLL1-low patients. In line with this observation, treatment or CLLrelated death occurred more often in the U-CLL COBLL1high patients than in U-CLL COBLL1-low patients (94% vs. 83%). To further confirm the difference in U-CLL patients

Figure 4. The survival difference between unmutated chronic lymphocytic leukemia (U-CLL) COBLL1-high and U-CLL COBLL1-low is not caused by recurrent mutations and chromosomal abnormalities. (A-D) Samples were analyzed by I-FISH [del(17p), del(11q), trisomy 12, del(11q)] and sequencing (mutations in IGHV, TP53, BIRC3, NOTCH1, SF3B1). TP53 defect - TP53 mutation, deletion or both. (A) U-CLL COBLL1-low and U-CLL COBLL1-high patients do not exhibit any differences in the occurrence of recurrent defects [TP53 defect, BIRC3, NOTCH1 and SF3B1 mutations, del(11q), trisomy 12, del(13q)] or (B) in cytogenetic aberrations evaluated according to the hierarchical model.36 (C) Expression of COBLL1 categorized according to the IGHV mutation load. (D) U-CLL COBLL1-high patients exhibit non-significantly higher incidence of TP53 defect at diagnosis or its later selection. (A) 41 mutated CLL (M-CLL), 37 U-CLL COBLL1-low, 29 U-CLL COBLL1-high. (B-D) 86 MCLL, 58 U-CLL COBLL1-low, 34 U-CLL COBLL1-high. (D) Adverse survival of U-CLL COBLL1-high patients is retained even in TP53 wild-type patients. (D) (Left) Overall survival: 46 U-CLL COBLL1-low, 24 U-CLL COBLL1-high. (D) (Right) Time to second treatment: 37 U-CLL COBLL1-low, 22 U-CLL COBLL1-high. *P≤0.05, **P≤0.01, ***P≤0.001. Aberrations frequency tested by Fisher’s exact test, survival data tested by Gehan-Breslow-Wilcoxon test, germline IGHV tested by Mann-Whitney test.

haematologica | 2018; 103(2)

319

H. Plešingerová et al.

overall survival, we analyzed our patient cohorts separately (cohort B vs. cohort A+C). The U-CLL COBLL1-high patients showed a shorter OS than U-CLL COBLL1-low in both cases (58 vs. 75 months in cohort B and 75 vs. 123 months in cohort A+C) (Online Supplementary Figure S4)

but the difference was significant only in cohort B (P=0.0314, Gehan-Breslow-Wilcoxon test); this is likely due to the relatively small number of patients. The striking difference in survival of U-CLL COBLL1-

Figure 5. U-CLL COBLL1-high cells show higher response upon BCR stimulation. (A and B). Chronic lymphocytic leukemia (CLL) cells (4 mutated CLL (M-CLL), 8 UCLL COBLL1-low, 6 U-CLL COBLL1-high) were stimulated for 4 minutes with anti-IgM and response to BCR stimulation was analyzed using phospho-specific antibodies targeted against pPLCγ2, pSYK and pBLNK. (A) Representative examples of M-CLL, U-CLL COBLL1-low and U-CLL COBLL1-high patients. Histograms show a negative control (unstimulated non-stained sample, dotted line), unstimulated sample (full line) and IgM-stimulated sample (full line, gray area). Percentage of positive cells is indicated (unstimulated sample → stimulated sample). (B) Quantification of changes in the pPLCγ2, pSYK and pBLNK. Phosphorylation increase (y-axis) was calculated as a ratio of positive cells in IgM-stimulated versus unstimulated samples. Box-and-Whisker plots show quartiles and median. Dashed line indicates phosphorylation increase in non-malignant peripheral blood (PB) B cells (mean), • outliers, *P≤0.05, **P≤0.01, ***P≤0.001. Mann-Whitney test. (C) Western blot analysis of representative U-CLL samples treated with anti-IgM and analyzed for activation of BCR components using phospho-specific antibodies - PLCγ2 (pY1217), pSYK (pY525/526), pAKT (pS473) and pERK1/2 (pT202/Y204). Loading control: β-actin (left), total PLCγ2 (right). (D) Correlation of the response at the level of individual kinases (Spearman correlation). Statistically significant P-values are highlighted in bold with gray background. See Online Supplementary Figure S7 for details and raw data.

320

haematologica | 2018; 103(2)

COBLL1 in CLL pathogenesis

high and U-CLL COBLL1-low patients leads to compare the genetic aberrations which could influence the patients’ prognosis in both cohorts. We analyzed cytogenetic aberrations [del(17p), del(11q), trisomy 12, del(13q)] and recurrent mutations (TP53, NOTCH1, BIRC3, SF3B1) in 107 patients (41 M-CLL, 37 U-CLL COBLL1-low, 29 U-CLL COBLL1-high) where all these data were available. We were unable to find any significant difference in U-CLL COBLL1-high versus COBLL1-low categories (see Figure 4A for brick plot, Online Supplementary Figure S5 for Circos plot, Figure 4B for hierarchically categorized cytogenetic aberrations;35 Fisher’s exact test). Since U-CLL patients with borderline IGHV mutations have been shown to have a better prognosis than patients with truly unmutated IGHV,36 we also compared mutation load in U-CLL patients. The worse prognosis of U-CLL COBLL1-high patients could not be explained by difference in mutation load; on the contrary, borderline mutated patients (98-99.9%) showed higher expression of COBLL1 than patients with 100% identity (P=0.0219, MannWhitney test) (Figure 4C). We further assessed the influence of TP53 aberrations (mutations, deletions or both) present either before treatment or evolving during disease progression. Although UCLL COBLL1-high patients harbored TP53 aberrations more often and also lost wild-type TP53 more often during disease evolution, the differences were not significant (P=0.4823; Fisher’s exact test) (Figure 4D). Furthermore, when we compared only wild-type TP53 patient survival, the U-CLL COBLL1-high patients still retained a worse OS and TTST (U-CLL COBLL1-low: median OS 122 months, median TTST 42 months; U-CLL COBLL1-high: median OS 88 months, median TTST 23 months; POS=0.0276, PTTST=0.0404; Gehan-Breslow-Wilcoxon test) (Figure 4E). To evaluate COBLL1 significance in U-CLL survival, we performed univariate and multivariate Cox regression analyses. The univariate analysis revealed COBLL1 status, age at diagnosis, Rai stage at diagnosis, CD38 expression

and cytogenetic aberrations as significant prognostic factors for OS in U-CLL. Multivariate Cox regression analysis confirmed COBLL1 as an independent molecular marker (Table 2). In multivariate Cox regression analysis for TTST, COBLL1 did not retain independence. Moreover, we did not find any difference in the clinical parameters such as leukocytosis, clinical stage, age or sex, explaining the short survival of U-CLL COBLL1-high patients. Due to the striking difference in TTST, we also investigated the administered treatment in detail (Online Supplementary Figure S6). We did not find any difference in patient treatment response, length or number of received treatment cycles or if categorized as full, reduced, interrupted or reduced therapy. Therefore, we assumed that the aggressive course of U-CLL COBLL1-high patients cannot be explained by any common unfavorable clinicobiological disease characteristics.

U-CLL COBLL1-high cells show higher phosphorylation upon BCR stimulation To understand our findings in context, we performed detailed bioinformatics analysis of publicly available RNA sequencing data of 44 U-CLL samples30 (see Methods section). A subset of 1240 significantly COBLL1-correlated genes (P<0.05; Spearman test) (Online Supplementary Table S1) was selected for KEGG pathway analyses. Among the transcripts positively correlating with COBLL1 in U-CLL the genes associated with various cancer-linked signaling pathways and metabolic processes, including the B-cell receptor (BCR) pathway, were enriched (Online Supplementary Table S2). Progressive phosphorylation of BCR pathway components promotes cell survival, differentiation and proliferation in CLL (for review see ten Hacken and Burger37). Given the crucial biological and clinical importance of BCR signaling in CLL cells, we hypothesized that U-CLL COBLL1-high patients might have a deregulated response to BCR stimulation. To investigate how U-CLL COBLL1-high cells respond to BCR stimulation, we adopted a previously described

Figure 6. Unmutated chronic lymphocytic leukemia (U-CLL) COBLL1-high cells show deregulated chemotaxis and motility. Migratory properties of 10 mutated CLL (M-CLL), 10 U-CLL COBLL1-low, and 6 U-CLL COBLL1-high samples were assessed using transwell plates. (A) Chemotaxis towards chemokine CCL19 expressed as migration index (MI). (B) Chemotaxis towards chemokine CXCL12 expressed as MI. (C) Basal migration. MI was calculated as the number of cells migrated towards chemokine divided by the number of cells migrated in chemokine-free media. Basal migration was calculated as the percentage of migrated cells from all seeded cells. Each measurement was performed in a technical triplicate. Bars represent mean+Standard Deviation (S.D.) (A and B) Individual dots represent individual patients (C). *P≤0.05, **P≤0.01. (Mann-Whitney test).

haematologica | 2018; 103(2)

321

H. Plešingerová et al. protocol,26 stimulated CLL cells and examined the phosphorylation level of selected BCR signaling pathway components (pPLCγ2, pSYK and pBLNK) via flow cytometry. Eighteen CLL samples (4 M-CLL, 8 U-CLL COBLL1-low, 6 U-CLL COBLL1-high) and 3 peripheral blood (PB) B-cell samples from healthy donors were analyzed. The response to anti-IgM was evaluated as a difference (fold change) in the number of positive cells in stimulated and unstimulated samples. Not surprisingly, the response to BCR stimulation in CLL cells from individual patients was rather heterogeneous but still showed clear trends in the individual groups (see representative examples in Figure 5A). When quantified (Figure 5B), the number of pPLCγ2-positive cells after BCR stimulation was dramatically increased only in U-CLL COBLL1-high [U-CLL COBLL1-high vs. UCLL COBLL1-low (P=0.0007), U-CLL COBLL1-high vs. M-CLL (P=0.0007), vs. M-CLL (P=0.0139)]. A similar trend could also be seen for pSYK [U-CLL COBLL1-high vs. UCLL COBLL1-low (P=0.0609), U-CLL COBLL1-high vs. M-CLL (P=0.0095), Mann-Whitney test] and pBLNK where U-CLL COBLL1-high cells responded best, albeit not with statistical significance. The non-malignant PB B-cell controls showed a uniform response, which was very similar to that of M-CLL (Figure 5B). Interestingly, the western blot analysis confirmed in principle the differences in the activation of upstream BCR signaling components, namely PLCγ2 and SYK, but we were unable to detect any differences between the groups at the activated AKT and ERK1/2 level (Figure 5C). Quantitative analysis of western blot data from a larger cohort of CLL samples (n=10 for pAKT, 11 for all others) showed that pPLCγ2 and pSYK signals correlated strongly with each other (Figure 5D, graphs in Online Supplementary Figure S7) but not with the pAKT and pERK1/2 signals that were almost uniformly induced in all patients (Online Supplementary Figure S7A). This suggests that regulating the upstream (PLCγ2, SYK) and downstream (ERK1/2, AKT) BCR pathway module can differ. We conclude that U-CLL COBLL1-high patients exhibit an enhanced response to BCR stimulation, in particular at the level of upstream components such as pPLCγ2 and pSYK.

U-CLL COBLL1-high cells exhibit impaired migration and chemotaxis COBLL1 can physically interact with ROR1 and thus represents a candidate regulator of the non-canonical Wnt/PCP pathway. Since the Wnt/PCP pathway was shown to be involved in the migration of CLL cells,11,13 we analyzed their ability to respond to chemokines CCL19 and CXCL12, known to stimulate the cells via CCR7 and CXCR4 receptors, respectively.38 CLL cells were stratified according to the expression of COBLL1 and IGHV mutational status (10 M-CLL, 10 U-CLL COBLL1-low, 6 U-CLL COBLL1-high). Generally, the chemotactic and migratory abilities differed according to the combination of IGHV status and COBLL1 expression (Figure 6). U-CLL COBLL1-high cells showed impaired chemotaxis towards chemokines CCL19 and CXCL12 and increased basal migration compared to U-CLL COBLL1-low and M-CLL cells [both CCL19 and CXCL12: U-CLL COBLL1-high vs. M-CLL (P=0.0017), U-CLL COBLL1-high vs. U-CLL COBLL1-low (P=0.0302); basal migration: U-CLL COBLL1-high vs. MCLL (P=0.0030), U-CLL COBLL1-high vs. U-CLL COBLL1322

low (P=0.0420) Mann-Whitney test]. U-CLL COBLL1-low exhibited an intermediate response to chemokine stimuli. The deregulated migratory abilities of U-CLL COBLL1high cells further point out their altered microenvironmental interactions.

Discussion In this study, we have identified COBLL1 as a novel binding partner for ROR1 in CLL. COBLL1 is an evolutionary conserved but very little known protein. Its mouse ortholog Cordon blue (Cobl) interacts with Vang-like protein 2 (Vangl2; a component of the Wnt/PCP pathway) and is required for neural tube closure, which is a process typically regulated by the Wnt/PCP pathway.39 The combination of these results and our findings suggests that COBLL1 can represent a Wnt/PCP pathway regulator in mammals. COBLL1 levels in CLL dramatically vary and do not correlate with ROR1. This opens up the possibility that the way COBLL1 affects ROR1 function may differ depending on the level of COBLL1. COBLL1 links to human pathological conditions are very limited and restricted to the observation that COBLL1 upregulation is associated with a better prognosis after surgery in malignant pleural mesothelioma, where it acts as a negative regulator of apoptosis.40 On the contrary, COBLL1 upregulation in chronic myeloid leukemia patients was recently associated with a reduction in nilotinib-dependent apoptosis, disease progression and shorter OS.41 Besides mature B cells, COBLL1 expression is detectable in various other cell types including other blood elements (such as T cells), although usually at a much lower level.42 Comparable or higher COBLL1 levels were detected in mast cells, adipocytes, placenta and esophagus.42 Our data show that in M-CLL, COBLL1 expression is uniformly high, whereas in U-CLL patients it ranges from low to high levels. The U-CLL COBLL1-high cohort showed a strikingly worse prognosis than the COBLL1low. Shorter OS and TTST of U-CLL COBLL1-high patients remained, even after excluding patients with aberrant TP53 and the independence of COBLL1 as a prognostic factor in U-CLL for OS was proven by multivariate analysis. IGHV mutational status and COBLL1 expression thus represents a novel marker combination, which efficiently identifies patients with short OS and TTST. We showed bimodal COBLL1 expression distribution in three independent cohorts. The U-CLL patients can be categorized according to a cut off close to a local distribution minimum which facilitates access to our marker combination by other laboratories if desirable. In addition to qRT-PCR-based assessment of COBLL1 expression, COBLL1 protein levels can in principle, be analyzed using flow cytometry. However, this would require fixation (COBLL1 is a cytoplasmic protein) and staining with a primary and secondary antibody, since there are currently no well-validated fluorescently-conjugated monoclonal antibodies. Interestingly, functional analysis of U-CLL COBLL1-high CLL cells showed a higher response to BCR stimulation and deregulated chemotaxis in this patient cohort. This is in line with a large body of evidence showing that increased in vitro response to BCR stimuli associhaematologica | 2018; 103(2)

COBLL1 in CLL pathogenesis

ates with aggressive CLL.43,44 COBLL1-high CLL cells preferentially responded by activation of BTK, SYK and ERK1/2 whereas COBLL1-low U-CLL cells induced only ERK1/2. It has been shown previously that, in healthy B cells, ERK1/2 can be efficiently phosphorylated by Iginduced BCR crosslinking even in cases when no detectable phosphorylation of BTK or SYK is seen.45 There is also some evidence that these different modes of BCR activation depend on the stimulus45 and also differ between healthy and malignant cells.46 This suggests that COBLL1 can regulate this balance and promote the BCR activation mode that involves the upstream BTK/SYK kinases. In addition, U-CLL COBLL1-high cells exhibit impaired migration towards chemokines CCL19 and CXCL12, a phenotype very similar to aggressive CLL cells expressing ROR1 ligand Wnt-5a.11 Although we were not able to correlate WNT5A and COBLL1 expression, both studies indicate lower chemotaxis in patients with aggressive CLL and a deregulated Wnt/PCP signaling pathway. Both observations support the generally accepted view that patients with inferior prognosis often exhibit deregulated interaction with the microenvironment and with other cell types. The clear difference in TTST in U-CLL COBLL1-high patients suggests that standard therapeutic schemes do indeed have limited efficiency in this cohort. Due to their unmutated IGHV (U-CLL patients have been described as more perceptive to ibrutinib than M-CLL47) and high BCR responsiveness, COBLL1 can thus help to identify patients that will benefit more from the new BCR inhibitor-based therapies. The role of COBLL1 in CLL pathogenesis and in B-cell development remains unclear. One striking observation is apparently the difference in importance of high COBLL1 in M-CLL and U-CLL. We were able to confirm previously reported uniformly high COBLL1 levels in M-CLL cells.48,49 Interestingly, M-CLL is generally more indolent than UCLL, where high COBLL1 rather counterintuitively defines patients with an inferior prognosis. Upregulation of COBLL1 in centroblasts and centrocytes compared to naïve and memory cells indicates that COBLL1 is switched on during B-cell maturation in the germinal center. Together with lower IGHV germline identity in U-CLL COBLL1-high patients (compared to

References 1. Klein U, Tu Y, Stolovitzky GA, et al. Gene expression profiling of B cell chronic lymphocytic leukemia reveals a homogeneous phenotype related to memory B cells. J Exp Med. 2001;194(11):1625-1638. 2. Rosenwald A, Alizadeh AA, Widhopf G, et al. Relation of gene expression phenotype to immunoglobulin mutation genotype in B cell chronic lymphocytic leukemia. J Exp Med. 2001;194(11):1639-1647. 3. Masiakowski P, Carroll RD. A novel family of cell surface receptors with tyrosine kinase-like domain. J Biol Chem. 1992; 267(36):26181-26190. 4. Fukuda T, Chen L, Endo T, et al. Antisera induced by infusions of autologous AdCD154-leukemia B cells identify ROR1 as an oncofetal antigen and receptor for Wnt5a. Proc Natl Acad Sci USA. 2008; 105(8):3047-3052.

haematologica | 2018; 103(2)

U-CLL COBLL1-low patients), it suggests that upregulation of COBLL1 expression may be linked to the process of IGHV mutation. This view is also supported by the gene profiling of monoclonal B-lymphocytosis cells (MBL) with mutated and unmutated IGHV50 where COBLL1 expression followed a similar pattern to CLL.48,49 This suggests that deregulating COBLL1 expression likely occurs prior to overt CLL, or, alternatively, points to a different origin of a U-CLL subset from a rare B-cell subset with low COBLL1 expression. This assumption is also supported by the observation that the levels of COBLL1 in U-CLL COBLL1-low samples are lower than any of the healthy Bcell populations analyzed in this study. In summary, we identified COBLL1 as a component of the ROR1 receptor system in CLL cells. COBLL1 expression combined with IGHV hypermutation status correlates with CLL prognosis, and identifies the U-CLL COBLL1-high patients as those having an adverse disease course. U-CLL COBLL1-high cells show an increased response to BCR stimulation and attenuated chemotaxis, which suggests a mutual interplay between Wnt/PCP and BCR pathways in the regulation of CLL response to microenvironmental stimuli. Acknowledgments We wish to thank Christian Arquint and Erich A. Nigg for providing pcDNA3.1-FLAG vector, Peter Koník (University of South Bohemia, České Budějovice) for assisting with the mass spectrometry, Jana Kotašková (University Hospital Brno) for processing survival and treatment data, and Matthew Smith for language correction. Funding Supported by projects of the Czech Science Foundation (1716680S, 17-09525S), Ministry of Health, Czech Republic (1529793A and FNBr 65269705), by Masaryk University (MUNI/A/1106/2016, MUNI/A/0988/2016) and by projects of MEYS CR n. CEITEC 2020 (LQ1601). Also supported by the Czech Leukemia Study Group for Life (CELL). AM was financed by the programme SoMoPro, which is jointly supported by the European Union as part of the 7th Framework Programme (FP/2007-2013, Grant Agreement n. 229603) and the South Moravian Region. VB is supported by Neuron – Fund for Support of Science.

5. Baskar S, Kwong KY, Hofer T, et al. Unique cell surface expression of receptor tyrosine kinase ROR1 in human B-cell chronic lymphocytic leukemia. Clin Cancer Res. 2008; 14(2):396-404. 6. Kotašková J, Pavlová Š, Greif I, et al. ROR1based immunomagnetic protocol allows efficient separation of CLL and healthy B cells. Br J Haematol. 2016;175(2):339-342. 7. Daneshmanesh A, Hojjat-Farsangi M, Khan A, et al. Monoclonal antibodies against ROR1 induce apoptosis of chronic lymphocytic leukemia (CLL) cells. Leukemia. 2012; 26(6):1348-1355. 8. Berger C, Sommermeyer D, Hudecek M, et al. Safety of targeting ROR1 in primates with chimeric antigen receptor-modified T cells. Cancer Immunol Res. 2015;3(2):206216. 9. Hudecek M, Schmitt TM, Baskar S, et al. The B-cell tumor associated antigen ROR1 can be targeted with T cells modified to

10.

11.

12.

13.

express a ROR1-specific chimeric antigen receptor. Blood. 2010;116(22):4532-4541. Kaucká M, Krejčí P, Plevová K, et al. Posttranslational modifications regulate signalling by Ror1. Acta Physiol. 2011; 203(3):351-362. Janovska P, Poppova L, Plevova K, et al. Autocrine signaling by Wnt-5a deregulates chemotaxis of leukemic cells and predicts clinical outcome in chronic lymphocytic leukemia. Clin Cancer Res. 2016; 22(2):459469. Khan AS, Hojjat-Farsangi M, Daneshmanesh AH, et al. Dishevelled proteins are significantly upregulated in chronic lymphocytic leukaemia. Tumor Biol. 2016; z37(9):11947-11957. Kaucká M, Plevová K, Pavlová Š, et al. The planar cell polarity pathway drives pathogenesis of chronic lymphocytic leukemia by the regulation of B-lymphocyte migration. Cancer Res. 2013;73(5):1491-1501.

323

H. Plešingerová et al. 14. Yu J, Chen L, Cui B, et al. Wnt5a induces ROR1/ROR2 heterooligomerization to enhance leukemia chemotaxis and proliferation. J Clin Invest. 2016;126(2):585-598. 15. Plevova K, Francova HS, Burckova K, et al. Multiple productive immunoglobulin heavy chain gene rearrangements in chronic lymphocytic leukemia are mostly derived from independent clones. Haematologica. 2014;99(2):329-338. 16. Baliakas P, Hadzidimitriou A, Sutton LA, et al. Clinical effect of stereotyped B-cell receptor immunoglobulins in chronic lymphocytic leukaemia: A retrospective multicentre study. Lancet Haematol. 2014; 1(2):e74-e84. 17. Baliakas P, Hadzidimitriou A, Sutton L, et al. Recurrent mutations refine prognosis in chronic lymphocytic leukemia. Leukemia. 2015;29(2):329-336. 18. Baliakas P, Iskas M, Gardiner A, et al. Chromosomal translocations and karyotype complexity in chronic lymphocytic leukemia: A systematic reappraisal of classic cytogenetic data. Am J Hematol. 2014; 89(3):249-255. 19. Malcikova J, Smardova J, Rocnova L, et al. Monoallelic and biallelic inactivation of TP53 gene in chronic lymphocytic leukemia: Selection, impact on survival, and response to DNA damage. Blood. 2009; 114(26):5307-5314. 20. Pospisilova S, Gonzalez D, Malcikova J, et al. ERIC recommendations on TP53 mutation analysis in chronic lymphocytic leukemia. Leukemia. 2012;26(7):1458-1461. 21. Cajanek L, Ganji RS, Henriques-Oliveira C, et al. Tiam1 regulates the Wnt/Dvl/Rac1 signaling pathway and the differentiation of midbrain dopaminergic neurons. Mol Cell Biol. 2013;33(1):59-70. 22. de Groot REA, Ganji RS, Bernatik O, et al. Huwe1-mediated ubiquitylation of dishevelled defines a negative feedback loop in the Wnt signaling pathway. Sci Signal. 2014;7(317):ra26. 23. Arquint C, Sonnen KF, Stierhof Y-D, Nigg EA. Cell-cycle-regulated expression of STIL controls centriole number in human cells. J Cell Sci. 2012;125(Pt 5):1342-1352. 24. Cervenka I, Valnohova J, Bernatik O, et al. Dishevelled is a NEK2 kinase substrate controlling dynamics of centrosomal linker proteins. Proc Natl Acad Sci USA. 2016; 113(33):9304-9309. 25. Bryja V, Schulte G, Arenas E. Wnt-3a utilizes a novel low dose and rapid pathway that does not require casein kinase 1-mediated phosphorylation of Dvl to activate β-catenin. Cell Signal. 2007;19(3):610-616. 26. Palomba ML, Piersanti K, Ziegler CGK, et al. Multidimensional single-cell analysis of BCR signaling reveals proximal activation defect as a hallmark of chronic lymphocytic leukemia B cells. PLoS One. 2014;

324

9(1):e79987. 27. R Development Core Team R. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing; 2011;p409.(R Foundation for Statistical Computing; vol.1). Available from: http://www.r-project.org. 28. Zhao S, Guo Y, Shyr Y. KEGGprofile: An annotation and visualization package for multi-types and multi-groups expression data in KEGG pathway. R package version 1.14.0. 29. Krzywinski MI, Schein JE, Birol I, et al. Circos: An information aesthetic for comparative genomics. Genome Res. 2009; 19(9):1639-1645. 30. Ramsay AJ, Martínez-Trillos A, Jares P, Rodríguez D, Kwarciak A, Quesada V. Next-generation sequencing reveals the secrets of the chronic lymphocytic leukemia genome. Clin Transl Oncol. 2013; 15(1):3-8. 31. Kotaskova J, Tichy B, Trbusek M, et al. High expression of lymphocyte-activation gene 3 (LAG3) in chronic lymphocytic leukemia cells is associated with unmutated immunoglobulin variable heavy chain region (IGHV) gene and reduced treatment-free survival. J Mol Diagn. 2010;12(3):328-334. 32. Zamò A, Ott G, Katzenberger T, et al. Establishment of MAVER-1 cell line, a model for leukemic and aggressie mantle cell lymphoma. Haematologica. 2006; 91(1):40-47. 33. Paganoni S, Ferreira A. Neurite extension in central neurons: a novel role for the receptor tyrosine kinases Ror1 and Ror2. J Cell Sci. 2005;118(Pt 2):433-446. 34. Cui B, Ghia EM, Chen L, et al. High-level ROR1 associates with accelerated diseaseprogression in chronic lymphocytic leukemia. Blood. 2016;128(25):2931-2940. 35. Döhner H, Stilgenbauer S, Benner A, et al. Genomic Aberrations and Survival in Chronic Lymphocytic Leukemia. N Engl J Med. 2000;343(26):1910-1916. 36. Hamblin TJ, Davis ZA, Oscier DG. Determination of how many immunoglobulin variable region heavy chain mutations are allowable in unmutated chronic lymphocytic leukaemia - Long-term follow up of patients with different percentages of mutations. Br J Haematol. 2008;140(3):320-323. 37. ten Hacken E, Burger JA. Microenvironment interactions and B-cell receptor signaling in Chronic Lymphocytic Leukemia: Implications for disease pathogenesis and treatment. Biochim Biophys Acta. 2016;1863(3):401-413. 38. Burger JA. Chemokines and chemokine receptors in chronic lymphocytic leukemia (CLL): From understanding the basics towards therapeutic targeting. Semin Cancer Biol. 2010;20(6):424-430. 39. Carroll EA, Gerrelli D, Gasca S, et al.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

Cordon-bleu is a conserved gene involved in neural tube formation. Dev Biol. 2003; 262(1):16-31. Gordon GJ, Jensen R V, Hsiao LL, et al. Using gene expression ratios to predict outcome among patients with mesothelioma. J Natl Cancer Inst. 2003;95(8):598-605. Han SH, Kim SH, Kim HJ, et al. Cobll1 is linked to drug resistance and blastic transformation in chronic myeloid leukemia. Leukemia. 2017;31(7):1532-1539. Benita Y, Cao Z, Giallourakis C, Li C, Gardet A, Xavier RJ. Gene enrichment profiles reveal T-cell development, differentiation, and lineage-specific transcription factors including ZBTB25 as a novel NF-AT repressor. Blood. 2010;115(26):5376-5384. Morabito F, Cutrona G, Gentile M, et al. Prognostic relevance of in vitro response to cell stimulation via surface IgD in binet stage a CLL. Br J Haematol. 2010; 149(1):160-163. Lanham S, Hamblin T, Oscier D, Ibbotson R, Stevenson F, Packham G. Differential signaling via surface IgM is associated with VH gene mutational status and CD38 expression in chronic lymphocytic leukemia. Blood. 2003;101(3):1087-1093. Irish JM, Czerwinski DK, Nolan GP, Levy R. Kinetics of B Cell Receptor Signaling in Human B Cell Subsets Mapped by Phosphospecific Flow Cytometry. J Immunol. 2006;177(3):1581-1589. Irish JM, Czerwinski DK, Nolan GP, Levy R. Altered B-cell receptor signaling kinetics distinguish human follicular lymphoma B cells from tumor-infiltrating nonmalignant B cells. Blood. 2006;108(9):3135-3142. Thompson PA, O’Brien SM, Wierda WG, et al. Complex karyotype is a stronger predictor than del(17p) for an inferior outcome in relapsed or refractory chronic lymphocytic leukemia patients treated with ibrutinibbased regimens. Cancer. 2015; 121(20):3612-3621. Abruzzo LV, Barron LL, Anderson K, et al. Identification and validation of biomarkers of IgV(H) mutation status in chronic lymphocytic leukemia using microfluidics quantitative real-time polymerase chain reaction technology. J Mol Diagn. 2007; 9(4):546-555. Plesingerova H, Librova Z, Plevova K, et al. COBLL1, LPL and ZAP70 expression defines prognostic subgroups of chronic lymphocytic leukemia patients with high accuracy and correlates with IGHV mutational status. Leuk Lymphoma. 2017; 58(1):70-79. Morabito F, Mosca L, Cutrona G, et al. Clinical monoclonal B lymphocytosis versus rai 0 chronic lymphocytic leukemia: A comparison of cellular, cytogenetic, molecular, and clinical features. Clin Cancer Res. 2013;19(21):5890-5900.

haematologica | 2018; 103(2)

ARTICLE

Plasma Cell Disorders

Maternal embryonic leucine zipper kinase is a novel target for proliferation-associated high-risk myeloma

Ferrata Storti Foundation

Arnold Bolomsky,1# Roy Heusschen,2# Karin Schlangen,3 Kathrin Stangelberger,1 Joséphine Muller,2 Wolfgang Schreiner,3 Niklas Zojer,1 Jo Caers2,4* and Heinz Ludwig1*

Wilhelminen Cancer Research Institute, Department of Medicine I, Wilhelminenspital, Vienna, Austria; 2Laboratory of Hematology, GIGA-I3, University of Liège, Belgium; Center for Medical Statistics, Informatics and Intelligent Systems, Section for Biosimulation and Bioinformatics, Medical University of Vienna, Austria and 4Division of Hematology, Department of Medicine, University and CHU of Liège, Belgium

1 3

Haematologica 2018 Volume 103(2):325-335

#AB and RH are co-first authors. *JC and HL contributed equally to this work.

ABSTRACT

reatment of high-risk patients is a major challenge in multiple myeloma. This is especially true for patients assigned to the gene expression profiling-defined proliferation subgroup. Although recent efforts have identified some key players of proliferative myeloma, genetic interactions and players that can be targeted with clinically effective drugs have to be identified in order to overcome the poor prognosis of these patients. We therefore examined maternal embryonic leucine zipper kinase (MELK) for its implications in hyper-proliferative myeloma and analyzed the activity of the MELK inhibitor OTSSP167 both in vitro and in vivo. MELK was found to be significantly overexpressed in the proliferative subgroup of myeloma. This finding translated into poor overall survival in patients with high vs. low MELK expression. Enrichment analysis of upregulated genes in myeloma cells of MELKhigh patients confirmed the strong implications in myeloma cell proliferation. Targeting MELK with OTSSP167 impaired the growth and survival of myeloma cells, thereby affecting central survival factors such as MCL-1 and IRF4. This activity was also observed in the 5TGM.1 murine model of myeloma. OTSSP167 reduced bone marrow infiltration and serum paraprotein levels in a dose-dependent manner. In addition, we revealed a strong link between MELK and other proliferation-associated high-risk genes (PLK-1, EZH2, FOXM1, DEPDC1) and MELK inhibition also impaired the expression of those genes. We therefore conclude that MELK is an essential component of a proliferative gene signature and that pharmacological inhibition of MELK represents an attractive novel approach to overcome the poor prognosis of high-risk patients with a proliferative expression pattern.

Introduction The implementation of novel treatment opportunities have continuously improved the outcome of multiple myeloma (MM) patients throughout the last decades.1 However, clinical progress is mainly based on superior outcome in standard-risk patients, while the outcome in high-risk patients is still limited.2,3 Deciphering gene networks and drug candidates in high-risk MM, in order to improve the prognosis of all MM patient subgroups, remains a major task. Common classifications use tumor load and the presence of fluorescence in situ hybridization (FISH)-determined cytogenetic aberrations to define high-risk patients.4 More sophisticated methods include flow cytometry and gene expression profiling (GEP) to characterize patients with poor prognosis.5–10 The latter enabled the classification of MM into distinct GEP-defined subgroups.10 These subgroups are typically linked to the cytogenetic profile of MM (presence of distinct immunoglobulin heavy chain [IgH] translocations or hyperdiploidy). However, GEP studies also elucidated a proliferation-associated subtype.10 GEP-defined haematologica | 2018; 103(2)

Correspondence: heinz.ludwig@aon.at/jo.caers@chu.ulg.ac.be

Received: May 17, 2017. Accepted: October 27, 2017. Pre-published: November 9, 2017. doi:10.3324/haematol.2017.172973 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/325 ©2018 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.

325

A. Bolomsky et al.

myeloma with a proliferative character is strongly associated with high-risk scores and, consequently, poor prognosis.8–10 Importantly, the outcome of this patient subgroup remains poor with current treatment strategies and thus requires the implementation of more specialized treatment approaches to improve survival rates. Recent efforts have identified several key proliferative genes in MM. Among others, aurora kinase A (AURKA),11 polokinase-1 (PLK1),12,13 pituitary tumor transforming gene 1 (PTTG1)14 and DEP domain containing 1a (DEPDC1A)15 overexpression has been reported in proliferative MM, and linked to poor prognosis. Targeting of these genes impaired the growth and survival of MM cells, but their functional relevance for the proliferative character in MM is unclear. Moreover, information about interactions or the hierarchy of individual candidate genes is limited at the moment. In this context, FOXM1 was recently reported to be a putative driver in high-risk MM.16 A close relationship between FOXM1, CDK6 and NEK2 suggested a functional role for this transcription factor in promoting high-risk disease. CDK6 and NEK2 are transcriptional targets of FOXM1, and co-regulation of FOXM1 with these genes was linked to poor outcome. In addition, physical interaction between CDK6 and FOXM1 was suggested to further promote FOXM1-mediated gene transcription.16 However, the central drivers of proliferation-associated high-risk MM remain undiscovered and clinical grade inhibitors for many recently characterized target genes (e.g., FOXM1) are missing. MELK, a serine/threonine kinase with strong implications in cell cycle regulation,17,18 was identified as an upstream regulator of FOXM1 in solid and hematological malignancies.19,20 MELK plays a functional role during cell cycle progression via a direct interaction with CDC25B and co-localization with key proteins such as cyclin B1 and CDK1.17 Overexpression of MELK as well as an association between MELK levels and poor prognosis has been reported in various malignancies.21–26 MELK was shown to play a role in the proliferation and survival of malignant cells and to support the growth of cancer stem cells.27–29 Mechanistically, MELK was found to regulate FOXM1 mediated expression of mitotic genes in a PLK1-dependent manner in glioblastoma and to induce EZH2 expression in irradiation-resistant glioma stem cells.19,30 More recent studies revealed additional MELK targets (e.g., DEPDC1), and demonstrated disruption of the MELK-associated gene network by using the MELK inhibitor OTSSP167 in solid and hematological malignancies.20,31,32 These reports placed MELK upstream of several genes independently linked to high-risk myeloma, including FOXM1, EZH2, PLK1 and DEPDC1.12,13,15,16,33 Considering the availability of a MELK small molecule inhibitor (OTSSP167) already undergoing clinical testing,34 we aimed to analyze the role of MELK in high-risk MM.

Methods See the Online Supplementary Materials and Methods for a description of the techniques.

Cells Human multiple myeloma cell lines (HMCLs) U266, KMS-12BM, OPM-2, NCI-H929, SK-MM-1, RPMI8226, MM.1S, and MM.1R as well as immortalized bone marrow (BM) mesenchymal 326

stromal cells (kindly provided by Dr. Dario Campana, St. Jude Children's Research Hospital, Memphis, TN, USA) were cultivated as previously described.35 5TGM.1GFP+ cells (kind gift of Dr. G. Mundy, Vanderbilt University, Nashville, TN, USA) and HEK293T cells were maintained in Dulbecco's modified eagle medium (DMEM), supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine and 1% penicillin/streptomycin.

Cytotoxicity and colony formation assays Cytotoxicity and colony formation assays were performed using Cell Counting Kit 8 (Sigma-Aldrich) and MethoCult Classic methylcellulose based medium (Stem Cell Technologies) as described previously.35

Flow cytometry All assays were performed according to the manufacturer’s instructions as described previously.35 Analyses were performed on a FACScan and FACS Canto II (BD Biosciences).

In vivo study For studies in the murine 5TGM.1 myeloma model, OTSSP167 was dissolved in 0.5% methylcellulose (Sigma-Aldrich) and administered by oral gavage at different dose levels. For every experimental cohort, mice were randomly divided into a naïve group (n=5, healthy controls), a vehicle group (n=10, myelomabearing mice receiving vehicle solution) and a treated group (n=10, myeloma-bearing mice receiving OTSSP167). These experiments were performed as previously described.36 The Institutional Animal Care and Use Committee (ICACUC) approval number is 1336. The accreditation number from the Belgian government is LA16100002/LA2610359.

Statistical analysis Patient groups exhibiting higher or lower target gene expression were defined with the maximally selected rank statistics, implemented in the maxstat R package. Statistical significance of differences in overall survival (OS) was calculated by the log-rank test, and survival curves were plotted using the Kaplan-Meier method. Pearson correlations were calculated using R´s cor.test function. For the analysis of in vitro and in vivo experiments, a two-tailed unpaired t-test was performed for the comparison of 2 means and one-way ANOVA followed by a Tukey’s post hoc test for comparison of multiple means by Prism 5 (GraphPad Software Inc., La Jolla, CA, USA). P-values <0.05 were considered to be statistically significant. Drug combinations were analyzed with CompuSyn software. Combination index (CI) values <0.85, 0.85-1.15, and >1.15 were interpreted as synergistic, additive, and antagonistic drug activity, respectively. All graphs represent the mean ±standard deviation of at least three independent experiments performed in triplicates unless otherwise indicated.

Results MELK expression is elevated in proliferation-associated high-risk myeloma and linked to poor outcome To study the clinical relevance of MELK in MM, we analyzed MELK gene expression levels in publically available GEP datasets. No significant difference was observed between gene expression levels of healthy donor bone marow plasma cells (BMPCs), monoclonal gammopathy of undetermined significance (MGUS) and smoldering multiple myeloma (SMM) patient cells, but we noted a stepwise increase from MGUS/SMM cells to newly diagnosed and relapsed myeloma (Figure 1A). Analysis of MELK expression in distinct GEP-defined subgroups haematologica | 2018; 103(2)

MELK in multiple myeloma

(according to Zhan et al.10) revealed significant overexpression of MELK in newly diagnosed patients categorized into the proliferation (PR) subgroup of MM (Figure 1B). Consequently, high MELK expression levels were associated with poor outcome in patients treated within the total therapy 2 (median OS not reached vs. 81.47 months, P=0.01), total therapy 3 (median OS not reached, P<0.0001) as well as bortezomib- and/or dexamethasonebased protocols (median OS 21.1 months vs. 11.2 months, P=0.02) (Figure 1C). A similar association was noted with PANP-defined detectable and absent MELK expression as cut-off (Online Supplementary Figure S1). Moreover, comparison of MELK expression levels in MM patients at baseline vs. relapse indicated significant MELK upregulation in CD138-purified BM cells of relapsed patients, suggesting selection of MELKhigh MM cells or increasing MELK expression upon treatment which could be implicated in drug resistance (Figure 1D). To further strengthen the link

between MELK and high-risk disease we analyzed MELK expression levels using independent GEP-datasets which contained samples from patients with plasma cell leukemia (PCL). This clearly demonstrated an upregulation of MELK in PCL compared to MM, underlining the strong association between MELK expression and aggressive disease (Figure 1E,F).

MELK expression is strongly associated with cell cycle regulation In order to confirm the association of MELK with proliferation in MM, we analyzed GEP data (GSE24080) of newly diagnosed MM patients (n=551) with high vs. low levels of MELK. This depicted 266 upregulated and 5 downregulated probe sets representing 235 genes (minimum fold-change >2) in patients with high compared to low MELK expression (Online Supplementary Table S1). High MELK expression levels were associated with an ele-

Figure 1. MELK is overexpressed in proliferation-associated myeloma and linked to poor outcome. (A) Analysis of publically available GEP data demonstrated significant overexpression of MELK in CD138+ purified cells of MM patients compared to MGUS and SMM patients as well as healthy donor bone marrow plasma cells (BMPCs). (B) Analysis of MELK expression in distinct GEP-defined subgroups revealed significant overexpression in the proliferation (PR)-associated subgroup of MM. CD1: CCND1 group; CD2: CCND2 group; HY: hyperdiploid group; LB: low bone disease group; MF: Maf/MafB group; MS: MMSET group; MY: myeloid signature group. (C) High MELK expression was associated with poor outcome in newly-diagnosed patients treated within the total therapy 2 and 3 protocols (GSE24080) as well as relapsed and/or refractory patients (GSE9782) treated with bortezomib or dexamethasone. (D) MELK expression was elevated at relapse compared to baseline in patients treated within the TT2 (n=127 and n=343, respectively) and TT3 (n=29 and n=453, respectively) protocols as well as other treatment strategies (n=98). (EF) MELK expression was significantly elevated in CD138+ purified cells of patients suffering from PCL compared to BMPC, MGUS and MM cell samples. Horizontal lines indicate geometric mean with 95% confidence interval. *P<0.05, **P<0.01, ***P<0.001. PCL: plasma cell leukemia; MGUS: monoclonal gammopathy of undetermined significance; MM: multiple myeloma; SMM: smoldering multiple myeloma; ND: normal donor; TT: total therapy.

haematologica | 2018; 103(2)

327

A. Bolomsky et al.

vated transcription of several genes implicated in cell cycle regulation, such as CDK1, CCNB1, CCNB2, AURKA, KIF11, or BUB1B. This was confirmed by MetaCore enrichment analysis. The top-10 GO processes, pathway maps and process networks demonstrated a significant enrichment for cellular processes involved in cell cycle regulation (Online Supplementary Table S2). In brief, these results demonstrate a significant association of MELK

with proliferation-associated high-risk myeloma, and therefore encouraged pre-clinical testing of MELK as a novel therapeutic target in MM.

Targeting of MELK impairs the growth and survival of myeloma cells In line with their proliferative character,9 MELK messenger ribonucleic acid (mRNA) and protein expression was

Figure 2. Targeting of MELK impairs myeloma cell growth. (A) MELK protein and mRNA expression in HMCLs. (B) KMS-12-BM and MM.1S cells transduced with MELK specific shRNA show significantly impaired cell growth compared to cells transduced with control vector carrying scrambled shRNA. *P<0.05, ***P<0.001. (C) Treatment with OTSSP167 reduces MELK protein expression in HMCLs in a dose-dependent manner. Viability of (D) HMCLs and (E) primary MM cells 96 hours posttreatment with OTSSP167. (F) Viable primary MM cells were assessed by Annexin V/7-AAD staining 72 hours posttreatment with OTSSP167 in the presence (coculture) or absence (mono-culture) of BSMCs.

328

haematologica | 2018; 103(2)

MELK in multiple myeloma

detected in all HMCLs analyzed (Figure 2A). The targeting of MELK via short hairpin (sh)RNA significantly impaired the growth of KMS-12-BM and MM.1S cells (Figure 2B). We therefore continued to study the impact of MELK inhibition on MM cells using a small molecule inhibitor of MELK (OTSSP167). Treatment with OTSSP167 leads to destabilization of MELK and a subsequent loss of MELK

protein levels.31 Accordingly, OTSSP167 downregulated MELK protein levels 24 hours posttreatment and reduced the viability of all tested HMCLs (median IC50: 10.2 nM, range: 7.6 â&#x20AC;&#x201C; 27.1 nM) (Figure 2C,D). Moreover, OTSSP167 showed similar activity in six out of seven primary MM cell samples obtained from patients with heavily pre-treated disease (Figure 2E; for patient characteristics see Online

Figure 3. OTSSP167 impairs the cell cycle at the G2/M phase and induces apoptosis in MM cells. (A) Transcript levels of MM related growth and survival genes 5 hours posttreatment with OTSSP167 at the indicated concentrations. Gene expression levels are displayed relative to the control (0.1% DMSO). (B) Cell cycle distribution of HMCLs 48 hours posttreatment with either 0.1% DMSO (control) or OTSSP167. (C-E) Induction of apoptosis was verified by (C) Annexin V/7-AAD staining, (D) loss of the mitochondrial membrane potential, and (E) increased levels of cleaved PARP. (F) Representative Western blot images of HMCLs 24 hours posttreatment with either 0.1% DMSO (control) or OTSSP167. (G) OTSSP167 induces apoptosis in MM cells in the presence of BMSCs. (H) OTSSP167 inhibited colony formation of HMCLs. Images are representative for three independent experiments. *P<0.05, **P<0.01, and ***P<0.001 compared to control.

haematologica | 2018; 103(2)

329

A. Bolomsky et al.

Supplementary Table S3). Due to the short-lived viability of in vitro propagated primary MM cells, we also analyzed the activity of OTSSP167 in co-culture with bone marrow stromal cells (BMSCs). In line with the reported supportive role of BMSCs this demonstrated a pro-survival effect on the viability of primary MM cells. Importantly, OTSSP167 completely abrogated the protective effect of BMSCs and eradicated viable MM cells obtained from patients with PCL or refractory MM (Figure 2F). In contrast, OTSSP167 displayed only a minor impact on the via-

bility of human peripheral blood mononuclear cells (PBMCs) or BMSCs at effective anti-MM concentrations (median IC50: 726 nM) (Online Supplementary Figure S2). Inhibition of MELK was accompanied by a rapid downregulation of central myeloma genes. In line with the proposed involvement of MELK in the G2/M phase of the cell cycle we observed reduced gene expression levels of CCNB1, AURKA and PLK1 5 hours posttreatment with OTSSP167 (Figure 3A). A significant correlation of MELK expression levels and those of CCNB1 (R=0.82,

Figure 4. OTSSP167 displays synergistic activity with IMiDs and dexamethasone. (A) HMCLs were treated with OTSSP167 in combination with IMiDs (lenalidomide, pomalidomide) or dexamethasone for 96 hours at the indicated concentrations. CI values were determined with CompuSyn. CI values <0.85, 0.85-1.15, or >1.15 indicate synergistic (*), additive (+) or antagonistic (#) drug activity, respectively. (B) The activity of OTSSP167 in combination with dexamethasone plus lenalidomide or pomalidomide is compared to the corresponding monodrugs and dual-combinations. All data points of the triple combination in all three cell lines displayed strong synergism (CIs< 0.5 and 0.3 for lenalidomide and pomalidomide containing treatments, respectively; data not shown). OTS: OTSSP167; IMiDs: immunomodulatory drugs; Len: lenalidomide; Pom: pomalidomide; Dex: dexamethasone; CI: combination index.

330

haematologica | 2018; 103(2)

MELK in multiple myeloma

P<0.00001), AURKA (R=0.70, P<0.00001) and PLK1 (R=0.38, P<0.00001) was noted in publically available GEP data (GSE24080) (data not shown). In addition, we observed a decreased expression of the prominent myeloma survival factors MCL-1 and IRF4 (Figure 3A). These findings translated into a significant accumulation of cells in the G2/M phase of the cell cycle as well as induction of apoptosis 48 hours and 72 hours posttreatment, respectively (Figure 3B,C). The latter was verified by a significant increase of AnnexinV/7-AAD positive cells (Figure 3C) and associated with a loss of the mitochondrial membrane potential, detection of cleaved PARP and cleaved caspase3 (Figure 3D-F). Moreover, decreased expression of IRF4 and MCL-1 translated into reduced protein levels 24 hours posttreatment (Figure 3F). Importantly, the anti-myeloma activity of OTSSP167 was upheld in the presence of BMSCs. Similar frequencies of apoptotic cells were observed in co-cultures compared to mono-cultures (Figure 3C and 3G). We also observed reduced clonogenic growth in OTSSP167-treated HMCLs (Figure 3H). This additionally suggests an impact of MELK inhibition on tumor propagating cells.

OTSSP167 displays strong synergism with immunomodulatory drugs and dexamethasone To examine a potential impact of OTSSP167 on the activity of established anti-myeloma drugs, we performed drug combination studies with immunomodulatory drugs (IMiDs; lenalidomide, pomalidomide), proteasome inhibitors (PIs; bortezomib, carfilzomib), dexamethasone, melphalan and bendamustine. These experiments demonstrated synergistic activity (CI<0.85) of OTSSP167 with IMiDs and dexamethasone, while combination studies with PIs displayed varying results. In detail, the combination of OTSSP167 with IMiDs displayed synergistic or additive drug activity in 22 of 24 combinations tested. Median CI values for lenalidomide and pomalidomide were 0.76 (range: 0.49-1.29) and 0.48 (range: 0.12-1.12), respectively. Strong synergistic activity of OTSSP167 was also observed in combination with dexamethasone. 11 out of 12 combinations were synergistic (one additive), with CI values ranging from 0.20-1.12 (median 0.43) (Figure 4A). In contrast, 13/21 evaluable combinations with PIs displayed antagonistic drug activity. Mainly additive effects were observed when OTSSP167 was com-

Figure 5. OTSSP167 impairs myeloma cell growth in vivo. (A) In vitro evaluation of OTSSP167 in murine 5TGM.1 myeloma cells reduced MELK protein levels and cell viability in a dose-dependent manner 24 hours and 72 hours posttreatment, respectively. This was accompanied by a significant increase of cells in the G2/M phase of the cell cycle 24 hours after treatment initiation and accumulation of apoptotic cells 72 hours posttreatment (P<0.001). (B) OTSSP167 treatment schedule in the 5TGM.1 murine model of myeloma. Treatment of myeloma bearing mice with OTSSP167 led to a dose-dependent reduction of (C) BMPC infiltration rate, (D) spleen weight and (E) serum IgG2b levels. In addition, the paraplegia score (based on presence and severity of paraplegia, altered posture and diminished activity) of OTSSP167-treated mice significantly improved compared to vehicle-treated mice (F). Box plots represent median (horizontal line) with min-max whiskers. **P<0.01 and ***P<0.001 compared to vehicle-treated control mice. BM: bone marrow; MM: multiple myeloma.

haematologica | 2018; 103(2)

331

A. Bolomsky et al.

bined with the alkylating drugs melphalan (median CI: 0.86, range: 0.34-1.27) and bendamustine (1.03, 0.63-1.62) (Online Supplementary Figure S3). We further confirmed the strong synergism between OTSSP167 and IMiDs as well as dexamethasone using a wider range of drug concentrations. This corroborated our results as OTSSP167 showed consistent synergism with IMiDs and dexamethasone independent of the concentrations used (Online Supplementary Table S4). Finally, we

examined whether OTSSP167 is synergistic in combination with lenalidomide/pomalidomide plus dexamethasone. OTSSP167 displayed strong synergism with this well-established treatment regimen (median CI of OTSSP167 with lenalidomide-dexamethasone: 0.15, range: 0.02-0.46; pomalidomide-dexamethasone: 0.09, range: 0.003-0.28). In spite of the potent combinatorial effect of IMiDs-dexamethasone, OTSSP167 further improved the efficacy of this combination; in particular in

332

Figure 6. MELK is an essential component of a proliferationassociated high-risk network. (A) Gene expression levels of PLK1, EZH2, FOXM1 and DEPDC1 are significantly elevated in the proliferation (PR)-associated subgroup of myeloma (*P<0.05, **P<0.01 and ***P<0.001 compared to all other subgroups) and show a significant correlation with MELK expression levels. Horizontal lines indicate geometric mean with 95% confidence interval. (B) Gene expression levels of PLK1, EZH2, FOXM1 and DEPDC1 in HMCLs 24 hours posttreatment with OTSSP167. *P<0.05, **P<0.01 and ***P<0.001 compared to control (0.1% DMSO). (C, D) Protein expression of PR-associated high-risk genes (C) 24 hours posttreatment with OTSSP167 or (D) in cells transduced with a MELK specific shRNA. (E) The five genes, MELK, PLK1, EZH2, FOXM1 and DEPDC1 (marked in red) were used to reveal an underlying network of genes using GeneMANIA. The derived network included genes also found to be overexpressed in patients with high MELK expression (marked in yellow) as well as genes which are among the top 50 overexpressed genes of the PR subgroup of myeloma (marked in blue). (F) Overall survival of newly-diagnosed MM patients treated within the total therapy 2 or 3 protocols based on MELK, PLK1, EZH2, FOXM1 and DEPDC1 expression. Patients were grouped according to overexpression of 0, 1-2 and â&#x2030;Ľ3 of these PR-associated high-risk genes. CD1: CCND1 group; CD2: CCND2 group; HY: hyperdiploid group; LB: low bone disease group; MF: Maf/MafB group; MS: MMSET group; MY: myeloid signature group.

haematologica | 2018; 103(2)

MELK in multiple myeloma

OPM-2 and NCI-H929 cells at low doses with less pronounced activity of lenalidomide/pomalidomide plus dexamethasone (Figure 4B).

OTSSP167 impairs myeloma cell growth in vivo Treatment of murine 5TGM.1 MM cells with OTSSP167 demonstrated anti-myeloma activity similar to the effects observed in HMCLs. MELK transcript levels were in the range of HMCLs (data not shown), and we observed a dose-dependent reduction of MELK protein levels and viability 24 hours and 96 hours posttreatment, respectively. Moreover, OTSSP167 induced G2/M phase cell cycle arrest and apoptosis in 5TGM.1 cells (Figure 5A). This impact translated to the in vivo setting. OTSSP167 strongly reduced tumor burden in the 5TGM.1 murine model of myeloma. This was evidenced by several indicators including BM plasma cell infiltration, spleen weight and serum IgG2b levels (Figure 5B-E). We observed a dosedependent reduction of myeloma growth in the BM and spleen weight as well as normalization of paraprotein levels. In addition, OTSSP167 significantly enhanced the well-being of myeloma-bearing mice indicated by the absence of paraplegia and increased activity (Figure 5F).

Inhibition of MELK impairs a proliferation-associated myeloma high-risk gene signature We next sought to decipher the relationship of MELK with other genes associated with high-risk disease. Similar to MELK, PLK1 is significantly upregulated in the GEPdefined PR subgroup of MM.12 Other genes associated with poor outcome in MM and/or known functional ties with MELK include FOXM1, EZH2, and DEPDC1.15,16,37 Strikingly, all of these genes are significantly upregulated in the PR subgroup and associated with poor outcome. Correlation analysis confirmed a strong association between MELK expression levels and those of the other high-risk genes in MM cells (Figure 6A, Online Supplementary Figure S4). We therefore tested the impact of MELK inhibition on the expression of those genes. Treatment with OTSSP167 for a period of 24 hours led to a significant downregulation of PLK1 and EZH2 transcript levels (Figure 6B). Moreover, OTSSP167 downregulated PLK1, FOXM1, EZH2 and DEPDC1 protein levels in concert with MELK, suggesting a functional relationship of these genes in MM. This observation was confirmed with shRNA mediated MELK knockdown (Figure 6C,D). To better understand the genetic network of these molecules and to reveal additional network partners, we analyzed the interactions of the five high-risk genes using GeneMANIA. This demonstrated close interactions of all five candidate genes in concert with several cell cycle-associated genes (Figure 6E). The top six annotated functions of this network were all linked to cell cycle regulation (cell cycle G2/M-phase transition, G2/M-phase transition of mitotic cell cycle, condensed chromosome kinetochore, nuclear division, condensed chromosome centromeric region, mitosis; false discovery rate <1.10x10-9). Of note, nine out of 20 proposed network genes (CENPA, PRC1, CCNB1, CCNB2, MKI67, TOP2A, CDK1, NEK2, GTSE1) were also found to be elevated in newly-diagnosed MM patients with high MELK expression (Online Supplementary Table S1), and nine genes of the proposed network plus input genes are among the top 50 overexpressed genes of the GEP-defined PR subgroup (DEPDC1, EZH2, FOXM1, NEK2, CCNB1, TOP2A, PRC1, haematologica | 2018; 103(2)

CCNB2, and BIRC5) (Figure 6E). The clinical relevance of this gene network was evident in patients treated within the total therapy 2 and 3 trials. Patients with elevated expression of three or more high-risk genes (MELK, PLK1, EZH2, FOXM1, DEPDC1) displayed a significantly shorter OS compared to patients with low expression levels of these genes, or only one or two genes with high expression (median OS 37.1 months vs. not reached, P<0.0001; Figure 6F). Taken together, these data strengthen our prior results and suggest a direct impact of MELK on other proliferation-associated genes as well as a functional role in proliferation-associated high-risk MM.

Discussion The characterization of novel treatment opportunities for high-risk patients is a major task in current myeloma research. Herein, we identified MELK as a putative antiMM target in the proliferation-associated high-risk subgroup of MM and underlined its role as an attractive drug target in vitro and in vivo. MELK was significantly overexpressed in MM patients of the GEP-defined PR subgroup and associated with several genes implicated in cell cycle progression, especially at the G2/M phase. Interestingly, we also observed the overexpression of MELK at relapse and in PCL, pointing to the reported acquisition of a proliferative character post anti-myeloma therapy.10 This finding underlines the association between MELK and aggressive disease. Moreover, the upregulation of MELK or the selection of MELKhigh MM cells during the course of the disease suggests that a broad patient population could benefit from OTSSP167 treatment compared to a relatively small fraction of patients at baseline (approximately 15% of patients are categorized into the PR subgroup at diagnosis38). These initial data, therefore, strongly supported pre-clinical testing of MELK as a novel drug target for high-risk MM. The targeting of MELK with specific shRNA significantly reduced the growth of HMCLs. Hence, we studied the impact of pharmacological inhibition of MELK using OTSSP167, currently under investigation in several preclinical and clinical studies. OTSSP167 impairs the autophosphorylation of MELK leading to a subsequent degradation and loss of endogenous MELK protein.31,39 Accordingly, we observed a dose-dependent decrease of MELK protein levels 24 hours posttreatment and significant anti-MM activity. This is in line with a recent report demonstrating potent anti-myeloma activity of OTSSP167 in a panel of MM cell lines in vitro.32 Our findings corroborate the impact of OTSSP167 on the viability of myeloma cells in mono- and co-culture, demonstrate potent activity in primary MM cells from high-risk patients and also validated the impact of OTSSP167 on tumor propagating cells using an independent (colony formation) assay. We also demonstrated that the induction of apoptosis coincides with the depolarization of the mitochondrial membrane and loss of MCL-1, a key anti-apoptotic protein in HMCLs.40 This corresponds with recent findings demonstrating MCL-1 protein synthesis in a MELK-dependent manner.41 The rapid loss of MCL-1 might also explain the adverse effects of combining OTSSP167 with PIs.42 In contrast, treatment with OTSSP167 in combination with IMiDs and dexamethasone demonstrated strong synergistic activity and therefore proved the applicability and ben333

A. Bolomsky et al.

efit of this drug in combination with established anti-MM agents. The strong synergism observed with lenalidomide/pomalidomide plus dexamethasone suggests that OTSSP167 might even enhance the activity of this backbone regimen. OTSSP167 was shown to specifically target MELK in different pre-clinical in vivo models without the induction of severe events.25,31,34 Two phase I studies assessing the safety and bioavailability of OTSSP167 have recently been completed (results not published); two additional studies in breast cancer and hematological malignancies are currently ongoing. In the case of no severe toxicities, OTSSP167 should be strongly considered for clinical testing in MM, especially due to its availability as an oral agent and its positive impact on the activity of established myeloma drugs. Significantly, the anti-myeloma activity of OTSSP167 was confirmed in vivo. We observed a dose-dependent reduction of tumor growth in the murine 5TGM.1 MMmodel. Moreover, treatment displayed a significant increase in the well-being of mice even at sub-optimal anti-myeloma activity. The potent in vivo activity of OTSSP167 confirmed the role of MELK as a novel target in the presence of stromal support, and suggests that OTSSP167 inhibits tumor initiating cells in vivo. The effective targeting of myeloma stem cells was reported by blocking the MELK-associated factor EZH2.33 Thus, MELK inhibition might exert its anti-myeloma activity via affecting key players of MM pathophysiology linked to tumor propagation and high-risk disease. Recent work in glioblastoma revealed a central role for MELK in the regulation of FOXM1-mediated cellular proliferation in a PLK1-dependent manner and EZH2-mediated irradiation resistance.19,30 In addition, OTSSP167 was recently shown to target DEPDC1.31 These genes were previously described as therapy targets in MM and reported to be associated with high proliferation and/or poor survival.12,13,15,16,37 We demonstrated a significant correlation of MELK with all four candidate genes and confirmed the negative impact of these genes on the outcome of myeloma patients. Notably, the overexpression of at least three out of five proliferation-associated high-risk genes was required to unfold their poor prognostic role. This is a key finding of the current study as it clearly demonstrates that it is not a single gene, but rather networks of closely interconnected genes, which drive aggressive disease. Importantly, the targeting of MELK affected the protein synthesis of all other high-risk genes. Of note, OTSSP167 reduced the gene expression levels of PLK1 and EZH2 but not those of DEPDC1 and FOXM1. In line with this observation, a recent study demonstrated that MELK stabilizes DEPDC1 at the protein level via phosphorylation without affecting DEPDC1 transcription.31 The stable expression of FOXM1 at the mRNA level, despite modulation of MELK, is also in line with a previous report.19 Moreover, two additional studies demonstrated a decrease of FOXM1 protein expression post OTSSP167 treatment, but did not provide data about FOXM1 mRNA levels.20,32 Given that MELK stimulates FOXM1 activation in a PLK1-dependent manner, we hypothesize two pathways that lead to the observed loss of FOXM1 protein expression. First, rapid downregulation of PLK1 (evident 6 hours posttreatment) likely

334

results in reduced FOXM1 activation. Second, FOXM1 is required for mitotic progression, and a loss of activity likely contributes to the observed arrest of MM cells at the G2/M stage of the cell cycle as well as a reduced expression of MELK-FOXM1 downstream target genes such as EZH2.30,43 As FOXM1 undergoes proteasomal degradation upon mitotic arrest, a process accelerated by FOXM1 SUMOylation,43 we believe that this explains the observed discrepancy between mRNA and protein levels. However, based on recent findings, we cannot exclude potential MELK-independent (off-target) effects of OTSSP167 on these cell cycle-associated genes,44 hence further research efforts are required to reveal the exact sequence of OTSSP167 mediated anti-tumor mechanisms and the hierarchy of the MELK-associated gene network. In silico analysis placed MELK in a network with strong enrichment for key genes of the GEP-defined PR subgroup (DEPDC1, EZH2, FOXM1, NEK2, CCNB1, TOP2A, PRC1, CCNB2, and BIRC5). This suggests that MELK is an important orchestrator of a whole set of proliferative network genes, and blocking MELK appears to represent a promising future strategy to target proliferation-associated myeloma. Independent confirmation of this assumption was obtained from a recent in silico analysis of 645 patients treated within the CoMMpass trial. This study revealed MELK as top driver of a cell cycle-associated pathway in high-risk MM.45 Although further studies are needed in order to decipher the exact interaction network and hierarchy of MELK with other high-risk genes, the study herein highlights the strong relation of MELK with proliferation-associated genes and its role as a potential drug target for this group of patients. Taken together, our data reveal MELK as a novel prognostic marker of proliferation-associated high-risk myeloma and an attractive drug target in MM. The targeting of MELK demonstrated potent anti-myeloma activity, enhanced the activity of IMiDs and dexamethasone, and impaired tumor propagating cells. Furthermore, we demonstrated a strong relationship between MELK and proliferation as well as other proliferation-associated highrisk genes. This suggests that MELK, in conjunction with other high-risk genes, plays an essential role in the regulation of the proliferative phenotype of MM and that selective targeting of MELK could impair a whole network of central drivers of proliferation-associated MM. These results therefore warrant further investigation into the role of MELK in myeloma and support the clinical testing of OTSSP167 in high-risk MM. Acknowledgments The authors would like to thank Waltraud Scherbler, Katharina Postel, Isrun Bolomsky, Sophie Dubois and the GIGA flow cytometry platform for excellent technical assistance. Funding This study was supported by the Austrian Forum against Cancer, the University of Liège and the Fonds National de la Recherche Scientifique (Belgium). RH and JM are Télévie researchers. JC is a post-doctorate clinical specialist funded by the Belgian Foundation against Cancer.

haematologica | 2018; 103(2)

MELK in multiple myeloma

References 1. Kumar SK, Dispenzieri A, Lacy MQ, et al. Continued improvement in survival in multiple myeloma: changes in early mortality and outcomes in older patients. Leukemia. 2014;28(5):1122–1128. 2. Barlogie B, Mitchell A, van Rhee F, et al. Curing myeloma at last: defining criteria and providing the evidence. Blood. 2014;124(20):3043–3051. 3. Weinhold N, Heuck CJ, Rosenthal A, et al. Clinical value of molecular subtyping multiple myeloma using gene expression profiling. Leukemia. 2016;30(2):423–430. 4. Sonneveld P, Avet-Loiseau H, Lonial S, et al. Treatment of multiple myeloma with highrisk cytogenetics: a consensus of the International Myeloma Working Group. Blood. 2016;127(24):2955–2962. 5. Paiva B, Almeida J, Pérez-Andrés M, et al. Utility of flow cytometry immunophenotyping in multiple myeloma and other clonal plasma cell-related disorders. Cytometry B Clin Cytom. 2010;78(4):239–252. 6. Paiva B, Gutiérrez NC, Rosiñol L, et al. High-risk cytogenetics and persistent minimal residual disease by multiparameter flow cytometry predict unsustained complete response after autologous stem cell transplantation in multiple myeloma. Blood. 2012;119(3):687–691. 7. Paiva B, Vídriales M-B, Rosiñol L, et al. A multiparameter flow cytometry immunophenotypic algorithm for the identification of newly diagnosed symptomatic myeloma with an MGUS-like signature and long-term disease control. Leukemia. 2013;27(10):2056–2061. 8. Shaughnessy JD, Zhan F, Burington BE, et al. A validated gene expression model of high-risk multiple myeloma is defined by deregulated expression of genes mapping to chromosome 1. Blood. 2007;109(6):2276–2284. 9. Hose D, Rème T, Hielscher T, et al. Proliferation is a central independent prognostic factor and target for personalized and risk-adapted treatment in multiple myeloma. Haematologica. 2011;96(1):87– 95. 10. Zhan F, Huang Y, Colla S, et al. The molecular classification of multiple myeloma. Blood. 2006;108(6):2020–2028. 11. Hose D, Rème T, Meissner T, et al. Inhibition of aurora kinases for tailored risk-adapted treatment of multiple myeloma. Blood. 2009;113(18):4331–4340. 12. McMillin DW, Delmore J, Negri J, et al. Microenvironmental influence on pre-clinical activity of polo-like kinase inhibition in multiple myeloma: implications for clinical translation. PLoS One. 2011;6(7):e20226. 13. Evans RP, Dueck G, Sidhu R, et al. Expression, adverse prognostic significance and therapeutic small molecule inhibition of Polo-like kinase 1 in multiple myeloma. Leuk Res. 2011;35(12):1637–1643. 14. Noll JE, Vandyke K, Hewett DR, et al. PTTG1 expression is associated with hyperproliferative disease and poor prognosis in multiple myeloma. J Hematol

haematologica | 2018; 103(2)

Oncol. 2015;8:106. 15. Kassambara A, Schoenhals M, Moreaux J, et al. Inhibition of DEPDC1A, a bad prognostic marker in multiple myeloma, delays growth and induces mature plasma cell markers in malignant plasma cells. PloS One. 2013;8(4):e62752. 16. Gu C, Yang Y, Sompallae R, et al. FOXM1 is a therapeutic target for high-risk multiple myeloma. Leukemia. 2016;30(4):873–882. 17. Ganguly R, Hong C, Smith L, Kornblum HI, Nakano I. Maternal embryonic leucine-zipper kinase: key kinase for stem cell phenotype in glioma and other cancers. Mol Cancer Ther. 2014;13(6):1393–1398. 18. Nakano I, Paucar AA, Bajpai R, et al. Maternal embryonic leucine zipper kinase (MELK) regulates multipotent neural progenitor proliferation. J Cell Biol. 2005; 170(3):413–427. 19. Joshi K, Banasavadi-Siddegowda Y, Mo X, et al. MELK-dependent FOXM1 phosphorylation is essential for proliferation of glioma stem cells. Stem Cells. 2013;31(6):1051–1063. 20. Alachkar H, Mutonga MB, Metzeler KH, et al. Preclinical efficacy of maternal embryonic leucine-zipper kinase (MELK) inhibition in acute myeloid leukemia. Oncotarget. 2014;15(5):12371-12382. 21. Marie SKN, Okamoto OK, Uno M, et al. Maternal embryonic leucine zipper kinase transcript abundance correlates with malignancy grade in human astrocytomas. Int J Cancer. 2008;122(4):807–815. 22. Pickard MR, Green AR, Ellis IO, et al. Dysregulated expression of Fau and MELK is associated with poor prognosis in breast cancer. Breast Cancer Res. 2009;11(4):R60. 23. Kuner R, Fälth M, Pressinotti NC, et al. The maternal embryonic leucine zipper kinase (MELK) is upregulated in high-grade prostate cancer. J Mol Med (Berl). 2013; 91(2):237–248. 24. Du T, Qu Y, Li J, et al. Maternal embryonic leucine zipper kinase enhances gastric cancer progression via the FAK/Paxillin pathway. Mol Cancer. 2014;13:100. 25. Li S, Li Z, Guo T, et al. Maternal embryonic leucine zipper kinase serves as a poor prognosis marker and therapeutic target in gastric cancer. Oncotarget. 2016;7(5):6266– 6280. 26. Xia H, Kong SN, Chen J, et al. MELK is an oncogenic kinase essential for early hepatocellular carcinoma recurrence. Cancer Lett. 2016;383(1):85–93. 27. Nakano I, Masterman-Smith M, Saigusa K, et al. Maternal embryonic leucine zipper kinase is a key regulator of the proliferation of malignant brain tumors, including brain tumor stem cells. J Neurosci Res. 2008;86(1):48–60. 28. Nakano I, Joshi K, Visnyei K, et al. Siomycin A targets brain tumor stem cells partially through a MELK-mediated pathway. Neuro Oncol. 2011;13(6):622–634. 29. Wang Y, Lee Y-M, Baitsch L, et al. MELK is an oncogenic kinase essential for mitotic progression in basal-like breast cancer cells. Elife. 2014;3:e01763. 30. Kim S-H, Joshi K, Ezhilarasan R, et al.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42. 43.

44.

45.

EZH2 protects glioma stem cells from radiation-induced cell death in a MELK/FOXM1-dependent manner. Stem Cell Rep. 2015;4(2):226–238. Chung S, Kijima K, Kudo A, et al. Preclinical evaluation of biomarkers associated with antitumor activity of MELK inhibitor. Oncotarget. 2016; 7(14):18171– 18182. Stefka AT, Park J-H, Matsuo Y, et al. Antimyeloma activity of MELK inhibitor OTS167: effects on drug-resistant myeloma cells and putative myeloma stem cell replenishment of malignant plasma cells. Blood Cancer J. 2016;6(8):e460. Zeng D, Liu M, Pan J, et al. Blocking EZH2 methylation transferase activity by GSK126 decreases stem cell-like myeloma cells. Oncotarget. 2017;8(2):3396-3411. Chung S, Suzuki H, Miyamoto T, et al. Development of an orally-administrative MELK-targeting inhibitor that suppresses the growth of various types of human cancer. Oncotarget. 2012;3(12):1629–1640. Bolomsky A, Schlangen K, Schreiner W, Zojer N, Ludwig H. Targeting of BMI-1 with PTC-209 shows potent anti-myeloma activity and impairs the tumour microenvironment. J Hematol Oncol. 2016;9:17. Heusschen R, Muller J, Binsfeld M, et al. SRC kinase inhibition with saracatinib limits the development of osteolytic bone disease in multiple myeloma. Oncotarget. 2016;7(21):30712–30729. Hernando H, Gelato KA, Lesche R, et al. EZH2 inhibition blocks multiple myeloma cell growth through upregulation of epithelial tumor suppressor genes. Mol Cancer Ther. 2016;15(2):287–298. Weinhold N, Heuck C, Rosenthal A, et al. The clinical value of molecular subtyping multiple myeloma using gene expression profiling. 2016;30(2):423-430. Inoue H, Kato T, Olugbile S, et al. Effective growth-suppressive activity of maternal embryonic leucine-zipper kinase (MELK) inhibitor against small cell lung cancer. Oncotarget. 2016;7(12):13621–13633. Gong J-N, Khong T, Segal D, et al. Hierarchy for targeting pro-survival BCL2 family proteins in multiple myeloma: pivotal role of MCL1. Blood. 2016; 128(14):1834-1844. Wang Y, Begley M, Li Q, et al. Mitotic MELK-eIF4B signaling controls protein synthesis and tumor cell survival. Proc Natl Acad Sci U S A. 2016;113(35):9810–9815. Podar K, Gouill SL, Zhang J, et al. A pivotal role for Mcl-1 in Bortezomib-induced apoptosis. Oncogene. 2008;27(6):721-731. Myatt SS, Kongsema M, Man CW, et al. SUMOylation inhibits FOXM1 activity and delays mitotic transition. Oncogene. 2014; 33(34):4316-4329. Ji W, Arnst C, Tipton AR, et al. OTSSP167 abrogates mitotic checkpoint through inhibiting multiple mitotic kinases. PLoS One. 2016;11(4):e0153518. Gruber F, Keats JJ, McBride K, et al. Bayesian network models of multiple myeloma: drivers of high risk and durable response. 2016; 128(22):4406–4406.

335

ARTICLE

Plasma Cell Disorders

Ferrata Storti Foundation

Prognostic significance of tumor burden assessed by whole-body magnetic resonance imaging in multiple myeloma patients treated with allogeneic stem cell transplantation Jennifer Mosebach,1* Sofia Shah,2* Stefan Delorme,1 Thomas Hielscher,3 Hartmut Goldschmidt,2 Heinz-Peter Schlemmer,1 Stefan SchĂśnland,2 Ute Hegenbart2* and Jens Hillengass2*

Haematologica 2018 Volume 103(2):336-343

Department of Radiology, German Cancer Research Center, Heidelberg; 2Department of Medicine V, Multiple Myeloma Section, University of Heidelberg and 3Department of Biostatistics, German Cancer Research Center, Heidelberg, Germany 1

JM, SS, UH and JH contributed equally to this work

ABSTRACT

Correspondence: j.mosebach@dkfz-heidelberg.de

Received: July 11, 2017. Accepted: December 5, 2017. Pre-published: December 7, 2017.

llogeneic stem cell transplantation is a therapeutic option under dispute but nonetheless chosen with increasing frequency for patients suffering from multiple myeloma in Europe. To study possible predictors of survival, 79 patients were investigated using whole-body magnetic resonance imaging to assess the visible tumor burden before and after allogeneic stem cell transplantation. Statistical analysis of clinical and imaging parameters included Cox regression models and distribution of survival time estimates (Kaplan-Meier method). Log rank test was used to determine the prognostic impact of the presence of focal lesions on survival. A higher tumor burden according to the lesion count was associated with a shorter overall survival (univariable/multivariable Cox regression: 1st magnetic resonance imaging P=0.028/P=0.048; 2nd magnetic resonance imaging P=0.008/ P=0.024). Focal infiltration pattern itself seemed to be an additional adverse prognostic factor for overall survival (2nd MRI P=0.048), although no definite cut-off could be defined. Kaplan-Meier estimates at 60 months of follow up show a significant difference (Log rank P=0.04) for overall survival rates between patients with focal infiltration (32%) and those without (75%). Since this subgroup of patients may benefit from maintenance therapy, adoptive immunotherapy, or local interventions, whole-body imaging is an appropriate and highly recommendable diagnostic approach for detection of prognostically relevant lesions before and after treatment.

doi:10.3324/haematol.2017.176073

Introduction Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/336 ÂŠ2018 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.

336

In the era of emerging immunologic treatment options in hematology and oncology, one of the first approaches in this field, namely allogeneic stem cell transplantation (alloSCT) remains a widely disputed but still promising therapeutic option. Results from clinical trials comparing outcome after autologous and alloSCT in patients with multiple myeloma (MM) have been ambiguous. Whilst in some studies allogeneic transplantation in first-line therapy led to at least a superior progression-free survival (PFS), the outcome was similar or even inferior to autologous SCT in others.1-6 In the relapsed setting only one small study was able to show a superior PFS.7 Nevertheless, alloSCT is being increasingly used in some European countries, especially as second-line treatment and beyond.8 In some of the studies, survival curves revealed a plateau with patients achieving a long-term remission interpreted by some researchers as cure. Treatment-related mortality, however, is high compared to autologous transplantation, ranging from 10% in some first-line studies to up to 33% in the relapsed setting. Therefore, the International Myeloma Working Group recommended alloSCT only for eligible patients with early relapse after autologous SCT and within the setting of clinical trials.9 This considered, it will be necessary to discriminate patients who will probably benefit from the treatment from those who will not. Measurement of tumor burden as a surrogate for possible haematologica | 2018; 103(2)

Tumor burden assessed by MRI in multiple myeloma

remission or relapse is an ongoing matter of debate and is mostly indirect, as through immunoglobulin production/free light chains, CRAB criteria etc. Quantification can also be attempted through the percentage of plasma cells in bone marrow, but this is prone to sampling errors due to a focal growth pattern, as found in a significant number of patients.10 Monitoring of minimal residual disease including multi-color flow cytometry (MFC) and next generation sequencing (NGS)-based detection provides prognostic information but comes with the same problem of potential sampling error.11-13 To identify the localization of malignant foci in the organism, modern imaging techniques play a major role in diagnostics and follow up. As a consequence, these methods have been included in the updated diagnostic and response criteria.14,15 Whole-body magnetic resonance imaging (WB-MRI) is of great value in early diagnosis and detection of residual disease since it has been shown that bone marrow infiltration detected by MRI is of prognostic significance.16-18 Previously, an agreement between serological response and changes in imaging has been proven, and it has also been shown that residual focal lesions after therapy (autologous SCT) are of prognostic significance for overall survival (OS).19 Additionally, MRI has the advantage that it implies neither radiation exposure nor contrast agent administration and can therefore be performed repeatedly without harm.20 In the present study, we examined bone marrow infiltration in WB-MRI in patients before and after allogeneic SCT in addition to clinical and molecular risk constellation. Our intention was to learn whether the number of focal lesions before alloSCT or the number of persisting focal lesions thereafter is a possible predictor of survival. Finally, MRI might help to identify patients who will benefit from this treatment, or, in the post-transplant setting,those who might need additional treatment.

Table 1. Therapy regime.

Systemic therapy and transplantation st

alloSCT after 1 relapse alloSCT after > 1st relapse Auto-allo SCT upfront Auto-allo after 1st relapse Auto-allo after > 1st relapse

20 10 16 20 13 79

25.3 12.7 20.2 25.3 16.5 100

allo: allogeneic; auto: autologous; n: number; SCT: stem cell transplantation.

waived due to the retrospective nature of this evaluation. For therapy regimes see Table 1. Clinical remission status is shown in Table 2.

Imaging protocol and evaluation Diagnostic whole-body-MRI examinations were performed on 1.5T scanners (Avanto, Siemens Medical Solutions, Erlangen/Germany) including a coronal T1-weighted turbo-spinecho sequence, coronal T2-weighted, fat-attenuated turbo-inversion-recovery magnitude (TIRM), and morphologic sagittal sequences (Table 3). No contrast medium was given. Protocol details have been previously published.17,22 Focal lesions and diffuse infiltration patterns were assessed separately for each acquisition date by two radiologists, with 4 and 25 years of experience in oncologic imaging, blinded to the response, in consensus reading as previously described.19,23,24 Focal lesions were counted as myeloma infiltrates if they were hypointense in T1w as well as hyperintense in T2w fat-attenuated sequences, and >5 mm in diameter. Online Supplementary information is available.

Statistical analysis Methods Patient cohort In this single-institution-imaging-study, 79 patients were evaluated and had undergone WB-MRI before and after alloSCT between 7/2004 and 9/2013. A total of 68 were in stage III (86%) and 11 in stage II (14%) according to Durie-Salmon.21 Study approval was obtained from the institutional review board of the University of Heidelberg/Germany, and informed consent was

For the analysis of prognostic significance of parameters at 1st MRI, PFS and OS were calculated from the date of allogeneic transplantation on, including 79 patients. OS was defined as time to death from any cause, and PFS as time to progression of disease or death, whichever occurred first. For parameters at 2nd MRI, OS and PFS were counted from the landmark time point 250 days after alloSCT. Patients who were in progression even before or at the 2nd MRI were excluded from this part of the analysis. The 2nd MRI was only included if it had been performed within 250 days of alloSCT.

Table 2. Remission status at baseline and follow up.

1st MRI CR nCR VGPR PR MR SD PD

2nd MRI

10 7 8 43 3 4 4 79

12.7 8.9 10.1 54.4 3.8 5.1 5.1 100

10 4 10 15 1 8 48

% 20.8 8.3 20.8 31.2 2.1 16.7 100

CR: complete remission; MR: minimal response; MRI: magnetic resonance imaging; n: number; nCR: near complete remission; PD: progressive disease; PR: partial response; SD: stable disease; VGPR,: very good partial remission.

haematologica | 2018; 103(2)

337

J. Mosebach et al.

Figure 1. 2nd lumbar vertebra with focal remission after therapy. (A) 39year-old patient before alloSCT; *magnification shows hypointens lesion in bone marrow (B) 6 months after alloSCT no focal lesion was detected. Also, note weight loss with reduction of abdominal and subcutaneous fat tissue after therapy.

Log-rank test was used to determine the prognostic impact of the presence of focal lesions on survival, and the distribution of survival times was estimated with the Kaplan-Meier method. Survival rates for PFS and OS at 24 and 60 months following alloSCT and landmark were compared. Prognostic impact was assessed with univariable and multivariable Cox regression models. Hazard ratio for the number of focal lesions was scaled to increments of 10 lesions. Multivariable models included the following additional covariates: Durie- Salmon stage, elevated LDHlevels, age, ISS (II/III vs. I), cytogenetic high-risk, treatment autoalloSCT upfront each before alloSCT, and remission status according to the IMWG-criteria (VGPR or better) at the corresponding examination date of MRI assessment. A separate model was fitted for each time point (1st/2nd MRI) and MRI parameter (focal lesions yes vs. no/number of focal lesions). For multivariable models, missing values of clinical parameter values were imputed using multivariate imputation by chained equations as implemented in the R package based on 100 imputation runs.25 All tests were twosided; P<0.05 was considered statistically significant. All analyses were carried out with statistical software R 3.2 (R Foundation for Statistical Computing, Vienna/Austria. URL https://www.Rproject.org/).

Results Clinical parameters Mean time interval between 1st MRI and alloSCT was 29 days (range 0 -113). Response was assessed according to the guidelines of the International Myeloma Working Group adding “near complete remission” (Table 2).15 Median follow up was 83.5 months (72.0-113.6). Fifty338

Table 3. Imaging protocol.

MRI-sequence

T1-w TSE cor

T2w-TIRM cor

T2-w FLASH sag

TR/TE

627ms/11ms

3340ms/109ms

402ms/12ms

cor: coronal; FLASH: T2*-weighted fast low angle shot; MRI: magnetic resonance imaging; sag: sagittal; TE: echo time; TIRM: turbo inversion recovery magnitude; TSE: turbo spin echo; TR: repetition time. Depending on the patient’s height, acquisition included only proximal parts of the lower extremities.

seven (72.2%) patients had recurrent disease during follow up. In total 65 events for PFS and 51 deaths were observed (64.5%). Out of the 79 patients (median age 53 years/ range 29-65, 30 female/49 male patients) who had an initial MRI, 63 also completed the follow up examination, 48 of them in an acceptable time frame (< 250 days after alloSCT). Of those 48 patients, 39 had no progression until the 2nd MRI. Median time between alloSCT and 2nd MRI was 183 days (range 105- 238 days). For these 39 patients, 32 PFS events and 23 deaths were observed during follow up. Median follow up time in this subgroup was 76 months. Univariable analysis of prognostic factors is shown in Table 4. A higher stage of disease according to the classification of Durie- Salmon (III versus II) resulted in earlier progression after alloSCT (HR 3.10, P=0.016). Prognostic factors before alloSCT influencing OS include an increase of LDH level (per 100U/L increment, HR 1.4, P=0.025). A less favorable outcome was also found for patients who did not undergo auto-alloSCT up front (HR 2.45, P=0.039). Multivariate analysis (Table 4) supported an influence of the Durie- Salmon stage at 1st MRI on PFS (HR 3.48, P=0.023), and of the therapy regime on OS (HR 2.73, haematologica | 2018; 103(2)

Tumor burden assessed by MRI in multiple myeloma

Table 4. Clinical parameters influencing progression-free survival and overall survival.

Clinical Parameters PFS Age Durie-Salmon Stage: III vs. II ISS: 2/3 vs. 1 ISS: 2 vs. 1 ISS: 3 vs. 1 High LDH Increase of LDH level (per 100 U/L increment) FISH: high risk vs. low risk Status of remission at baseline: VGPR and better vs. other Status of remission at 2nd MRI: VGPR and better vs. other Therapy: other vs. auto-alloSCT upfront OS Age Durie-Salmon Stage: III vs. II ISS: 2/3 vs. 1 ISS: 2 vs. 1 ISS: 3 vs. 1 High LDH Increase of LDH level (per 100 U/L increment) FISH: high risk vs. low risk Status of remission at baseline: VGPR and better vs. other Status of remission at 2nd MRI: VGPR and better vs. other Therapy: other vs. auto-alloSCT upfront

Univariable Cox model HR (LCL-UCL) P 1.02 (0.99-1.05) 3.10 (1.23-7.77) 0.98 (0.55-1.77) 1.23 (0.66-2.31) 0.63 (0.26-1.50) 0.94 (0.49-1.82)

0.248 0.016 0.956 0.516 0.295 0.862

1.18 (0.90-1.55) 1.41 (0.78-2.57) 0.73 (0.42-1.27)

Multivariable Cox model HR (LCL-UCL) P 1.02 (0.99-1.06) 3.48 (1.19-10.16) 0.88 (0.47-1.65)

0.232 0.023 0.687

1.13 (0.57-2.26)

0.723

0.221 0.255 0.269

1.17 (0.62-2.21) 0.70 (0.37-1.33)

0.633 0.275

0.50 (0.24-1.06)

0.069

0.53 (0.17-1.69)

0.285

1.87 (0.95-3.68)

0.071

1.68 (0.79-3.55)

0.178

1.01 (0.98-1.05) 2.10 (0.76-5.85) 0.97 (0.49-1.93) 1.03 (0.49-2.15) 0.85 (0.31-2.35) 1.53 (0.72-3.26)

0.515 0.154 0.941 0.939 0.757 0.267

1.00 (0.96-1.04) 2.83 (0.78-10.24) 0.96 (0.47-1.99)

0.968 0.113 0.919

1.33 (0.58-3.08)

0.500

1.40 (1.04-1.87) 1.56 (0.79- 3.10) 1.13 (0.62-2.06)

0.025 0.202 0.698

1.23 (0.52-2.92) 1.32 (0.65-2.71)

0.634 0.443

0.99 (0.43-2.30)

0.981

0.91 (0.31-2.66)

0.861

2.45 (1.05-5.76)

0.039

2.73 (1.08-6.95)

0.035

Analysis included univariable and multivariable Cox regression model. Results for models at 1st MRI (prior to alloSCT) are given, except for Status of remission at 2nd MRI, which is based on the model at landmark. Results for multivariable model are based on the model with number of focal lesions as MRI parameter. No relevant differences in results were found when considering presence of focal lesions (yes/no) as MRI parameter instead. Deletion 17p13 and translocation t(4;14) were considered high-risk cytogenetic aberrations. The influence of translocation t(14;16) was not investigated due to a high number of missing values. The only two patients with documented t(14;16) also had del17p13. AlloSCT: allogeneic stem cell transplantation; auto: autologous; FISH: fluorescence in situ hybridization; HR: hazard ratio; ISS: international staging system; LCL: lower 95% confidence level; LDH: lactate dehydrogenase; PFS: progression-free survival; OS: overall survival; UCL: upper 95% confidence level; U/L: units per litre; VGPR: very good partial response.

P=0.035). Other factors such as age, LDH, remission status and cytogenetic risk constellation did not reach statistical significance.

MRI findings At initial imaging, focal lesions were detected in 66 out of 79 patients (83.5%), and diffuse infiltration patterns in 60 patients (76%). After alloSCT, myeloma-suspicious focal lesions were visible in 27 out of 39 patients (69.2%), none in 12 (30.8%), and 28 patients had signal alterations compatible with diffuse infiltration (71.8%). A figure with T1-weighted images of a patient with multiple lesions at various locations is included in the Online Supplementary Material. Of the 39 patients without clinical progression, 8 had no lesions in neither the baseline nor the follow up MRI scan. In 27 patients, one or more lesions were found at baseline haematologica | 2018; 103(2)

and at the follow up scan. In 4 patients, one or more lesions were present at baseline and resolved after alloSCT. An example of a focal remission is shown in Figure 1, images of a patient with progressive disease in follow up MRI is included in the Online Supplementary Material. Univariable regression analysis (Table 5) could not detect statistically significant influence of MRI findings on PFS. Statistical results for presence of one or more focal lesion after therapy and of increasing number of focal lesions at 1st and 2nd MRI suggested an effect with HR >1 but this did not reach statistical significance. A higher number of focal lesions at baseline and follow up, on the other hand, were associated with a shorter OS (Table 5, HR 1.22, P=0.028; HR 1.46, P=0.008, respectively, per 10 lesion increase). Presence of at least one focal lesion after therapy also yielded a negative prognostic 339

J. Mosebach et al.

effect on OS (Table 5, HR 2.98; P=0.048). This is also seen in the Kaplan-Meier plot for OS and presence of any focal lesion at second MRI, which is shown in Figure 2. KaplanMeier survival rates for PFS and OS at 24 and 60 months of follow up are presented in Table 6. If any focal lesion were detectable on the second MRI, the OS rate was 63% after 24 months and 32% after 60 months. If no focal lesions were detected (log-rank, P=0.04), 92% of the patients were alive after 24 months, and 75% after 60 months. Multivariable analysis supported the adverse prognostic influence on OS for increasing number of lesions (by 10) at both time points (HR 1.24, P=0.048; HR 1.56, P=0.024, respectively). Furthermore, increased risk without reaching statistical significance was found for PFS and OS considering presence of focal lesion at 1st MRI (HR 1.96, P=0.097; and HR 2.26, P=0.098), and for higher numbers at baseline (PFS, increase by 10 lesions: HR 1.21, P=0.058). Diffuse infiltration pattern showed no impact on PFS (1st MRI P=0.720, 2nd MRI P=0.699) or OS (1st MRI P=0.151, and 2nd MRI P=0.238). Patients with decreasing numbers of focal lesions (19/39) between MRIs did not have a better prognosis than other patients with focal infiltration at baseline imaging. Patients with resolving lesions showed slightly better PFS than patients with no lesions at both MRIs but without statistical significance (HR 0.42 P=0.280). No statistically significant difference was found for patients with none versus persisting lesions (HR 1.41, P=0.44), meaning radiologically stable patients.

mendation to apply this treatment only in eligible patients with early relapse after autologous SCT and within clinical trials.9 Hence, it is important to identify those patients who might benefit from alloSCT.

Prognostic significance of tumor burden The intention of the current analysis was to study the prognostic relevance of focal lesions as a measure of tumor burden in multiple myeloma in the setting of alloSCT. MRI examinations before and after allogeneic stem cell transplantation were therefore retrospectively reviewed with a long follow up. Results of univariable and multivariable analyses verified that a higher tumor load at baseline as well as follow up-MRI is of adverse prognostic significance for OS. This is supported by indirect measure-

Log-rank: P=0.04*

Discussion Given that multiple myeloma is as yet not curable in the majority of patients, not even with autologous SCT, alloSCT remains a last resort in the attempt to definitely eradicate the disease. However, a relatively high treatment-related mortality and morbidity and a still significant percentage of relapsing patients has led to the recom-

Figure 2. Kaplan-Meier graph for influence of presence or absence of focal lesions in 2nd MRI on OS. Censored patients are indicated with small vertical marks. Supplementary information is available online.

Table 5. Imaging findings.

PFS MRI 1 Presence of focal lesion Increasing number of focal lesions by 10 MRI 2 Presence of focal lesion Increasing number of focal lesions by 10 OS MRI 1 Presence of focal lesion Increasing number of focal lesions by 10 MRI 2 Presence of focal lesion Increasing number of focal lesions by 10

Univariable Cox model HR (LCL-UCL) P

Multivariable Cox model HR (LCL-UCL) P

1.56 (0.80-3.08) 1.15 (0.97-1.36)

0.195 0.099

1.96 (0.89-4.31) 1.21 (0.99-1.47)

0.097 0.058

1.83 (0.83-4.03) 1.19 (0.90-1.57)

0.131 0.223

1.40 (0.52-3.78) 1.19 (0.75-1.89)

0.506 0.457

1.96 (0.83-4.60) 1.22 (1.02-1.45)

0.123 0.028

2.26 (0.86-5.94) 1.24 (1.00-1.54)

0.098 0.048

2.98 (1.01-8.79) 1.46 (1.10-1.94)

0.048 0.008

2.79 (0.84-9.33) 1.56 (1.06-2.28)

0.095 0.024

Progression-free survival (PFS) and overall survival (OS). Univariable and multivariable analysis of hazard ratio (HR) are shown with lower and upper 95% confidence level (LCL and UCL). MRI1: 79 patients; MRI2: 39 patients.

340

haematologica | 2018; 103(2)

Tumor burden assessed by MRI in multiple myeloma

ments of tumor burden, like increasing LDH-levels or a high M-component production rate, as being included in the Durie- and Salmon Staging system, which was used for these patients at the time of recruitment. In patients who did not progress immediately after transplantation, the detection of a focal infiltration pattern (at least one focal lesion/ any lesion) in bone marrow after therapy seems to be an additional adverse factor. OS after 5 years was 75% in patients without focal lesions in second MRI, compared to 32% for patients with detectable lesions. This remarkable difference is in line with the research by Walker et al. who concluded that a higher number of focal lesions in untreated newly diagnosed patients was unfavorable for survival, although in our present study no cut-off point for number of lesions could be defined. Patients with resolving lesions resembling imaging response showed slightly better PFS without statistical significance, probably due to limited eligible patients with complete imaging. Patients with any focal lesion after therapy and especially a higher tumor load on MRI are at higher risk of progression and shorter OS, independently of molecular tumor activity. Therefore, patients might be selected for, and hopefully profit from, continuous therapy to prevent or at least delay relapse. Additionally, localized relapse has been shown to occur despite sustained molecular remission, which can be reliably detected through follow-up imaging.26

Comparison to findings in PET/CT and autologous SCT The results of the present analysis also support previous findings that residual lesions after autologous SCT are of adverse prognostic significance. This is true for MRI as well as PET/CT.2 The mentioned results have led to the implementation of imaging findings into the updated recommendations for assessment of treatment response in patients with multiple myeloma.15 The rationale behind this recommendation is also that an assessment of minimal residual disease is performed usually on bone marrow samples acquired from the iliac crest. These samples, however, might miss accumulations of malignant cells i.e., focal lesions in other parts of the body. Since alloSCT aims to cure myeloma, the achievement and therefore the assessment of the deepest possible response is crucial. Our findings in the alloSCT setting support the results of Patriarca et al., who evaluated 54 patients before and after allogeneic SCT with PET/CT and were able to show that patients with a complete remission in imaging have a sig-

nificantly longer PFS and OS than those in whom any PET-positive lesions had remained (2- year PFS: 51% versus 25%, P=0.03; 2-year OS: 81% versus 47%, P=0.001; 29). In recently published data of the IFM/DFCI Trial, PET/CT normalization before maintenance was also associated with better PFS and OS.30 Combined results so far suggest that residual disease after therapy increases the risk of relapse, as we also previously discussed for patients after autologous SCT, although results for PFS were only of borderline statistical significance in our current study.19

Role of cytogenetic risk factors and therapy regime Interestingly, some of the well-established risk factors like high risk cytogenetics by Fluorescence in situ hybridization and ISS had no prognostic effect in our cohort. Although we did observe an increased risk (HR >1) for high-risk FISH, this did not reach statistical significance. Deletion 17p13 and translocation t(4;14) were considered high-risk cytogenetic aberrations. The influence of translocation t(14;16) was not investigated due to a high number of missing values because at the time of the first diagnosis of most of the patients (beginning in 2004) FISH was not yet standard of care in our department. It has nonetheless also been shown that a possible success of alloSCT is independent of the cytogenetic risk profile.31 Furthermore, a less favorable outcome was seen in patients who did not undergo auto-alloSCT up front. Poorer outcome in the relapsed setting has been previously reported by Franssen and colleagues, who also did not see any differences in outcome for patients with high risk cytogenetics, as was the case in our own investigation.32

Limitations and future directions A limitation of the present MRI study is the limited number of cases. It must be noted, however, that in comparison to other treatment options, few patients are eligible for alloSCT and, recruited in one of the biggest myeloma centers in the world, we herein present the biggest cohort with MR imaging to date. Also, we would like to discuss the mere morphologic evaluation applied in this study, which makes it difficult to separate active tumor lesions from pre-treated lesions without residual vital cells. Our own investigations (not published) which attempt to differentiate between these types of lesions have been unsuccessful to date, and caution is advised as progression can arise from inactive cystic-transformed

Table 6. Survival rates.

Focal lesion

Follow up (time/months)

PFS Survival rates (LCL-UCL)

OS Survival rates (LCL-UCL)

yes

24 60 24 60 24 60 24 60

0.34 (0.24-0.48) 0.16 (0.09-0.29) 0.54 (0.33-0.89) 0.37 (0.18-0.77) 0.16 (0.07-0.39) 0.16 (0.07-0.39) 0.50 (0.28-0.88) 0.37 (0.17-0.83)

0.62 (0.51-0.75) 0.36 (0.26-0.51) 0.77 (0.57-1.00) 0.68 (0.47-1.00) 0.63 (0.47-0.84) 0.32 (0.18-0.56) 0.92 (0.77-1.00) 0.75 (0.54-1.00)

MRI 1

no MRI 2

yes no

Abbreviations: see Table 4.

haematologica | 2018; 103(2)

341

J. Mosebach et al.

lesions as well. Furthermore, repopulating blood-building bone marrow in vertebrae or even in long bones can morphologically resemble focal myeloma lesions and makes interpretation challenging. Functional MRI sequences such as diffusion weighted imaging (DWI) were not regularly available in this study, but are highly recommended in a scientific setting. Since the use of contrast agents is limited in myeloma patients, due to renal impairment as a potential symptom of the disease, DWI seems especially promising. It does not require contrast agents, but can still give qualitative and quantitative information about the bone marrow and has been shown to be a useful technique for detecting diffuse and multifocal marrow infiltration in patients with myeloma, with equal or higher sensitivity, when compared to PET.33,34 According to Cassou-Mounat et al. the detection rate of PET can be improved by the implementation of 18F-fluorocholine in diagnostics. In their pilot study, the recent metabolic tracer could detect more lesions compared to 18F-fluorodeoxyglucose.35 Further studies are needed, and we are currently seeking to assess the development and remission of lesions in DWI and PET in this context in our institutions. In addition to focal infiltration, diffuse bone marrow infiltration is seen in many myeloma patients. In our cohort, we could not detect an impact of a diffuse infiltration pattern on the patientsâ&#x20AC;&#x2122; outcome. Although conven-

References 1. Bruno B, Rotta M, Patriarca F, et al. A comparison of allografting with autografting for newly diagnosed myeloma. N Engl J Med. 2007;356(11):1110-1120. 2. Bjorkstrand B, Iacobelli S, Hegenbart U, et al. Tandem autologous/reduced-intensity conditioning allogeneic stem-cell transplantation versus autologous transplantation in myeloma: long-term follow-up. J Clin Oncol. 2011;29(22):3016-3022. 3. Gahrton G, Iacobelli S, Bjorkstrand B, et al. Autologous/reduced-intensity allogeneic stem cell transplantation vs autologous transplantation in multiple myeloma: longterm results of the EBMT-NMAM2000 study. Blood. 2013;121(25):5055-5063. 4. Garban F, Attal M, Michallet M, et al. Prospective comparison of autologous stem cell transplantation followed by dosereduced allograft (IFM99-03 trial) with tandem autologous stem cell transplantation (IFM99-04 trial) in high-risk de novo multiple myeloma. Blood. 2006;107(9):34743480. 5. Krishnan A, Pasquini MC, Logan B, et al. Autologous haemopoietic stem-cell transplantation followed by allogeneic or autologous haemopoietic stem-cell transplantation in patients with multiple myeloma (BMT CTN 0102): a phase 3 biological assignment trial. Lancet Oncol. 2011; 12(13):1195-1203. 6. Rosinol L, Perez-Simon JA, Sureda A, et al. A prospective PETHEMA study of tandem autologous transplantation versus autograft followed by reduced-intensity conditioning allogeneic transplantation in newly diagnosed multiple myeloma. Blood. 2008; 112(9):3591-3593.

342

tional MRI has previously been shown to be more accurate than FDG PET/CT for the detection of diffuse marrow infiltration, due to the possible reconstitution of the bone marrow after previous therapy and transplantation, pathological or therapeutically induced diffuse signal changes could not be reliably distinguished.36 Therefore, further analysis will surely be a topic of future research.

Conclusion In general, it seems that a focal infiltration pattern and an increased tumor load represented by increasing focal myeloma bone marrow lesions, shortens OS. In conclusion, we recommend imaging using whole-body MRI before and after allogeneic SCT, since patients with prognostically relevant lesions and higher tumor burden before and after treatment independently of serological response may benefit from maintenance therapy, donor lymphocyte infusions (DLI), or local interventions to consolidate remission. Funding We gratefully acknowledge the German Research Foundation (Deutsche Forschungsgemeinschaft) for Research/Grant Support (JM) as part of the CRC/Transregio 79.

7. de Lavallade H, El-Cheikh J, Faucher C, et al. Reduced-intensity conditioning allogeneic SCT as salvage treatment for relapsed multiple myeloma. Bone Marrow Transplant. 2008;41(11):953-960. 8. Sobh M, Michallet M, Gahrton G, et al. Allogeneic hematopoietic cell transplantation for multiple myeloma in Europe: trends and outcomes over 25 years. A study by the EBMT Chronic Malignancies Working Party. Leukemia. 2016; 30(10):2047-2054. 9. Giralt S, Garderet L, Durie B, et al. American Society of Blood and Marrow Transplantation, European Society of Blood and Marrow Transplantation, Blood and Marrow Transplant Clinical Trials Network, and International Myeloma Working Group consensus conference on salvage hematopoietic cell transplantation in patients with relapsed multiple myeloma. Biol Blood Marrow Transplant. 2015; 21(12):2039-2051. 10. Joshua DE. Tumor Burden. In: Berenson JR, ed. Biology and management of multiple myeloma. Totowa, NJ: Humana Press, 2004:127-136. 11. Ladetto M, Ferrero S, Drandi D, et al. Prospective molecular monitoring of minimal residual disease after non-myeloablative allografting in newly diagnosed multiple myeloma. Leukemia. 2016;30(5):12111214. 12. Paiva B, Cedena MT, Puig N, et al. Minimal residual disease monitoring and immune profiling using second generation flow cytometry in elderly multiple myeloma. Blood. 2016;127(25):3165-74. 13. Sarasquete ME, Garcia-Sanz R, Gonzalez D, et al. Minimal residual disease monitoring in multiple myeloma: a comparison between allelic-specific oligonucleotide real-time quantitative polymerase chain

14.

15.

16.

17.

18.

19.

20.

21.

reaction and flow cytometry. Haematologica. 2005;90(10):1365-1372. Rajkumar SV, Dimopoulos MA, Palumbo A, et al. International Myeloma Working Group updated criteria for the diagnosis of multiple myeloma. Lancet Oncol. 2014; 15(12):e538-e548. Kumar S, Paiva B, Anderson KC, et al. International Myeloma Working Group consensus criteria for response and minimal residual disease assessment in multiple myeloma. Lancet Oncol. 2016;17(8):e328e346. Walker R, Barlogie B, Haessler J, et al. Magnetic resonance imaging in multiple myeloma: diagnostic and clinical implications. J Clin Oncol. 2007;25(9):1121-1128. Hillengass J, Fechtner K, Weber MA, et al. Prognostic significance of focal lesions in whole-body magnetic resonance imaging in patients with asymptomatic multiple myeloma. J Clin Oncol. 2010;28(9):16061610. Mariette X, Zagdanski AM, Guermazi A, et al. Prognostic value of vertebral lesions detected by magnetic resonance imaging in patients with stage I multiple myeloma. Br J Haematol. 1999;104(4):723-729. Hillengass J, Ayyaz S, Kilk K, et al. Changes in magnetic resonance imaging before and after autologous stem cell transplantation correlate with response and survival in multiple myeloma. Haematologica. 2012; 97(11):1757-1760. Derlin T, Peldschus K, Munster S, et al. Comparative diagnostic performance of (1)(8)F-FDG PET/CT versus whole-body MRI for determination of remission status in multiple myeloma after stem cell transplantation. Eur Radiol. 2013;23(2):570-578. Durie BG, Salmon SE. A clinical staging system for multiple myeloma. Correlation of

haematologica | 2018; 103(2)

Tumor burden assessed by MRI in multiple myeloma

22.

23.

24.

25. 26.

27.

measured myeloma cell mass with presenting clinical features, response to treatment, and survival. Cancer. 1975;36(3): 842-854. Bauerle T, Hillengass J, Fechtner K, et al. Multiple myeloma and monoclonal gammopathy of undetermined significance: importance of whole-body versus spinal MR imaging. Radiology. 2009;252(2):477485. Baur A, Stabler A, Nagel D, et al. Magnetic resonance imaging as a supplement for the clinical staging system of Durie and Salmon? Cancer. 2002;95(6):1334-1345. Stabler A, Baur A, Bartl R, Munker R, Lamerz R, Reiser MF. Contrast enhancement and quantitative signal analysis in MR imaging of multiple myeloma: assessment of focal and diffuse growth patterns in marrow correlated with biopsies and survival rates. AJR Am J Roentgenol. 1996; 167(4):1029-1036. Stef van Buuren KG-O. mice: Multivariate imputation by chained equations in R. J Stat Softw. 2011;45(3):1-67. Byrne JL, Fairbairn J, Davy B, Carter IG, Bessell EM, Russell NH. Allogeneic transplantation for multiple myeloma: late relapse may occur as localised lytic lesion/plasmacytoma despite ongoing molecular remission. Bone Marrow Transplant. 2003;31(3):157-161. Bartel TB, Haessler J, Brown TL, et al. F18fluorodeoxyglucose positron emission

haematologica | 2018; 103(2)

28.

29.

30.

31.

tomography in the context of other imaging techniques and prognostic factors in multiple myeloma. Blood. 2009;114(10): 2068-2076. Zamagni E, Patriarca F, Nanni C, et al. Prognostic relevance of 18-F FDG PET/CT in newly diagnosed multiple myeloma patients treated with up-front autologous transplantation. Blood. 2011;118(23):59895995. Patriarca F, Carobolante F, Zamagni E, et al. The role of positron emission tomography with 18F-fluorodeoxyglucose integrated with computed tomography in the evaluation of patients with multiple myeloma undergoing allogeneic stem cell transplantation. Biol Blood Marrow Transplant. 2015;21(6):1068-1073. Moreau P, Attal M, Caillot D, et al. Prospective evaluation of magnetic resonance imaging and [18F]fluorodeoxyglucose positron emission tomography-computed tomography at diagnosis and before maintenance therapy in symptomatic patients with multiple myeloma included in the IFM/DFCI 2009 Trial: Results of the IMAJEM Study. J Clin Oncol. 2017; 35(25):2911-2918. Kroger N, Badbaran A, Zabelina T, et al. Impact of high-risk cytogenetics and achievement of molecular remission on long-term freedom from disease after autologous-allogeneic tandem transplantation in patients with multiple myeloma.

32.

33.

34.

35.

36.

Biol Blood Marrow Transplant. 2013;19(3): 398-404. Franssen LE, Raymakers RA, Buijs A, et al. Outcome of allogeneic transplantation in newly diagnosed and relapsed/refractory multiple myeloma: long-term follow-up in a single institution. Eur J Haematol. 2016; 97(5):479-488. Pawlyn C, Fowkes L, Otero S, et al. Wholebody diffusion-weighted MRI: a new gold standard for assessing disease burden in patients with multiple myeloma? Leukemia. 2016;30(6):1446-1448. Sachpekidis C, Mosebach J, Freitag MT, et al. Application of (18)F-FDG PET and diffusion weighted imaging (DWI) in multiple myeloma: comparison of functional imaging modalities. Am J Nucl Med Mol Imaging. 2015;5(5):479-492. Cassou-Mounat T, Balogova S, Nataf V, et al. 18F-fluorocholine versus 18F-fluorodeoxyglucose for PET/CT imaging in patients with suspected relapsing or progressive multiple myeloma: a pilot study. Eur J Nucl Med Mol Imaging. 2016; 43(11):1995-2004. Zamagni E, Nanni C, Patriarca F, et al. A prospective comparison of 18F-fluorodeoxyglucose positron emission tomography-computed tomography, magnetic resonance imaging and whole-body planar radiographs in the assessment of bone disease in newly diagnosed multiple myeloma. Haematologica. 2007;92(1):50-55.

343

ARTICLE

Coagulation & Its Disorders

Ferrata Storti Foundation

Clustered F8 missense mutations cause hemophilia A by combined alteration of splicing and protein biosynthesis and activity Irving Donadon,1,2 John H. McVey,3 Isabella Garagiola,4 Alessio Branchini,1 Mimosa Mortarino,4 Flora Peyvandi,4,5 Francesco Bernardi1 and Mirko Pinotti1,6 Department of Life Sciences and Biotechnology, University of Ferrara, Italy; 2Human Molecular Genetics, International Centre for Genetic Engineering and Biotechnology, Trieste, Italy; 3School of Bioscience & Medicine, University of Surrey, Guildford, UK; 4 Angelo Bianchi Bonomi Hemophilia and Thrombosis Center, Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico and Fondazione Luigi Villa, Milan, Italy; 5 Department of Pathophysiology and Transplantation, University of Milan, Italy and 6 Laboratorio per le Tecnologie delle Terapie Avanzate, University of Ferrara, Italy 1

Haematologica 2018 Volume 103(2):344-350

ABSTRACT

Correspondence: pnm@unife.it

Received: August 10, 2017. Accepted: November 15, 2017. Pre-published: November 23, 2017. doi:10.3324/haematol.2017.178327 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/344

issection of pleiotropic effects of missense mutations, rarely investigated in inherited diseases, is fundamental to understanding genotype-phenotype relationships. Missense mutations might impair mRNA processing in addition to protein properties. As a model for hemophilia A, we investigated the highly prevalent F8 c.6046c>t/p.R2016W (exon 19) mutation. In expression studies exploiting lentiviral vectors, we demonstrated that the amino acid change impairs both Factor VIII (FVIII) secretion (antigen 11.0±0.4% of wildtype) and activity (6.0±2.9%). Investigations in patients’ ectopic F8 mRNA and with minigenes showed that the corresponding nucleotide change also decreases correct splicing to 70±5%, which is predicted to lower further FVIII activity (4.2±2%), consistently with patients’ levels (<1-5%). Masking the mutated exon 19 region by antisense U7snRNA supported the presence of a splicing regulatory element, potentially affected by several missense mutations causing hemophilia A. Among these, the c.6037g>a (p.G2013R) reduced exon inclusion to 41±3% and the c.6053a>g (p.E2018G) to 28±2%, similarly to a variant affecting the 5’ splice site (c.6113a>g, p.N2038S, 26±2%), which displayed normal protein features upon recombinant expression. The p.G2013R reduced both antigen (7.0±0.9%) and activity (8.4±0.8%), while the p.E2018G produced a dysfunctional molecule (antigen: 69.0±18.1%; activity: 19.4±2.3%). In conclusion, differentially altered mRNA and protein patterns produce a gradient of residual activity, and clarify genotype-phenotype relationships. Data detail pathogenic mechanisms that, only in combination, account for moderate/severe disease forms, which in turn determine the mutation profile. Taken together we provide a clear example of interplay between mRNA and protein mechanisms of disease that operate in shaping many other inherited disorders.

344

Introduction It is widely accepted that mutations can have pleiotropic effects that define the overall pathological phenotype.1,2 However, understanding in detail the contribution of each of these effects is complex and compromises clarification of genotype-phenotype relationships. Missense mutations are the second most common cause of severe hemophilia A (HA)3 (http://www.eahad-db.org/; http://factorviiidb.org/) and the most frequent cause of the rare coagulation factor deficiencies,4,5 as well as of other human diseases, and are known to impair several properties of mutated proteins.6,7 However, coding sequences overlap with exonic pre-mRNA splicing regulatory elements and thus missense changes may also affect premRNA processing.8,9 haematologica | 2018; 103(2)

Pleiotropic effects of HA-causing mutations

To quantitatively evaluate the interplay between the effects of missense mutations on mRNA and protein biosynthesis and their impact on HA coagulation phenotype, as a relevant model we initially investigated the F8 p.R2016W (c.6046c>t) missense mutation (HGVS numbering; reference sequences NP_000123.1 and NM_000132.3 for protein and cDNA, respectively).10 This mutation in exon 19 is one of the most highly reported amino acid substitutions (>60 patients), particularly in Italy.11,12 It is suggested to affect splicing13 and candidate to impair FVIII protein structure, but has never been characterized. Here, we evaluated the potential pleiotropic effects by extensive expression studies and dissected the effects on FVIII protein secretion and activity as well as on exon 19 definition. Intriguingly, several missense mutations clustered in the p.R2016W region impacted on each mechanism to a different extent, consistent with the variable degree of HA severity.

made by overlapping PCR15 and subsequent cloning through the MluI and SbfI restriction sites. The coFVIII variants were cloned in the lentivirus plasmid pLNT backbone (Figure 1A) to produce lentiviruses (LV).14 CHO cells16 were transduced with LV-rFVIII at MOI 2. Cells and media were harvested 72 hours (h) post infection to evaluate lentiviral copy number, FVIII antigen (ELISA) and chromogenic activity levels, as described.14

Creation of F8 minigenes and splicing assays The genomic cassette consisting of F8 exon 19 (117 bp) and the surrounding intron 18 (343 nucleotides) and intron 19 (332 nucleotides) sequences was amplified from DNA of a normal subject and cloned in the pTB vector17 through NdeI restriction sites (Figure 2) within primers. Mutations were introduced by mutagenesis to create minigene variants. Expression vectors for the U1snRNA and U7snRNA variants were created as described.18,19 HepG2 or HEK293 cells were transfected9,20 with F8 minigenes alone or equimolar amounts of pU1snRNA/pU7snRNAs to evaluate exon 19 inclusion by RT-PCR with primers alpha 2,3 and Bra2.17

Methods Lentiviral vectors and FVIII protein expression studies

Evaluation of FVIII plasma levels and F8 splicing profile in HA patients

Mutations were introduced by site-specific mutagenesis9 into the codon-optimized human FVIII cDNA lacking B-domain (coFVIII).14 The F8 exon 19 deleted cDNA fragment (FVIIIÎ&#x201D;19) was

FVIII activity was evaluated by the 2-stage amidolytic assay (Coamatic Factor VIII; Chromogenix, Lexington, MA, USA) and the 1-stage coagulation assay as described.11

Figure 1. The splicing-defective F8 missense mutations differentially impair Factor VIII (FVIII) protein secretion and function. (A) Schematic representation (upper part) of the lentiviral vector backbone harboring the codon-optimized cDNA of human FVIII lacking the B-domain (coFVIII) and sequence of the affected region and of the investigated mutations (middle part). The alignment of FVIII sequence across species is reported (lower part) together with affected residues (red). (B) Secreted FVIII antigen (upper) and co-factor activity (lower) levels of rFVIII variants expressed as % of rFVIIIwt. Secreted protein levels were normalized on virus copy number per cell determined by qPCR.14 Results are reported as meanÂąStandard Deviation from three independent experiments. (C) Structure of the human FVIII (PDB: 2R7E). Domain overview (A1-A2-A3-C1-C2 domains, inset) and interface between the A1 (blue) and A3 (white) domains (ribbon). The clustered residues under investigation (HGVS numbering) are represented by ball and stick. The R2016 residue is shown in space-filling.

haematologica | 2018; 103(2)

345

I. Donadon et al. Ectopic mRNA was isolated from fresh leukocytes21 and evaluated by RT-PCR using primers in F8 exon 17 and 22. Studies of patients’ sample were approved by the Review Board of the Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico of Milan, and patients gave their written informed consent. Sequences of all primers used are listed in the Online Supplementary Appendix.

Results The impact of the F8 c.6046c>t/p.R2016W mutation on FVIII protein and F8 mRNA splicing were extensively analyzed using recombinant technologies.

The R2016W substitution impairs both FVIII secretion and function So far, few HA-causing missense variants have been characterized22-26 because of the very low secretion efficiency of recombinant full-length rFVIII, which makes the evaluation of the residual levels associated with variants very difficult. To overcome this limitation, we exploited the lentiviral-mediated delivery of the expression cassette consisting of the codon-optimized FVIII cDNA lacking the B domain (Figure 1A).14 In this experimental system, the rFVIII-2016W missense variant (Figure 1B) was secreted in medium at reduced levels (11.0±0.4% of rFVIIIwt) and, in chromogenic assays, displayed an activity of 6.0±2.9%. Although reduced, these FVIII activity values were slightly higher than those measured in patients. The activity levels

Table 1. Comprehensive overview of detrimental effects of in vivo and in vitro on F8 mRNA and FVIII protein expression.FVIII exon 19

Patients' features Variants HA patients HGVS Legacy AA n. n. p.G2013R p.R2016W p.E2018G p.N2038S

1994 1997 1999 2019

1 61 7 10

rFVIII/splicing DATA FVIII activity (%) rFVIII activity Classification (%) <1 <1-5 1-4 5-20

s s/mo mo mo/mi

8.4±0.8 6.0±2.9 19.4±2.3 99.6±12.5

ex19 inclusion (%) 41±3 70±5 28±2 26±2

Inferred FVIII activity (%) Classification 2.8-4.0 2.0-6.7 4.4-6.5 20.9-31.4

mo mo/mi mo/mi mi

Inferred Factor VIII (FVIII) activity levels, reported as a range, are calculated according to rFVIII activity corrected for the % of exon inclusion. FVIII activity levels in patients affected by the p.R2016W mutation were measured by our group while those associated with the other mutations have been previously reported as referenced in the text,30-35 and as summarized in the HA mutation database (http://factorviii-db.org/). Severe (s), moderate (mo) or mild (mi) were defined as reported by Bolton-Maggs and Pasi.4 HGVS: Human Genome Variation Society.

Figure 2. Features of the F8 minigene and of the sequences under investigation. Schematic representation (bottom) of the F8 exon 19 region cloned into the pTB vector through the NdeI restriction sites (indicated). The sequences of the exon-intron boundary and of the engineered U1snRNA (U1 F8ex19) are reported together with 3’ss and 5’ss scores (numbers) (splice site prediction by neural network; www.fruitfly.org). (Top) The region masked by the engineered U7 F8exon19 is magnified. The investigated mutations are indicated as nucleotide and amino acid changes.

346

haematologica | 2018; 103(2)

Pleiotropic effects of HA-causing mutations

from1-stage clotting assays ranged from 1 to 6% [mean±Standard Deviation (SD): 3.6±1.8%; median: 4%] whereas those from chromogenic assays ranged from less than 1 to 5%. The mean value of the latter cannot be estimated because several patients display undefined values below 1%. When the comparison was conducted for each single patient, the chromogenic activity was, with the above mentioned limitation, roughly half that of the clotting one.

the c.6046c>t mutation further decreased exon 19 inclusion to 70% (Figure 3A, lower panel). To test the mechanistic hypothesis that exon 19 is inherently prone to skipping, we artificially strengthened exon 19 definition by acting on the recognition of the 5’ss by the spliceosomal U1snRNA, which is crucial in exon definition in the earliest splicing step.28 First, we optimized the authentic exon 19 5’ss by mutagenesis (high complementary 5’ss, 5’ssHC) to make it fully-complementary to the wild-type U1snRNA binding sequence, mimicking a perfect 5’ss. Conversely, we challenged the F8 minigene with an engineered U1snRNA (U1 F8ex19) designed to recognize the authentic exon 19 5’ss (Figure 2, lower part). Both approaches counteracted the effects of the c.6046c>t change (Figure 3A, lower panel), indicating that the poorly defined exon 19 strongly favored the variant’s pathogenic role.

The c.6046c>t change promotes skipping of the poorly defined F8 exon 19 The availability of RNA from leukocytes from a panel (n=20) of HA patients affected by the F8 c.6046c>t mutation prompted us to investigate the F8 mRNA splicing patterns.27 The RT-PCR analysis showed, with the limitation of studying ectopic expression, that the c.6046c>t change alters splicing and decreases the proportion of correct transcripts (74.1±14.5%) by inducing exon 19 skipping (Figure 3A and B, upper panels). To experimentally demonstrate the causative nature of the c.6046c>t change and investigate splicing mechanisms, we exploited F8 minigenes preserving the authentic exon 19 nucleotide sequence (Figure 2, lower panel). Splicing assays in two different human cell lines, HepG2 and HEK293, produced overlapping results indicating that 93±2% of transcripts were correctly spliced in the wild-type context (Figure 3A, lower panel). Computational analysis showed that the 5’ and 3’ splice sites (ss) defining exon 19 diverge significantly from the consensus sequence, as witnessed by the low scores (5’ss, 0.55; 3’ss 0.14 vs. 1 of the consensus) (Figure 2, lower panel), which might explain the observed exon skipping rate (~7%). In this nucleotide context, the introduction of

The c.6046c>t mutation marks a splicing regulatory region that is affected by several HA-causing missense changes To investigate whether the c.6046c>t change affects a splicing regulatory element we took advantage of the potent antisense approach based on the U7snRNA,29 which we designed (U7 F8exon19) to mask by base-pairing the region including the variant (Figure 2, upper part). Co-expression of the U7 F8exon19 (Figure 3A, lower panel) induced complete exon skipping even in the wildtype context, strongly suggesting the presence of a positive splicing regulatory element. We therefore selected five missense variants reported as causing HA clustered in the region covered by the U7 F8exon19 (Figure 2). Minigene assays demonstrated that

Figure 3. The c.6046c>t mutation as well as other adjacent missense changes promote exon 19 skipping differentially. (A) (Top) F8 splicing patterns in white blood cells from a normal control (N) and from 6 representative hemophilia A (HA) patients (P1-6) affected by the c.6046c>t mutation. The RT-PCR was conducted with primers in exons 17 and 22, and amplified fragments were separated on 2% agarose gel. (Bottom) Splicing assays with the wild-type (wt) and the mutated (c.6046c>t) minigenes. Effects of the antisense U7 F8ex19 (+) or the compensatory U1 F8ex19 (+) and of the strengthened 5’ss (5’ssHC) are shown. The RT-PCR was conducted with primers alpha 2,3 and Bra2 in the pTB construct (see arrows in Figure 2). Amplified fragments were separated on 2% agarose gel. Numbers report the percentage of exon 19 inclusion measured in at least three independent experiments; expressed as mean ± Standard Deviation. The percentage of exon 19 inclusion was estimated by densitometric analysis of bands. n: normal; s: exon 19 skipped. (B) (Top) Comparison of the exon 19 inclusion rates measured in 20 HA patients affected by the c.6046c>t mutation (light gray column) and in HepG2 cells expressing F8 minigene variants (dark gray columns). (Bottom) Representative example of splicing patterns of the selected missense variants. RT-PCR and analysis was conducted as in (A).

haematologica | 2018; 103(2)

347

I. Donadon et al.

two of them affected splicing (Figure 3B) and reduced exon inclusion to 41±3% (p.G2013R/c.6037g>a)30 and 28±2% (p.E2018G/c.6053a>g).31-33 We extended the investigation to missense variants close to (p.V2035A/c.6104t>c) or at (p.N2038S/c.6113a>g; -3 position)32,34,35 the 5’ss, and only the latter reduced exon inclusion (26±2%).

The splicing-defective F8 missense mutations differentially impair FVIII protein secretion and function To detail the impact of the underlying amino acid changes on FVIII protein we expressed and characterized the splicing-defective rFVIII missense variants. Results from expression studies (Figure 1B) demonstrated a spectrum of secreted levels, from virtually normal (rFVIII2018G and rFVIII-2038S) to poor (rFVIII-2013R, 7.0±0.9% of rFVIIIwt). Activity levels ranged from normal (rFVIII2038S) to significantly reduced (rFVIII-2018G, 19.4±2.3%; rFVIII-2013R, 8.4±0.8%). The specific co-factor activity (activity/antigen ratio) ranged from normal (rFVIII-2013R) to 50% (rFVIII-2016W) or was significantly (rFVIII-2018G) reduced, indicating dysfunctional features of the latter FVIII variants. The study at the protein level was completed with the expression of the in-frame transcript deriving from exon 19 skipping, clearly detectable in patients’ mRNA and associated with all splicing-defective variants. The rFVIIIΔ19 variant was not appreciably secreted, and, therefore, did not contribute to the FVIII protein levels.

Discussion This study stems from the notion that amino acid and splicing codes overlap36 and thus the frequent missense

changes, commonly considered only for the impact of the amino acid substitution on protein biology, might also have a detrimental effect on mRNA splicing. To provide qualitative and quantitative insights into the potential pleiotropic effect of missense mutations we chose to analyze the frequent p.R2016W/c.6046c>t variant reported to cause HA.11,12 It has previously been suggested to alter splicing13 and the amino change affects a partially (R2016) conserved (Figure 1A, lower panel) residue that in the FVIII structure (PDB: 2R7E)37 is surface exposed on the A3 domain at the interface with A1 domain (Figure 1C). Previous studies have shown that amino acid substitutions in this region have been shown to destabilize the FVIIIa structure and impair FVIII function,38 with a discrepancy in the functional activity values that were higher when measured by 1-stage clotting assays than with the chromogenic method.39-41 This activity profile was consistent with that observed in plasma from several HA patients bearing the F8 p.R2016W variant, which strongly supported an impact on the FVIII protein. Lentiviral-mediated expression of rFVIII led us to demonstrate that the R2016W substitution differentially impairs secreted (~11% of FVIIIwt) and functional (~6%) protein levels, a finding that confirms the dysfunctional features of the molecule. On the other hand, investigation at the ectopic F8 mRNA level in several HA patients, and through minigene assays, indicated that the corresponding nucleotide change promotes exon 19 skipping. Interestingly, the c.6046c>t change manifested the negative impact because of the weak context of exon 19, as witnessed by the observation that its effect was removed when placed in an artificially strengthened exon 19. Moreover, the observation that masking the region by an antisense U7snRNA induced exon-skipping revealed the presence of a splicing regulatory element with enhancer features not predicted by computational analysis.42 This

Figure 4. Pleiotropic effects of F8 variants. Detrimental effects of variants on F8 splicing and on Factor VIII (FVIII) protein expressed as % reduction extent of wildtype (wt) of correct transcripts (top: splicing impairment) or of co-factor activity (bottom: protein impairment).

348

haematologica | 2018; 103(2)

Pleiotropic effects of HA-causing mutations

was further supported by the identification of two adjacent missense changes causing HA that remarkably reduced exon inclusion (p.G2013R/c.6037g>a and p.E2018G/c.6053a>g). Altogether these findings suggest that splicing impairment by missense changes is largely under-estimated. The identified splicing-defective missense variants occur at residues that in the FVIII sequence alignment are completely (G2013, E2018) conserved (Figure 1A, lower panel), and in the FVIII structure37 are buried (G2013, E2018) within the A3 domain at the interface with A1 domain (Figure 1C). Recombinant FVIII protein expression demonstrated that they alter FVIII secretion and function differentially. We also identified the p.N2038S/c.6113a>g at the 5’ss (-3 position) that, as predicted, affected splicing. However, the amino acid change at the poorly conserved N2038 residue (Figure 1A) did not impact on FVIII protein biology, a finding that supports the relevance of aberrant splicing triggered by missense mutations. The combination of detrimental effects on splicing and protein observed in vitro (Figure 4) led us to infer FVIII functional levels (Table 1) that are consistent, albeit slightly over-estimated, with the FVIII activity measured in patients affected by the mutations, in spite of the considerable biological variation in patients and some limitations of the study. In particular, the exploitation of hybrid mini-

References 1. Branchini A, Rizzotto L, Mariani G, et al. Natural and engineered carboxy-terminal variants: decreased secretion and gain-offunction result in asymptomatic coagulation factor VII deficiency. Haematologica. 2012;97(5):705-709. 2. Branchini A, Campioni M, Mazzucconi MG, et al. Replacement of the Y450 (c234) phenyl ring in the carboxyl-terminal region of coagulation factor IX causes pleiotropic effects on secretion and enzyme activity. FEBS Lett. 2013;587(19):3249-3253. 3. Margaglione M, Castaman G, Morfini M, et al. The Italian AICE-Genetics hemophilia A database: results and correlation with clinical phenotype. Haematologica. 2008; 93(5):722-728. 4. Bolton-Maggs PH and Pasi KJ. Haemophilias A and B. Lancet. 2003; 361(9371):1801-1809. 5. Peyvandi F, Kunicki T, Lillicrap D. Genetic sequence analysis of inherited bleeding diseases. Blood. 2013;122(20):3423-3431. 6. Lenting PJ, van Mourik JA, Mertens K. The life cycle of coagulation factor VIII in view of its structure and function. Blood. 1998; 92(11):3983-3996. 7. Furlan Freguia C, Toso R, Pollak ES, Arruda VR, Pinotti M, Bernardi F. Characterization of mild coagulation factor VII deficiency: activity and clearance of the Arg315Trp and Arg315Lys variants in the Cys310-Cys329 loop (c170s). Haematologica. 2004; 89(12):1504-1509. 8. Balestra D, Barbon E, Scalet D, et al. Regulation of a strong F9 cryptic 5'ss by intrinsic elements and by combination of tailored U1snRNAs with antisense oligonucleotides. Hum Mol Genet. 2015; 24(17):4809-4916.

haematologica | 2018; 103(2)

genes and the expression of FVIII molecules lacking the B domain, necessary to appreciate the residual FVIII levels. However, the 'improved' B-domainless features could minimize the detrimental impact of amino acid substitutions on intracellular FVIII biosynthesis, potentially relevant for FVIII variants poorly secreted and functionally normal (e.g. rFVIII-2013R). In summary, data clearly identify an F8 exonic region that has evolved in relation to both mRNA maturation (nucleotide splicing code) and protein (amino acid code) constraints, a feature potentially shared with other exons and with implications for the HA pathogenic mechanisms and mutation pattern. For the first time, the combination of differentially altered mRNA processing, FVIII biosynthesis and co-factor activity was experimentally evaluated for clustered variants, and showed to substantially contribute to variably significant FVIII deficiency. This accounts for a gradient of residual FVIII activity that provides valuable information as to how to interpret the HA coagulation phenotypes. Funding The authors would like to thank Novo Nordisk (“Access to Insight Basic Research Grant 2014” to MP), Telethon-Italy (GGP14190 to MP) for the financial support and Prof. Shumperli for U7snRNA expressing vector.

9. Tajnik M, Rogalska ME, Bussani E, et al. Molecular Basis and Therapeutic Strategies to Rescue Factor IX Variants That Affect Splicing and Protein Function. PLoS Genet. 2016;12(5):e1006082. 10. Goodeve AC, Reitsma PH, McVey JH, Working Group on Nomenclature of the S, Standardisation Committee of the International Society on T, Haemostasis. Nomenclature of genetic variants in hemostasis. J Thromb Haemost. 2011;9(4):852855. 11. Santagostino E, Mancuso ME, Tripodi A, et al. Severe hemophilia with mild bleeding phenotype: molecular characterization and global coagulation profile. J Thromb Haemost. 2010;8(4):737-743. 12. Garagiola I, Seregni S, Mortarino M, et al. A recurrent F8 mutation (c.6046C>T) causing hemophilia A in 8% of northern Italian patients: evidence for a founder effect. Mol Genet Genomic Med. 2016;4(2):152-159. 13. Theophilus BD, Enayat MS, Williams MD, Hill FG. Site and type of mutations in the factor VIII gene in patients and carriers of haemophilia A. Haemophilia. 2001; 7(4):381-391. 14. Ward NJ, Buckley SM, Waddington SN, et al. Codon optimization of human factor VIII cDNAs leads to high-level expression. Blood. 2011;117(3):798-807. 15. Casari C, Pinotti M, Lancellotti S, et al. The dominant-negative von Willebrand factor gene deletion p.P1127_C1948delinsR: molecular mechanism and modulation. Blood. 2010;116(24):5371-5376. 16. Oberbek A, Matasci M, Hacker DL, Wurm FM. Generation of stable, high-producing CHO cell lines by lentiviral vector-mediated gene transfer in serum-free suspension culture. Biotechnol Bioeng. 2011; 108(3):600-610. 17. Pagani F, Stuani C, Tzetis M, et al. New

18.

19.

20.

21.

22.

23.

24.

25.

type of disease causing mutations: the example of the composite exonic regulatory elements of splicing in CFTR exon 12. Hum Mol Genet. 2003;12(10):1111-1120. Pinotti M, Rizzotto L, Balestra D, et al. U1snRNA-mediated rescue of mRNA processing in severe factor VII deficiency. Blood. 2008;111(5):2681-2684. Pagani F, Buratti E, Stuani C, Bendix R, Dork T, Baralle FE. A new type of mutation causes a splicing defect in ATM. Nat Genet. 2002;30(4):426-429. Kudaravalli R, Tidd T, Pinotti M, et al. Polymorphic changes in the 5' flanking region of factor VII have a combined effect on promoter strength. Thromb Haemost. 2002;88(5):763-767. Lunghi B, Pinotti M, Maestri I, Batorova A, Bernardi F. Evaluation of factor V mRNA to define the residual factor V expression levels in severe factor V deficiency. Haematologica. 2008;93(3):477-478. O'Brien DP, Pattinson JK, Tuddenham EG. Purification and characterization of factor VIII 372-Cys: a hypofunctional cofactor from a patient with moderately severe hemophilia A. Blood. 1990;75(8):16641672. Pipe SW, Eickhorst AN, McKinley SH, Saenko EL, Kaufman RJ. Mild hemophilia A caused by increased rate of factor VIII A2 subunit dissociation: evidence for nonproteolytic inactivation of factor VIIIa in vivo. Blood. 1999;93(1):176-183. Pipe SW, Saenko EL, Eickhorst AN, Kemball-Cook G, Kaufman RJ. Hemophilia A mutations associated with 1-stage/2stage activity discrepancy disrupt proteinprotein interactions within the triplicated A domains of thrombin-activated factor VIIIa. Blood. 2001;97(3):685-691. Nogami K, Zhou Q, Wakabayashi H, Fay PJ. Thrombin-catalyzed activation of factor

349

I. Donadon et al.

26.

27.

28.

29.

30.

31.

350

VIII with His substituted for Arg372 at the P1 site. Blood. 2005;105(11):4362-4368. Jourdy Y, Nougier C, Roualdes O, et al. Characterization of five associations of F8 missense mutations containing FVIII B domain mutations. Haemophilia. 2016; 22(4):583-589. Liang Q, Xiang M, Lu Y, et al. Characterisation and quantification of F8 transcripts of ten putative splice site mutations. Thromb Haemost. 2015;113(3):585592. Roca X, Krainer AR, Eperon IC. Pick one, but be quick: 5' splice sites and the problems of too many choices. Genes Dev. 2013;27(2):129-144. Marquis J, Meyer K, Angehrn L, Kampfer SS, Rothen-Rutishauser B, Schumperli D. Spinal muscular atrophy: SMN2 premRNA splicing corrected by a U7 snRNA derivative carrying a splicing enhancer sequence. Mol Ther. 2007;15(8):1479-1486. Repessé Y, Slaoui M, Ferrandiz D, et al. Factor VIII (FVIII) gene mutations in 120 patients with hemophilia A: detection of 26 novel mutations and correlation with FVIII inhibitor development. J Thromb Haemost. 2007;5(7):1469-1476. Hill M, Deam S, Gordon B, Dolan G. Mutation analysis in 51 patients with

32.

33.

34.

35.

36.

haemophilia A: report of 10 novel mutations and correlations between genotype and clinical phenotype. Haemophilia. 2005;11(2):133-141. Green PM, Bagnall RD, Waseem NH, Giannelli F. Haemophilia A mutations in the UK: results of screening one-third of the population. Br J Haematol. 2008; 143(1):115-128. Markoff A, Gerke V, Bogdanova N. Combined homology modelling and evolutionary significance evaluation of missense mutations in blood clotting factor VIII to highlight aspects of structure and function. Haemophilia. 2009;15(4):932-941. Liu M, Murphy ME, Thompson AR. A domain mutations in 65 haemophilia A families and molecular modelling of dysfunctional factor VIII proteins. Br J Haematol. 1998;103(4):1051-1060. Schwaab R, Oldenburg J, Schwaab U, et al. Characterization of mutations within the factor VIII gene of 73 unrelated mild and moderate haemophiliacs. Br J Haematol. 1995;91(2):458-464. Pagani F, Buratti E, Stuani C, Baralle FE. Missense, nonsense and neutral mutations define juxtaposed regulatory elements of splicing in CFTR Exon 9. J Biol Chem. 2003;278(29):26580-26588.

37. Shen BW, Spiegel PC, Chang CH, et al. The tertiary structure and domain organization of coagulation factor VIII. Blood. 2008;111(3):1240-1247. 38. Armstrong E and Hillarp A. Assay discrepancy in mild haemophilia A. Eur J Haematol Suppl. 2014;76:48-50. 39. Pavlova A, Delev D, Pezeshkpoor B, Müller J, Oldenburg J. Haemophilia A mutations in patients with non-severe phenotype associated with a discrepancy between one-stage and chromogenic factor VIII activity assays. Thromb Haemost. 2014;111(5):851-861. 40. Duncan EM, Rodgers SE, McRae SJ. Diagnostic testing for mild hemophilia a in patients with discrepant one-stage, twostage, and chromogenic factor VIII:C assays. Semin Thromb Hemost. 2013; 39(3):272-282. 41. Trossaërt M, Boisseau P, Quemener A, et al. Prevalence, biological phenotype and genotype in moderate/mild hemophilia A with discrepancy between one-stage and Chromogenic factor VIII activity. J Thromb Haemost. 2011;9(3):524-530. 42. Cartegni L, Wang J, Zhu Z, Zhang MQ, Krainer AR. ESEfinder: A web resource to identify exonic splicing enhancers. Nucleic Acids Res. 2003;31(13):3568-3571.

haematologica | 2018; 103(2)

ARTICLE

Coagulation & Its Disorders

Complement C3 is a novel modulator of the anti-factor VIII immune response

Ferrata Storti Foundation

Julie Rayes,1,2,3* Mathieu Ing,1,2,3* Sandrine Delignat,1,2,3 Ivan Peyron,1,2,3 Laurent Gilardin,1,2,3 Carl-Wilhelm Vogel,4,5 David C. Fritzinger,4 Véronique Frémeaux-Bacchi,1,2,3,6 Srinivas V. Kaveri1,2,3 Lubka T. Roumenina1,2,3 and Sébastien Lacroix-Desmazes1,2,3

INSERM, UMR S 1138, Centre de Recherche des Cordeliers, Paris, France; 2Université Pierre et Marie Curie-Paris6, UMR S 1138, France; 3Université Paris Descartes, UMR S 1138, France; 4University of Hawaii Cancer Center, University of Hawaii at Manoa, Honolulu, HI, USA; 5Department of Pathology, John A. Burns School of Medicine, University of Hawaii at Manoa, Honolulu, HI, USA and 6Assistance Publique-Hôpitaux de Paris, Service d’Immunologie Biologique, Hôpital Européen Georges-Pompidou, France 1

Haematologica 2018 Volume 103(2):351-360

*JR and MI contributed equally to the work.

ABSTRACT

evelopment of neutralizing antibodies against therapeutic Factor VIII (FVIII) is the most serious complication of the treatment of hemophilia A. There is growing evidence to show the multifactorial origin of the anti-FVIII immune response, combining both genetic and environmental factors. While a role for the complement system on innate as well as adaptive immunity has been documented, the implication of complement activation on the onset of the anti-FVIII immune response is unknown. Here, using in vitro assays for FVIII endocytosis by human monocyte-derived dendritic cells and presentation to T cells, as well as in vivo complement depletion in FVIIIdeficient mice, we show a novel role for complement C3 in enhancing the immune response against therapeutic FVIII. In vitro, complement C3 and its cleavage product C3b enhanced FVIII endocytosis by dendritic cells and presentation to a FVIII-specific CD4+ T-cell hybridoma. The C1 domain of FVIII had previously been shown to play an important role in FVIII endocytosis, and alanine substitutions of the K2092, F2093 and R2090 C1 residues drastically reduce FVIII uptake in vitro. Interestingly, complement activation rescued the endocytosis of the FVIII C1 domain triple mutant. In a mouse model of severe hemophilia A, transient complement C3 depletion by humanized cobra venom factor, which does not generate anaphylatoxin C5a, significantly reduced the primary anti-FVIII immune response, but did not affect anti-FVIII recall immune responses. Taken together, our results suggest an important adjuvant role for the complement cascade in the initiation of the immune response to therapeutic FVIII.

Correspondence: sebastien.lacroix-desmazes@crc.jussieu.fr

Received: January 31, 2017. Accepted: November 10, 2017. Pre-published: November 16, 2017. doi:10.3324/haematol.2017.165720 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/352

Introduction Hemophilia A is a rare X-linked inherited hemorrhagic disorder consecutive to the absence of functional pro-coagulant Factor VIII (FVIII). Prevention or treatment of bleeding consists of replacement therapy with exogenous FVIII. However, up to 30% of the patients develop antibodies of IgG isotype that neutralize the procoagulant activity of therapeutic FVIII, and are referred to as FVIII inhibitors.1 The immune response that develops against therapeutic FVIII is thought to be a classical T-cell dependent immune response to a foreign glycoprotein antigen. Thus, the antibody response and the nature of the implicated T cells have been extensively characterized.2 The initiation of a naïve T-cell-dependent immune response requires that the antigen-presenting cells (APCs) endocytose the target antigen. The APCs also need to receive appropriate maturation signals, referred to as “danger signals”, in order to mature and present the processed antigen to CD4+ T cells in an immunogenic context. In the case of the immune response to therapeutic FVIII, the haematologica | 2018; 103(2)

351

J. Rayes et al.

nature of the danger signals that adjuvant the anti-FVIII immune response and permit activation of naïve FVIIIspecific CD4+ T cells is not known. A role for bleeding and inflammation in the development of FVIII inhibitor has been proposed by several authors.3-5 The in vitro study of FVIII endocytosis has consistently been performed using serum-free medium or medium containing heat-inactivated serum,6-10 thus ignoring a potential role for the complement system in the observed mechanism. The complement system plays a major role in the development of immune responses.11 It is an integral part of the innate and adaptive host defense. Complement activation occurs through different pathways: the classical pathway is triggered by C1q binding to immune complexes, the lectin pathway is triggered by the binding of mannose binding lectin to mannose residues on pathogens, and the alternative pathway is spontaneously and continuously activated at a low rate (i.e. spontaneous C3 tickover).12,13 Inappropriate complement triggering is pathogenic and has been associated with autoimmune reactions.14 In the present work, we investigated the role of the complement system in the initiation and development of the anti-FVIII immune response. We demonstrate that transient depletion of complement using humanized cobra venom factor (hCVF) dampens the intensity of the primary anti-FVIII immune response in FVIII-deficient mice. We propose that initiation of the anti-FVIII immune response involves, at least in part, facilitation of FVIII endocytosis by C3 and its activation fragment C3b.

Methods Antibodies and reagents Full length FVIII was either a kind gift from CSL-Behring (Helixate® NexGen, Marburg, Germany) or from Baxter (Recombinate®, Maurepas, France). Recombinant human A disintegrin and metalloprotease with thrombospondin type I repeats13 (ADAMTS-13) was a kind gift from Baxter. Complement human proteins Factor B, Factor D, C3, C3b and C3-depleted serum were purchased from Complement Technology (Comptech, TX, USA) and Merck Millipore (Merck Chemicals Ltd., Nottingham, UK). Human serum was obtained from AB blood type healthy donors. Antibodies against CD1a, CD3, CD14, CD40, CD83, CD86, HLA-DR, CD206, low density lipoprotein receptor-related protein (LRP, CD91), CD209, CD68 and APClabeled Annexin V were purchased from BD Pharmingen (San Jose, CA, USA). Antibody against CD20 was purchased from eBiosciences (San Diego, CA, USA). The biotinylated monoclonal mouse anti-human FVIII antibody GMA-8015 and sheep polyclonal anti-human FVIII (SAF8C) were from Green Mountain Antibodies (Burlington, VT, USA) and Affinity Biological (Ancaster, Canada). The monoclonal antihuman FVIII antibody 77IP52H7 was a kind gift from LFB (Les Ulis, France). The biotinylated monoclonal mouse anti-human ADAMTS-13 antibody (20A5) and polyclonal goat anti-mouse C3b/iC3b (clone A209) were from Clinisciences (Nanterre, France) and Quidel (San Diego, USA), respectively.

Generation and production of recombinant wild-type or mutated FVIII, and of humanized cobra venom factor The wild-type human B-domain-deleted (BDD) FVIII (FVIIIHSQ) and the R2090A-K2092A-F2093A FVIII mutant (FVIIIC1) were generated and purified as described previously.15,16 Preparation of the 352

plasmid expressing HC3-1496, and expression and purification of HC3-1496 were performed as described previously for the preparation of pMB-HC3-1348.17 Details are provided in the Online Supplementary Appendix.

Generation of human immature monocyte-derived dendritic cells and monocyte-derived macrophages Monocytes were isolated from the blood of healthy donors using anti-CD14+ magnetic microbeads (Miltenyi Biotec, Paris, France). Monocytes were incubated in RPMI-1640 (Lonza, Verviers, Belgium) supplemented with GM-CSF (1000 IU/106 cells) and IL-4 (500 IU/106 cells) (Miltenyi Biotec) for five days to generate immature monocyte-derived dendritic cells (MO-DCs). The immature status was confirmed by flow cytometry (LSR II, BD Biosciences, Le Pont au Claix, France) with a CD1a, CD14, CD80, CD86, CD83, CD40 and HLA-DR staining. To differentiate monocytes into macrophages (MO-Φ), cells were plated in RPMI-1640 supplemented with M-CSF (ImmunoTools, Friesoythe, Germany, 100 ng/106 cells) for five days. Macrophage phenotype was confirmed by flow cytometry with CD68 staining.

Isolation of circulating human blood dendritic cells Circulating dendritic cells (DCs) were isolated from the blood of healthy donors using the Blood Dendritic Cell Isolation Kit II (Miltenyi Biotec). Purity was assessed as more than 80% by flow cytometry using exclusion staining based on CD3, CD14 and CD20.

Uptake of antigens by immature MO-DCs FVIII (50 nM, Helixate®), FVIIIHSQ (50 nM), FVIIIC1 (50 nM), ADAMTS-13 (50 nM) conjugated to Dylight 633 (Thermo Scientific, Courtaboeuf, France) or human Factor IX (50 nM, Benefix, Pfizer, France) were incubated with 20% normal serum (NS), heat-inactivated (56°C for 30 min) serum (HIS) or C3-depleted serum (ΔC3) in RPMI-1640 at 37°C for 1 hour (h). When indicated, FVIII or Dylight 633-labeled ADAMTS-13 were incubated with C3 (250 mg/mL), Factor B (50 mg/mL) and Factor D (1 mg/mL) or with C3b (250 mg/mL) in 20 mM HEPES 150 mM NaCl 10 mM MgCl2 for 1 h at 37°C. Samples were then incubated with 5-dayold immature MO-DCs (2.105 cells/100 mL) for 2 h at 4°C or 37°C. Cells were washed with ice-cold phosphate buffer saline, fixed with 4% paraformaldehyde (Sigma-Aldrich). In the case of FVIII, cells were permeabilized using 0.5% saponin (Sigma-Aldrich) and stained with an FITC-conjugated monoclonal mouse anti-human FVIII IgG (77IP52H7). Cells were analyzed by flow cytometry and data were computed using the BD FACS Diva software (v.6.1, BD Biosciences). The uptake was quantified as the difference in the median fluorescence intensities between 37°C and 4°C (ΔMFI37°C4°C) to exclude the signal generated by the binding of FVIII to the cell surface.

Stimulation of an FVIII-specific HLA-DRB1*0101restricted mouse CD4+ T-cell hybridoma Activation of the HLA-DRB1*0101-restricted mouse CD4+ Tcell hybridoma specific for human FVIII (1G8-A2), was assessed as described previously.18 FVIII (10 nM) was incubated with 20% NS, HIS or ΔC3 serum in RPMI-1640 supplemented with 2 mM CaCl2 and 10 mM MgCl2 for 1 h at 37°C. Samples were then incubated with 5-day-old MO-DCs (104 cells/condition) generated from healthy donors with the DRB1*0101 haplotype, for 2 h at 37°C to allow FVIII uptake. MO-DCs were then washed, suspended in XVIVO15 medium, and co-cultured with 105 1G8-A2 CD4+ T cells for 18 h at 37°C. Controls included T cells incubated alone, or incubated with MO-DCs in the presence of concanavalin A (2 mg/mL, Sigma-Diagnostics, MO, USA) or in the absence of FVIII. haematologica | 2018; 103(2)

Complement adjuvants the anti-FVIII immune response

Levels of interleukine-2 secreted by T cells were assessed using BD OptEIA™ mouse IL-2 ELISA set (BD Biosciences).

FVIII binding to immobilized C3b by ELISA C3b was coated on ELISA plates at 10 nM in sodium bicarbonate buffer (pH 9.4) overnight at 4°C. Wells were blocked in PBS5% bovine serum albumin (BSA). FVIII was diluted in blocking buffer and incubated on C3b-coated wells for 1 h at 37°C. Bound FVIII was revealed using a biotinylated monoclonal mouse antihuman FVIII antibody GMA-8015, streptavidin-HRP (R&D Systems, Lille, France) and the HRP o-phenylenediamine dihydrochloride substrate (OPD, Sigma-Aldrich, St. Louis, MO, USA). As a control, bound ADAMTS-13 was revealed using the biotinylated mouse anti-human ADAMTS-13 antibody, 20A5.

ma (Siemens Healthcare Diagnostics, Marburg, Germany) for 2 h at 37°C. The residual FVIII pro-coagulant activity was measured using a chromogenic assay (Siemens Healthcare Diagnostics). Results are expressed in Bethesda Units (BU/mL) that correspond to the reciprocal dilution of the mouse plasma that yields 50% residual FVIII activity.

Ethical considerations Mice were handled in agreement with French ethical authorities (authorization ns. 02058.04 and 8275.02). Ethical committee permission was obtained for the use of buffy bags from healthy donors to isolate monocytes.

Results Co-localization of FVIII and C3b by immunofluorescence FVIII (50 nM, Helixate®) or buffer alone were incubated in the presence of 20% HIS or NS for 1 h at 37°C in RPMI-1640. The mixture was added on MO-DCs (250,000 cells) for 2 h at 37°C. After washing with ice-cold PBS, cells were fixed using 3.7% paraformaldehyde and permeabilized. FVIII and C3b/iC3b were recognized after incubation with anti-human FVIII SAF8C (2 µg/mL) and a polyclonal anti-mouse C3b/iC3b (10 mg/mL) antibody for 30 minutes (min) at room temperature under agitation followed by the addition of the secondary A647-conjugated donkey anti-sheep IgG and A488-conjugated goat anti-mouse IgG (Molecular Probes, Eugene, USA), respectively. After washing, nuclei were stained with Hoecht®33342 (Sigma-Aldrich) and dendritic cells were plated onto poly-L-lysine coated slides using cytospin centrifugation (5 min at 1500 rpm), and mounted in ProLong Diamond Anti-Fade Mountant (Molecular Probes). Images were acquired with LSM 710 confocal microscope (Zeiss, Oberkochen, Germany).

Animals and administration of FVIII Exon 16 FVIII-deficient mice on the C57Bl/6 background (a gift from Prof. H.H. Kazazian, University of Pennsylvania School of Medicine, Philadelphia, PA, USA) were 8-15 weeks old. Mice were injected intravenously with 1 mg human recombinant FVIII (Helixate®) once a week for four weeks. Blood was collected five days after the last FVIII administration. Plasma was isolated and kept at -80°C until use.

In vivo complement blockade Complement was depleted in FVIII-deficient mice by intraperitoneal injection of 20 mg of hCVF. Importantly, hCVF does not cleave C5.19 C3 levels in plasma were measured by sandwich ELISA, using a polyclonal goat anti-mouse C3 antibody (MP Biomedicals, Illkirch, France) to capture C3 and a biotinylated polyclonal goat anti-mouse C3 antibody, followed by streptavidin-HRP and OPD substrate, to reveal bound C3. hCVF administration occurred 6 h prior to FVIII administration.

Titration of anti-FVIII IgG and FVIII inhibitors ELISA plates (Nunc, Roskilde, Denmark) were coated with FVIII (1 mg/mL, Recombinate®) overnight at 4°C. After blocking with PBS-1% BSA, plasma was incubated for 1 h at 37°C. Bound IgG were revealed using an HRP-coupled polyclonal goat anti-mouse IgG antibody (Southern Biotech, Anaheim, CA, USA) and the OPD substrate. Absorbance was read at 492 nm. The monoclonal mouse FVIII heavy chain-specific IgG mAb6 (a gift from Dr J.M. Saint-Remy, Katholieke Universiteit Leuven, Leuven, Belgium) was used as a standard. FVIII inhibitors were measured by incubating heat-inactivated mouse plasma with human standard plashaematologica | 2018; 103(2)

Complement in FVIII endocytosis by MO-DCs and blood DCs but not MO-Φ We first assessed FVIII endocytosis by MO-DCs using flow cytometry. Incubation of FVIII with immature human MO-DCs in the presence of normal serum (NS) resulted in a 1.86±0.61-fold increase in FVIII uptake as compared to incubation in the presence of heat-inactivated serum (HIS) (P<0.01) (Figure 1A). The deposition of C3b at the cell surface upon incubation of immature MODCs in the presence of NS at 37°C (Figure 1B) confirmed complement activation. No deposition of C3b occurred when cells were incubated in the presence of C3-depleted serum. Opsonization of the cells by C3b was associated with an increased binding of FVIII to MO-DCs, as assessed by incubation at 4°C (P<0.05) (Figure 1C). Interestingly, a direct binding of FVIII to immobilized C3b was observed by ELISA (Figure 1D), while ADAMTS-13, used as a control antigen, failed to bind to C3b. In line with this, immunofluorescence confirmed the co-localization of FVIII and C3b on MO-DCs (Figure 1E and Online Supplementary Figure S2). Taken together, the data suggest a role for C3b in facilitating FVIII capture and internalization by MO-DCs. Of note, incubation with NS did not significantly modify the expression of LRP (CD91) and CD206 (two endocytic receptors for FVIII) or DC-SIGN (CD209) (Online Supplementary Figure S1). Incubation of FVIII with purified circulating blood DCs from healthy donors in the presence of NS yielded a 3-fold increase in FVIII uptake (P<0.05) (Figure 2A), while it had only a marginal effect on FVIII uptake in the case of MO-Φ (Figure 2B). Interestingly, the presence of complement did not alter the endocytosis by MO-DCs of human recombinant ADAMTS-13 and Factor IX, used as control antigens (Figure 2C and D).

Complement C3 participates in FVIII endocytosis by MO-DCs leading to presentation to CD4+ T cells Because the complement C3 molecule is the central component of the complement cascade, and because FVIII directly binds to C3b in ELISA and co-localizes with C3b on the cells, we assessed the involvement of C3 in FVIII endocytosis by dendritic cells. Levels of FVIII endocytosis by MO-DCs were similar in the presence of C3-depleted serum (ΔC3) and in the presence of HIS (Figure 3A and B). Importantly, incubation of MO-DCs with HIS, NS or ΔC3 lead to similar degrees of cell surface expression of phosphatidylserine, a ligand for activated FVIII at the surface of activated platelets20 (Online Supplementary Figure S3A). Differences in FVIII endocytosis were thus not related to 353

J. Rayes et al.

different levels of phosphatidylserine expression (i.e. apoptosis) on MO-DCs. To confirm that the reduced FVIII endocytosis observed upon C3 depletion results in a reduced presentation to CD4+ T lymphocytes, we assessed the activation of a CD4+ T-cell hybridoma (1G8-A2) by MO-DCs incubated with FVIII in the presence of the different serum preparations. Importantly, the use of the T-cell hybridoma provides an opportunity to work with FVIII concentrations closer to those reached in patients with replacement therapy. The activation of 1G8-A2, assessed by the secretion of interleukine-2 (IL-2),18 was significantly higher when FVIII was incubated with MO-DCs in the presence of NS (541±123 pg/mL of IL-2) (Figure 3C) as compared to HIS (217±132 pg/mL; P≤0.05) or to ΔC3 serum (238±61 pg/mL; P≤0.01). There was no activation of 1G8-A2 when MODCs were incubated with the different serum prepara354

Figure 1. Complement enhances Factor VIII (FVIII) endocytosis by human monocyte-derived dendritic cells (MO-DCs). (A, C and E) FVIII (50 nM) was incubated in 20% normal serum (NS) or in 20% heatinactivated serum (HIS) for 1 hour (h) at 37°C. Mixtures were added on 5-day-old immature MO-DCs (2.105 cells in RPMI1640) for 2 h at 4°C or 37°C. Mean fluorescence intensity (MFI) was measured by flow cytometry using an FITC-coupled antiFVIII antibody. (A) MFI values measured at 4°C were subtracted from values measured at 37°C, in order to eliminate the signal due to the binding to the cell surface. Results represent the fold change of ∆MFI37°C-4°C measured for each condition versus ∆MFI37°C-4°C obtained in the presence of HIS. Data are shown as means±Standard Deviation (SD) of fold change of MFI measured in the case of NS versus that measured in the case of HIS (n=6). Dead cells were excluded using fixable viability dye. (B) MO-DCs were incubated alone (gray area), with NS (full line curve), HIS (broken line curve) or C3-deficient plasma (∆C3, dotted line curve) for 1 h at 37°C. Surface-bound C3b was detected by flow cytometry using an antiC3b antibody. Results are shown as histograms. (C) FVIII bound at the surface of the cells was detected by flow cytometry following incubation of FVIII with the cells at 4°C. (D) The binding of FVIII and ADAMTS-13 to immobilized C3b (10 nM) was measured by ELISA using specific anti-FVIII or anti-ADAMTS-13 IgG. The results are expressed in optical density (OD) measured at 492 nm. (C and D) Mean±SD of at least 3 independent experiments. Statistical significance was assessed using the double-sided nonparametric Mann-Whitney test (A) or paired Wilcoxon test (C). (E) The co-localization of FVIII (red) and C3b (green) was assessed by immunofluorescence, using polyclonal anti-human FVIII and antiC3b/iC3b antibodies and the appropriate fluorescent-labeled secondary antibodies. Images were acquired by confocal microscopy. Control images obtained in the absence of FVIII are shown in Online Supplementary Figure S1.

tions in the absence of FVIII. Of note, incubation of immature human MO-DCs with HIS, NS or ΔC3 serum did not induce DC maturation in vitro, as assessed by the surface expression of co-stimulatory CD40, CD80, CD86, and maturation CD83 molecules, and of HLA-DR (Online Supplementary Figure S4). We then evaluated the role of C3b in FVIII endocytosis by MO-DCs in the absence of serum proteins. We coincubated FVIII with either purified C3b or with an artificially reconstituted C3 convertase containing C3, Factor B and Factor D. The reconstituted C3 convertase leads to the generation of C3b as confirmed by western blot (data not shown). Samples were then added to MO-DCs and FVIII endocytosis was monitored by flow cytometry (Figure 3D). Co-incubation of immature MO-DCs with the C3 convertase or with C3b did not induce detectable apoptosis as assessed by the surface expression of phoshaematologica | 2018; 103(2)

Complement adjuvants the anti-FVIII immune response

Figure 2. Complement enhances FVIII endocytosis by human professional antigen-presenting cells. (A and B) FVIII (50 nM) Dylight 633-conjugated ADAMTS-13 (50 nM, panel C) or human recombinant FITC-labeled Factor IX (50 nM, panel D) were incubated in 20% normal serum (NS) or in 20% heat-inactivated serum (HIS) for 1 hour (h) at 37°C. Mixtures were added on purified blood dendritic cells (DCs) (A), 5-day-old MO-Φ (B) or 5-day-old immature human monocyte-derived dendritic cells (MO-DCs) (2.105 cells in RPMI-1640, panels C and D), for 2 h at 4°C or 37°C. Mean fluorescence intensity (MFI) was measured by flow cytometry using an FITC-coupled anti-FVIII antibody in the case of FVIII, or directly in the case of labeled ADAMTS-13 and Factor IX. MFI values measured at 4°C were subtracted from values measured at 37°C in order to eliminate the signal due to the binding to the cell surface. Results represent the fold change of ∆MFI37°C-4°C measured for each condition versus ∆MFI37°C-4°C obtained in the presence of HIS. Graphs depict mean±Standard Deviation (SD) of at least 3 independent experiments. Dead cells were excluded. Statistical significance was assessed using the double-sided non-parametric Mann-Whitney test. ns: not significant.

phatidylserine (Online Supplementary Figure S3B). FVIII endocytosis was significantly higher when FVIII was incubated in the presence of the whole C3 convertase (1.88±0.28-fold increase) or C3b (1.78±0.40-fold increase) than when FVIII was incubated with Factors B and D alone (P<0.05). Consistent with the fact that ADAMTS-13 did not bind to C3b, incubation of ADAMTS-13 with the C3 convertase or with C3b did not result in increased endocytosis by MO-DCs (Figure 3E). We then compared the endocytosis by MO-DCs of a recombinant B-domain-deleted FVIII (FVIIIHSQ) to that of a triple R2090A/F2092A/K2093A FVIII mutant (FVIIIC1) in the absence and presence of complement activation. As previously reported using serum-free medium,9 FVIIIC1 was not endocytosed by MO-DCs when incubated in HIS (Figure 4). In contrast, in the presence of NS, FVIIIC1 was endocytosed to levels similar to those reached with FVIIhaematologica | 2018; 103(2)

IHSQ incubated in HIS. Taken together, the data confirm a role for complement C3 in FVIII endocytosis by MO-DCs and presentation to T cells, independently of known endocytic receptors and critical FVIII residues.

Complement depletion decreases the primary immune response against FVIII in FVIII-deficient mice Naïve FVIII-deficient mice were treated with hCVF prior to each administration of FVIII, in order to transiently deplete the complement C3 molecule in the blood. Injection of hCVF was followed by a 90% or more decrease of circulating C3 levels that lasted for at least 5 h (Online Supplementary Figure S5). The efficacy of hCVFmediated C3 depletion was retained following repeated weekly administration to the same mice, indicating the absence of induction of a neutralizing immune response against hCVF (data not shown). Of note, hCVF lacks C5355

J. Rayes et al. A

Figure 3. Complement C3 enhances Factor VIII (FVIII) endocytosis by dendritic cells and presentation to CD4+ T cells. (A and B) FVIII (50 nM) was incubated for 1 hour (h) at 37°C in 20% normal serum (NS), heat-inactivated serum (HIS) or C3-depleted serum (ΔC3). Mixtures were added on immature human monocyte-derived dendritic cells (MO-DCs) (2.105 cells in RPMI-1640) for 2 h at 4°C or 37°C. FVIII was then detected with an FITC-coupled anti-FVIII antibody. Mean fluorescence intensity (MFI) values measured at 4°C were subtracted from values measured at 37°C. (A) Histograms from FVIII-positive cells incubated with HIS (solid line), NS (bold line) and ΔC3 (dashed line). The gray histogram depicts the fluorescence of cells incubated alone. (B) Fold-change of ∆MFI37°C-4°C measured for each condition versus ∆MFI37°C-4°C obtained in the presence of HIS. (C) DCs were differentiated from monocytes of healthy donors with the HLA-DRB1*0101 haplotype. Five-day-old immature MO-DCs (104 cells in X-VIVO15) were incubated alone (Control), with concanavalin A (2 mg/mL, ConA) or with 10 nM FVIII in the presence of 20% normal serum (NS), heat-inactivated serum (HIS) or C3-depleted serum (ΔC3) for 2 h at 37°C. Cells were washed and incubated with an FVIII-specific HLADRB1*0101-restricted mouse CD4+ T-cell hybridoma for 18 h at 37°C. Culture supernatants were collected and analyzed for IL-2 secretion. (D and E) FVIII (50 nM, panel D) or Dylight 633-conjugated ADAMTS-13 (50 nM, panel E) were incubated alone (Control), in the presence of Factor B (FB, 50 µg/mL) and Factor D (FD, 1 mg/mL) with or without C3 (250 mg/mL), or in the presence of C3b (250 mg/mL) for 1 h at 37°C. Samples were then incubated with 5-day-old immature MO-DCs (2.105 cells in X-VIVO15) for 2 h at 4°C or 37°C. MFI was measured by flow cytometry using an FITC-coupled anti-FVIII antibody in the case of FVIII or directly in the case of labeled ADAMTS-13. MFI values measured at 4°C were subtracted from values measured at 37°C. Results represent the fold change of ∆MFI37°C-4°C measured for each condition versus ΔMFI MFI37°C-4°C measured with FVIII alone (Control). All results are expressed as mean±Standard Deviation (SD) and are representative of at least 5 independent experiments. Statistical significance was assessed using the double-sided non-parametric Mann-Whitney test. ns: non-significant.

convertase activity and hence does not generate the potent pro-inflammatory C5a anaphylatoxin.19,21 FVIIIdeficient mice treated with hCVF prior to FVIII infusion demonstrated significantly reduced levels of anti-FVIII IgG titers in plasma, as compared to the PBS control group [(5±67 vs. 267±245 mg/mL mAb6-equivalent, respectively, mean±Standard Deviation (SD); P<0.0001] (Figure 5A), as well as reduced inhibitory titers (66±55 vs. 195±97 BU/mL), respectively; depletion decreases the primary immune response against FVIII in FVIII-deficient mice, P<0.0001) (Figure 5B). Of note, depletion of C3 using hCVF did not significantly affect the half-life of FVIII in Figure 4 (Left). Complement rescues the endocytosis of a triple Factor VIII (FVIII) FVIIIC1 mutant by dendritic cells. Recombinant wild-type (FVIIIHSQ) and mutated (FVIIIC1) B-domain-deleted FVIII (50 nM) were incubated for 1 hour (h) at 37°C in 20% normal serum (NS) or heat-inactivated serum (HIS). Mixtures were added on immature monocyte-derived dendritic cells (MO-DCs) (2.105 cells in RPMI-1640) for 2 h at 4°C or 37°C. FVIII was then detected with an FITC-coupled anti-FVIII antibody directed to the A2 domain of FVIII. Mean fluorescence intensity (MFI) values measured at 4°C were subtracted from values measured at 37°C. Horizontal bars depict mean±Standard Deviation (SD) of 5 independent experiments. Statistical significance was assessed using the double-sided non-parametric Mann-Whitney test.

356

haematologica | 2018; 103(2)

Complement adjuvants the anti-FVIII immune response

FVIII-deficient mice (Online Supplementary Figure S6). In contrast, administration of hCVF to primed FVIII-deficient mice did not have any effect on the recall anti-FVIII immune response (Figure 5C and Online Supplementary Figure S7).

Discussion Complement C3 has been shown to mediate antigen uptake by professional APCs and presentation to CD4+ and CD8+ T cells.22 Therefore, we investigated the role of complement in the endocytosis and presentation of FVIII by human APCs in vitro. The use of heat-treated serum, wherein the complement system is inactivated, was asso-

ciated with baseline levels of FVIII endocytosis by MODCs, as previously described.7,23,24 Similar levels of endocytosis were observed when the serum was immune depleted from the C3 component. In contrast, the use of normal serum that allows activation of complement leads to an increased uptake of FVIII in the case of both MO-DCs and circulating blood DCs. In line with these data, normal serum enhanced presentation of FVIII to an FVIII-specific T-cell hybridoma, as compared to serum lacking active C3. Since heating of serum may affect proteins other than complement, we reconstituted the C3-convertase in vitro using purified proteins. In vitro reconstitution of the C3convertase, or addition of the C3-activation fragment C3b, in the absence of other complement molecules was sufficient to restore close to maximal levels of FVIII endo-

Figure 5. Complement modulates the primary immune response to therapeutic Factor VIII (FVIII) in vivo. (A and B) FVIII (1 mg/mouse) was injected intravenously to naïve FVIII-deficient mice 6 hours (h) after an intraperitoneal injection of hCVF (20 μg) or phosphate buffered saline (PBS), once a week for four weeks. At day 28, blood was collected. (C) Naïve FVIII-deficient mice were injected once a week for four weeks with FVIII (1 mg/mouse). At week 9, FVIII-primed mice were treated with either 20 mg hCVF or PBS, and 6 h later with FVIII. Blood was collected before hCVF/PBS injection and at weeks 10 and 11 (see Online Supplementary Figure S7 for the protocol). Anti-FVIII IgG titers in plasma were assessed by ELISA (panels A and C, using the monoclonal mouse anti-FVIII antibody mAb6 as a standard) and FVIII inhibitory titers by chromogenic assay [panel B, expressed in Bethesda Units (BU) per milliliter]. Results are representative of 3 independent experiments. In all panels, horizontal bars represent medians, and statistical difference weas assessed using the double-sided non-parametric Mann-Whitney test. ns: non-significant.

haematologica | 2018; 103(2)

357

J. Rayes et al.

cytosis. In agreement with in vitro data, transient depletion of component C3 in naĂŻve FVIII-deficient mice using hCVF resulted in a drastically reduced production of antibodies to exogenous FVIII. Since the complement system is not expected to preferentially affect the specific immune response towards a particular antigen, we investigated the effect of complement on the endocytosis of ADAMTS-13. Human recombinant ADAMTS-13 was used as a control antigen for human recombinant FVIII, with the rationale that: i) recombinant ADAMTS-13 is being developed as a drug for replacement therapy to be used in patients with congenital thrombotic thrombocytopenic purpura (TTP), which is reminiscent of the development of recombinant FVIII for hemophilia A patients; ii) ADAMTS-13, like FVIII, is endocytosed by MO-DCs in a CD206-dependent manner;24 and, iii) as is seen in the case of FVIII, autoantibodies to endogenous ADAMTS-13 may develop, thus leading to acquired TTP. In contrast to the results obtained

in the case of FVIII, ADAMTS-13 did not bind to C3b, and in vitro complement activation or addition of C3b had no effect on the endocytosis of ADAMTS-13 by MO-DCs. The data suggest a selective role of complement in immune responses towards some glycoproteins, including FVIII. The effect of complement on endocytosis was dependent on the cell type (i.e. blood DCs as well as MO-DCs, but not MO-ÎŚ) and on the antigen (i.e. FVIII, but not ADAMTS-13 or Factor IX), suggesting the involvement of particular endocytic pathway(s) in complement activation. It is not likely that activation of complement induces expression of endocytic receptors for FVIII at the surface of the cells due to the short incubation time used in our study. Accordingly, activation of complement on MODCs did not alter the surface expression of LRP (CD91) and CD206, which are known endocytic receptors for FVIII.7,25,26 Particular FVIII moieties have been implicated in its uptake by APCs. Charged residues in the C1 and C2

Figure 6. Endocytosis pathways for Factor VIII (FVIII) in the absence or presence of complement activation. (A) In the absence of complement activation (i.e. use of decomplemented serum or of serum-free medium in cell culture), FVIII is endocytosed by three known routes: key residues in the C1 and in the C2 domain interact with yet unknown endocytic receptors,27,28 while high mannose-ending glycans at position 2118 on the C1 domain interact with the mannose-sensitive receptor CD206 (MR).7 Of note, alteration of either of these structures (e.g. mutation of key residues in either the C1 domain or the C2 domain) leads to a more than 80% reduction in FVIII uptake, suggesting that these three endocytic pathways are interdependent. (B) In the presence of complement activation, C3b deposition on dendritic cells is associated with an increase in FVIII binding and internalization by dendritic cells (Figure 1A and C). Importantly, complement activation partially restores the uptake of the C1 FVIII mutant. The data suggest that complement engages yet unidentified endocytic pathways rather than those at play in the absence of complement activation.

358

haematologica | 2018; 103(2)

Complement adjuvants the anti-FVIII immune response

domains of FVIII were recently shown to play critical roles in the endocytosis process. Masking of the R2090, F2092 and K2093 residues of the C1 domain by the monoclonal antibody Km33, or of the R2215 and R2220 residues of the C2 domain by the monoclonal antibody BO2C11, or mutation towards alanine residues, independently abrogate FVIII uptake by MO-DCs.9,27,28 Importantly, the experiments leading to the identification of the latter pathways of FVIII uptake had been performed systematically in the presence of serum-free medium, and therefore did not allow assessment of whether complement has a role in the endocytic process. The fact that complement activation rescues, at least in part, the internalization of the triple FVIIIC1 mutant reveals the existence of novel endocytic pathways for FVIII, independently of charged residues in C1.27 Together, the deposition of C3b at the surface of MO-DCs upon complement activation and the binding of FVIII to immobilized C3b in ELISA, suggest that FVIII interacts with the C3b deposited at the cell surface; this would in turn allow the recruitment of those as yet unidentified endocytic receptors (Figure 6). The question as to the origin of complement C3 activation at the time of FVIII administration in vivo remains. Injection of FVIII into FVIII-deficient mice did not lead to detectable complement activation. Conversely, immune responses to FVIII developed in FVIII-deficient mice without any overt sign of spontaneous hemorrhage or inflammation,29 and this was also seen in wild-type mice. However, the anti-FVIII immune response was drastically reduced in hemophilic mice upon exhaustion of C3 using hCVF. Under physiological conditions, a small fraction of the C3 molecules is hydrolyzed to C3(H2O), exposing new binding sites. The Factor B protease then binds C3(H2O) and is cleaved by Factor D, generating an initial C3 convertase, C3(H2O)Bb, that activates complement by cleaving C3 into its active fragments, C3a and C3b.30,31 The continuously generated C3b indiscriminately binds to host cells and pathogens and initiates a set of cascade reactions.12,13 Importantly, the C3(H2O)Bb convertase is more resistant to inactivation by Factor H and Factor I than is the C3bBb convertase.12 Taken together, these data suggest that such a spontaneous tick-over of C3, leading to permanent C3b generation, is sufficient to facilitate FVIII uptake by professional APCs, thus setting the stage for the initiation of the anti-FVIII immune response in vivo. More gen-

References 1. Ehrenforth S, Kreuz W, Scharrer I, et al. Incidence of development of factor VIII and factor IX inhibitors in haemophiliacs. Lancet. 1992;339(8793):594-598. 2. Lacroix-Desmazes S, Navarrete AM, Andre S, Bayry J, Kaveri SV, Dasgupta S. Dynamics of factor VIII interactions determine its immunologic fate in hemophilia A. Blood. 2008;112(2):240-249. 3. Skupsky J, Zhang AH, Su Y, Scott DW. A role for thrombin in the initiation of the immune response to therapeutic factor VIII. Blood. 2009;114(21):4741-4748. 4. Dimitrov JD, Dasgupta S, Navarrete AM, et al. Induction of heme oxygenase-1 in factor

haematologica | 2018; 103(2)

erally, our data illustrate a novel biological role of C3 tickover in the recognition of some non-particulate foreign antigens. In addition, the immune response to FVIII might be modulated by the cross-talks between the complement and coagulation cascades32 or by the influence of von Willebrand factor on complement activation.33-35 Obviously, part of the anti-FVIII immune response does not depend on C3b and on activation of the complement cascade, as is suggested by the fact that: i) FVIII demonstrates substantial levels of endocytosis in the absence of complement activation; and ii) FVIII-deficient mice still develop anti-FVIII IgG when treated with hCVF. However, our results do highlight the existence of a new endocytic route for FVIII which is independent of the endocytosis mediated by the C1 domain of the molecule,27 and which should lead to the identification of the endocytic receptors involved in vivo. Future studies should be conducted to decipher the contribution of each of the different pathways of complement activation in the inhibitory anti-FVIII immune response, and should also investigate whether temporary C3 depletion with hCVF is a promising therapeutic strategy to prevent the development of anti-FVIII antibody in patients with hemophilia A. Funding This study was supported by Institut National de la Santé et de la Recherche Médicale (INSERM), Centre National de la Recherche Scientifique (CNRS), Université Pierre et Marie Curie (UPMC) Paris 6, and by a grant from Pfizer (Aspire Haemophilia Research award 2016 WI212828). MI and IP were recipients of fellowships from Ministère de l'Enseignement Supérieur et de la Recherche. LG was the recipient of a ‘poste d’accueil INSERM’ fellowship. FVIIIHSQ in the ReNeo plasmid and BHK-M cells were kind gifts from Prof. Pete Lollar (Aflac Cancer and Blood Disorders Center, Department of Pediatrics, Emory University, Atlanta, GA, USA). Recombinant ADAMTS13 and FVIII were kind gifts from Baxter Innovations GmbH (Vienna, Austria), and CSL-Behring (Marburg, Germany), respectively. Acknowledgments We also would like to thank the staff from the Centre d’Imagerie Cellulaire et Cytométrie platform and Centre d'Expérimentation Fonctionnelle for assistance (Centre de Recherche des Cordeliers, Paris).

VIII-deficient mice reduces the immune response to therapeutic factor VIII. Blood. 2010;115(13):2682-2685. 5. Kurnik K, Auerswald G, Kreuz W. Inhibitors and prophylaxis in paediatric haemophilia patients: focus on the German experience. Thromb Res. 2014;134 (Suppl 1):S27-32. 6. Castro-Nunez L, Dienava-Verdoold I, Herczenik E, Mertens K, Meijer AB. Shear stress is required for the endocytic uptake of the factor VIII-von Willebrand factor complex by macrophages. J Thromb Haemost. 2012;10(9):1929-1937. 7. Dasgupta S, Navarrete AM, Bayry J, et al. A role for exposed mannosylations in presentation of human therapeutic self-proteins

to CD4+ T lymphocytes. Proc Natl Acad Sci USA. 2007;104(21):8965-8970. 8. Dasgupta S, Repesse Y, Bayry J, et al. VWF protects FVIII from endocytosis by dendritic cells and subsequent presentation to immune effectors. Blood. 2007;109(2):610612. 9. Herczenik E, van Haren SD, Wroblewska A, et al. Uptake of blood coagulation factor VIII by dendritic cells is mediated via its C1 domain. J Allergy Clin Immunol. 2012; 129(2):501-509. 10. Sorvillo N, Hartholt RB, Bloem E, et al. von Willebrand factor binds to the surface of dendritic cells and modulates peptide presentation of factor VIII. Haematologica. 2016;101(3):309-318.

359

J. Rayes et al. 11. Merle N, Church S, Fremeaux-Bacchi V, Roumenina L. Complement system part I molecular mechanisms of activation and regulation. Front Immunol. 2015;6:262. 12. Bexborn F, Andersson PO, Chen H, Nilsson B, Ekdahl KN. The tick-over theory revisited: formation and regulation of the soluble alternative complement C3 convertase (C3(H2O)Bb). Mol Immunol. 2008; 45(8):2370-2379. 13. Zewde N, Gorham RD Jr, Dorado A, Morikis D. Quantitative Modeling of the Alternative Pathway of the Complement System. PLoS One. 2016;11(3):e0152337. 14. Ballanti E, Perricone C, Greco E, et al. Complement and autoimmunity. Immunol Res. 2013;56(2-3):477-491. 15. Horton RM, Ho SN, Pullen JK, Hunt HD, Cai Z, Pease LR. Gene splicing by overlap extension. Methods Enzymol. 1993; 217:270-279. 16. Doering CB, Healey JF, Parker ET, Barrow RT, Lollar P. High level expression of recombinant porcine coagulation factor VIII. J Biol Chem. 2002;277(41):3834538349. 17. Fritzinger DC, Hew BE, Thorne M, et al. Functional characterization of human C3/cobra venom factor hybrid proteins for therapeutic complement depletion. Dev Comp Immunol. 2009;33(1):105-116. 18. Delignat S, Repesse Y, Gilardin L, et al. Predictive immunogenicity of Refacto AF. Haemophilia. 2014;20(4):486-492. 19. Vogel CW, Finnegan PW, Fritzinger DC. Humanized cobra venom factor: structure, activity, and therapeutic efficacy in preclinical disease models. Mol Immunol.

360

2014;61(2):191-203. 20. Gilbert GE, Furie BC, Furie B. Binding of human factor VIII to phospholipid vesicles. J Biol Chem. 1990;265(2):815-822. 21. Vogel CW, Fritzinger DC. Cobra venom factor: Structure, function, and humanization for therapeutic complement depletion. Toxicon. 2010;56(7):1198-1222. 22. Jacquier-Sarlin MR, Gabert FM, Villiers MB, Colomb MG. Modulation of antigen processing and presentation by covalently linked complement C3b fragment. Immunology. 1995;84(1):164-170. 23. Repesse Y, Dasgupta S, Navarrete AM, Delignat S, Kaveri SV, Lacroix-Desmazes S. Mannose-sensitive receptors mediate the uptake of factor VIII therapeutics by human dendritic cells. J Allergy Clin Immunol. 2012;129(4):1172-1173. 24. Sorvillo N, Pos W, van den Berg LM, et al. The macrophage mannose receptor promotes uptake of ADAMTS13 by dendritic cells. Blood. 2012;119(16):3828-3835. 25. Lenting P, Neels J, van den Berg B, et al. The light chain of factor VIII comprises a binding site for low density lipoprotein receptor-related protein. J Biol Chem. 1999;274:23734-23739. 26. Saenko E, Yakhyaev A, Mikhailenko I, Strickland D, Sarafanov A. Role of the low density lipoprotein-related protein receptor in mediation of factor VIII catabolism. J Biol Chem. 1999;274:37685-37692. 27. Wroblewska A, van Haren SD, Herczenik E, et al. Modification of an exposed loop in the C1 domain reduces immune responses to factor VIII in hemophilia A mice. Blood. 2012;119(22):5294-5300.

28. Gangadharan B, Ing M, Delignat S, et al. The C1 and C2 domains of blood coagulation factor VIII mediate its endocytosis by dendritic cells. Haematologica. 2017; 102(2):271-281. 29. Bi L, Sarkar R, Naas T, et al. Further characterization of factor VIII-deficient mice created by gene targeting: RNA and protein studies. Blood. 1996;88:3446-3450. 30. Lachmann PJ, Halbwachs L. The influence of C3b inactivator (KAF) concentration on the ability of serum to support complement activation. Clin Exp Immunol. 1975; 21(1):109-114. 31. Pangburn MK, Schreiber RD, MullerEberhard HJ. Formation of the initial C3 convertase of the alternative complement pathway. Acquisition of C3b-like activities by spontaneous hydrolysis of the putative thioester in native C3. J Exp Med. 1981; 154(3):856-867. 32. Foley JH, Conway EM. Cross Talk Pathways Between Coagulation and Inflammation. Circ Res. 2016;118(9):13921408. 33. Rayes J, Roumenina LT, Dimitrov JD, et al. The interaction between factor H and VWF increases factor H cofactor activity and regulates VWF prothrombotic status. Blood. 2014;123(1):121-125. 34. Feng S, Liang X, Kroll MH, Chung DW, Afshar-Kharghan V. von Willebrand factor is a cofactor in complement regulation. Blood. 2015;125(6):1034-1037. 35. Noone DG, Riedl M, Pluthero FG, et al. Von Willebrand factor regulates complement on endothelial cells. Kidney Int. 2016; 90(1):123-134.

haematologica | 2018; 103(2)

ARTICLE

Blood Transfusion

Hypoxia modulates the purine salvage pathway and decreases red blood cell and supernatant levels of hypoxanthine during refrigerated storage

Travis Nemkov,1# Kaiqi Sun,2# Julie A. Reisz,1# Anren Song,2 Tatsuro Yoshida,3 Andrew Dunham,3 Matthew J. Wither,1 Richard O. Francis,4 Robert C. Roach,5 Monika Dzieciatkowska,1 Stephen C. Rogers,6 Allan Doctor,6 Anastasios Kriebardis,7 Marianna Antonelou,8 Issidora Papassideri,8 Carolyn T. Young,9 Tiffany A. Thomas,4 Kirk C. Hansen,1 Steven L. Spitalnik,4 Yang Xia,2 James C. Zimring,10 Eldad A. Hod4 and Angelo D’Alessandro1,11

Ferrata Storti Foundation

Haematologica 2018 Volume 103(2):361-372

Department of Biochemistry and Molecular Genetics, University of Colorado Denver – Anschutz Medical Campus, Aurora, CO, USA; 2Department of Biochemistry, University of Texas Houston – School of Medicine, Houston, TX, USA; 3New Health Sciences Inc, Boston, MA, USA; 4Department of Pathology & Cell Biology, Columbia University Medical Center, New York, NY, USA; 5Altitude Research Center, University of Colorado Denver – Anschutz Medical Campus, Aurora, CO, USA; 6Division of Critical Care Medicine, Department of Pediatrics, School of Medicine, Washington University in St Louis, St Louis, MO, USA; 7Department of Medical Laboratories, Technological and Educational Institute of Athens, Greece; 8Department of Biology, National and Kapodistrian University of Athens, Greece; 9Rhode Island Blood Center, Providence, RI, USA and 10 BloodWorks Northwest, Seattle, WA, USA. 11Boettcher Investigator. 1

TN, KS and JAR contributed equally to this work and share the first authorship.

ABSTRACT

Correspondence:

ypoxanthine catabolism in vivo is potentially dangerous as it fuels production of urate and, most importantly, hydrogen peroxide. However, it is unclear whether accumulation of intracellular and supernatant hypoxanthine in stored red blood cell units is clinically relevant for transfused recipients. Leukoreduced red blood cells from glucose-6-phosphate dehydrogenase-normal or -deficient human volunteers were stored in AS-3 under normoxic, hyperoxic, or hypoxic conditions (with oxygen saturation ranging from <3% to >95%). Red blood cells from healthy human volunteers were also collected at sea level or after 1-7 days at high altitude (>5000 m). Finally, C57BL/6J mouse red blood cells were incubated in vitro with 13C1-aspartate or 13C5-adenosine under normoxic or hypoxic conditions, with or without deoxycoformycin, a purine deaminase inhibitor. Metabolomics analyses were performed on human and mouse red blood cells stored for up to 42 or 14 days, respectively, and correlated with 24 h post-transfusion red blood cell recovery. Hypoxanthine increased in stored red blood cell units as a function of oxygen levels. Stored red blood cells from human glucose-6-phosphate dehydrogenase-deficient donors had higher levels of deaminated purines. Hypoxia in vitro and in vivo decreased purine oxidation and enhanced purine salvage reactions in human and mouse red blood cells, which was partly explained by decreased adenosine monophosphate deaminase activity. In addition, hypoxanthine levels negatively correlated with post-transfusion red blood cell recovery in mice and – preliminarily albeit significantly - in humans. In conclusion, hypoxanthine is an in vitro metabolic marker of the red blood cell storage lesion that negatively correlates with post-transfusion recovery in vivo. Storage-dependent hypoxanthine accumulation is ameliorated by hypoxia-induced decreases in purine deamination reaction rates. haematologica | 2018; 103(2)

angelo.dalessandro@ucdenver.edu

Received: August 11, 2017. Accepted: October 24, 2017. Pre-published: October 27, 2017. doi:10.3324/haematol.2017.178608 Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/103/2/361 ©2018 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.

361

T. Nemkov et al.

Introduction Packed red blood cell (RBC) transfusions are life-saving interventions for millions of recipients every year (~11.3 million units transfused/year in the USA alone1). Refrigerated RBC storage is required to make the ~100 million units collected worldwide every year available for transfusion. However, refrigerated RBC storage induces many biochemical and morphological alterations, collectively termed the “storage lesion.”2–5 Some alterations are promoted by oxidative stress, arising within the first 2 weeks of storage,6,7 targeting proteins,8–11 lipids,12–15 and various small molecules, including purines.16–19 These observations, together with early depletion of high energy phosphate compounds, including adenosine triphosphate (ATP) and 2,3-diphosphoglycerate (2,3-DPG), prompted many to question whether the storage lesion compromises RBC transfusion safety and efficacy. Despite evidence documenting the evolution and severity of the storage lesion,2–4 whether prolonged RBC storage duration adversely affects transfused recipients is debatable. Recent randomized clinical trials concluded that the general standard of care would not be improved by exclusively issuing fresh RBC, at least for the studied clinical indications and within the statistical power of these studies.20 However, these clinical trials did not examine blood that was particularly old (e.g., >35 days). Recent evidence indicates that the storage lesion does not develop linearly and only transfusing >35-day old RBC induces increases in circulating non-transferrin-bound iron in healthy recipients.21 In parallel, increases in adverse outcomes have been observed in high-risk patients receiving blood units aged >35 days.22 Thus, to the extent that in vivo hemolysis and non-transferrin-bound iron are mediators of the adverse effects of transfusion, the clinical trials to date have not shown that transfusing old blood is safe. Current storage solutions make it possible to store RBC for up to 42 days with an average ~17% loss of transfusion potency at outdate,23 based on 51Cr-labeled post-transfusion recovery (PTR) studies in healthy human volunteers,24 which provide information on the ability of RBC to circulate, but not necessarily their ability to deliver oxygen. This is relevant when considering the effects of the loss of potency in massively transfused recipients, such as trauma patients.23 Despite reassuring evidence from clinical trials, further improvement in RBC storage strategies are possible, as recommended by the US National Heart, Lung, and Blood Institute.25 To this end, advances in the molecular understanding of the storage lesion have fostered the design of novel storage solutions (e.g., alkaline additives26) and strategies (e.g., hypoxic storage27) to improve storage quality. In parallel, recently identified “omics” markers of storage age28–30 may prove useful for benchmarking potential improvements in storage quality, once their association with post-transfusion outcomes has been clearly demonstrated. The present study addresses this by focusing on hypoxanthine,17,28,31 a deaminated purine resulting from the metabolism of ATP, adenosine monophosphate (AMP), and adenosine in mature RBC. Recently, Casali et al.,31 Bordbar et al.16 and Paglia et al.28 reported that end-of-storage SAGM RBC and supernatants are characterized by levels of hypoxanthine as high as ~450 and 1000 mM, respectively. As previously noted,31 under physiological conditions the concentration of hypoxanthine is very low, 362

both inside erythrocytes (9.3 nM) and in human plasma (1–8 mM). Transfusion of a single unit of end-of-storage blood containing almost millimolar levels of hypoxanthine could thus result in circulating hypoxanthine levels of ~100 mM, an amount that would further increase proportionally to the number of older units transfused to the recipient. Notably, concentrations as high as 100 mM are used in cytotoxicity assays to produce toxic amounts of hydrogen peroxide generated through the xanthine dehydrogenase/oxidase-catalyzed conversion of hypoxanthine to urate.31 On this background, in the present study, we provide absolute quantitative measurements of hypoxanthine in cells and supernatants of AS-3 packed RBC. We report negative correlations between RBC hypoxanthine levels in vitro and PTR in vivo in 14 different mouse strains and, preliminarily, in healthy human volunteers, indicating the potential clinical relevance of this metabolic lesion. We also provide a possible mechanistic explanation regarding the role of AMP deaminase (AMPD) activation in human and mouse RBC as a function of hemoglobin oxygen saturation (SO2) and resulting oxidative stress in vitro and in vivo. Finally, by combining state-of-the-art deep proteomics, quantitative mass spectrometry (MS), tracing experiments with stable isotope-labeled substrates, and pharmacological inhibition of purine deaminases, we identify the role of hypoxia in preventing AMPD activation, thereby decreasing the storage-dependent accumulation of deaminated purines, particularly hypoxanthine.

Methods Blood samples were collected from healthy donor volunteers upon receiving written informed consent and in conformity with the Declaration of Helsinki under the protocol approved by the relative institutions, including the University of Texas Houston and University of Colorado Denver institutional review boards (n. AWC-14-0127 and 11-1581, respectively). Commercial reagents were purchased from Sigma-Aldrich (Saint Louis, MO, USA) unless otherwise noted.

Glucose-6-phosphate dehydrogenase-normal and -deficient human red blood cells, stored under normoxic or hypoxic conditions Blood was collected from healthy glucose-6-phosphate dehydrogenase (G6PD)-normal donors at the Bonfils Blood Center (Denver, CO, USA) or from G6PD-deficient donors (Mediterranean variant) in Athens (Greece) according to the Declaration of Helsinki. Filter leukocyte-reduced (Pall Medical, Braintree, MA, USA) packed RBC were stored in CP2D-AS-3 (n=4; Haemonetics Corp., Braintree, MA, USA) or CPD-SAGM (n=6). Units were sampled in a sterile manner (15 mL per time point) on days 0, 21, 42, and cells and supernatants were separated by centrifugation at 2000 x g for 10 min at 4°C.

Mouse red blood cell storage under normoxic and hypoxic conditions with an adenosine monophosphate deaminase inhibitor RBC were collected aseptically by exsanguination from C57BL/6J mice (pool of n=5 per group) and stored for 14 days32 in CPD-AS-3 under normoxic or hypoxic conditions (O2 = 21% or 8%, respectively), in the presence or absence haematologica | 2018; 103(2)

Hypoxanthine in stored RBC of 13C5-adenosine (5 mM) and deoxycoformycin (500 mM), an AMPD inhibitor (500 mM), as described.33

Post-transfusion recovery studies in healthy human donor volunteers PTR studies were performed at Columbia University Medical Center-New York Presbyterian Hospital in healthy volunteers receiving autologous packed RBC (n=52), and were previously published21 without accompanying metabolomics data. Briefly, immediately before issue, a 25 mL sample of blood, obtained from the unit using a sterile docking device, was radiolabeled with 51Cr,21 while a matching 500 mL sample was immediately frozen for metabolomics analyses. At 1-4 h after transfusion of the unit, the 51Cr–labeled RBC sample was infused over 1 min. Blood specimens were then obtained every 2.5 min between 5 and 15 min after infusion and used to extrapolate time zero and the final time point to calculate PTR.21 Hypoxanthine levels were measured in the transfusates of the subjects in the previously published study.21

Post-transfusion recovery studies in mice The PTR studies in mice were performed as described previously,34 using multiple strains from Jackson Labs (Bar Harbor, ME, USA): KK/HIJ, LG/J, AKR/J, FVB/NJ, C3H/HeJ, DBA/2J, NOD/ShiLtJ, 129X1/SvJ, 129S1/SvImJ, A/J, BTBR/ T+ tf/J, Balb/cByJ, C57Bl/6J. UbiC-GFP male mice, on a C57BL/6 background, were bred to FVB/NJ females in the Bloodworks NW Research Institute Vivarium (Seattle, WA, USA) and offspring were used as transfusion recipients at 24-28 weeks of age.34

Human red blood cell oxygen saturation 8 hours after donation CP2D-AS-3 RBC with controlled SO2 [i.e., >95% (hyperoxic) to <3% (hypoxic)] on day 0 were prepared in vented chambers (Difco BLL, Detroit, MI, USA). SO2 levels in 977 RBC units at the Rhode Island Blood Center were determined within 8 h of donation and routine processing (i.e. leukofiltration and storage in AS-3 under standard normoxic blood bank conditions), with methods previously described and validated.35

High altitude studies RBC were collected from 12 male and nine female healthy human volunteers at sea level or after 3 h (ALT1 am), >8 h (ALT1 pm), or 7 days (ALT7) of exposure to high altitude hypoxia (5260 m) in Mt. Chacaltaya, Bolivia, within the framework of the AltitudeOmics study.36

Red blood cell treatment with xanthine dehydrogenase/oxidase and hypoxanthine Human RBC were exposed to 1.5 mM hypoxanthine in the presence of xanthine oxidase (0.8 U/mL) at 37°C for up to 6 h in a shaking water bath, as reported elsewhere.37

Proteomic analyses Leukocyte-reduced human RBC from healthy donor volunteers were washed five times in phosphate-buffered saline prior to lysis in distilled water with sonication. Proteomic analyses of RBC membranes and cytosols were performed as described elsewhere38 using 30 mg of protein per time point and a 4-12% gradient SDS-PAGE gel.39 Bands were reduced, alkylated, trypsin digested, and then analyzed by nanoUHPLC-MS/MS (nanoEasy LC II coupled to haematologica | 2018; 103(2)

a Q Exactive HF – Thermo Fisher, Bremen, Germany). Alternatively, RBC cytosolic proteins were depleted of hemoglobin using HemovoidTM (Biotech support group, Monmouth Junction, NJ, USA), prior to high-pH reversed phase fractionation;40 64 fractions were collected (32 each for cytosol and membranes) over a 3 h gradient, prior to nano-UHPLC-MS/MS proteomics, as described previously.9 Error tolerant searches were performed using Mascot (v. 2.4) against the human UniprotKB database (release date 2015.1.8), including decoy sequences (cysteine carbamidomethylation and methionine oxidation set as fixed and variable modifications, respectively). Mass tolerances for membrane and vesicle data were set to ±15 ppm for precursor ions and ±0.01 Da for fragment ions. For all Mascot search results, peptide spectral matches were filtered at a 95% confidence threshold (excluding matches with an expectation value >0.05).

Sample processing and metabolite extraction RBC were separated by centrifugation (10 min at 4ºC and 2500 x g) and then 50 mL were extracted in 450 mL of lysis buffer (methanol:acetonitrile:water 5:3:2) via ice cold extraction by vortexing for 30 min at 4ºC.14,41 Insoluble proteins were pelleted by centrifugation (10 min at 4ºC and 10,000 x g) and supernatants were collected and stored at -80°C until analysis.

Ultrahigh performance liquid chromatography – mass spectrometry metabolomics Analyses were performed using a Vanquish UHPLC system coupled online to a Q Exactive mass spectrometer (Thermo Fisher, Bremen, Germany). Samples were resolved over a Kinetex C18 column (2.1 x 150 mm, 1.7 µm; Phenomenex, Torrance, CA, USA) at 25ºC using a 3 min isocratic condition of 5% acetonitrile, 95% water, and 0.1% formic acid flowing at 250 mL/min,42 or using a 9 min gradient at 400 mL/min from 5-95% B (A: water/0.1% formic acid; B: acetonitrile/0.1% formic acid).14 MS analysis and data processing were performed as described elsewhere.14 Metabolite assignments, absolute quantification against heavy labeled internal standards and calibration curves of external standards (Cambridge Isotopes, Tewksbury, MA, USA), isotopologue distributions, and correction for expected natural abundances of 13C isotopes were performed using MAVEN (Princeton, NJ, USA), as already described.42 Absolute quantities of hypoxanthine were determined using the following formula: HypoxanthineAbsoluteQuant = (Integrated Peak AreaLight / Integrated Peak AreaHeavy) x ConcentrationHeavy x 10 where 10 is the dilution factor Graphs and statistical analyses (t-test, repeated measures ANOVA, or Spearman correlation) were prepared with GraphPad Prism 5.0 (GraphPad Software, Inc., La Jolla, CA, USA).

Results Hypoxanthine accumulates during murine and human red blood cell storage and negatively correlates with post-transfusion recovery Intracellular and supernatant hypoxanthine levels progressively increase during standard storage of human RBC in AS-3 (Figure 1A), reaching concentrations as high as 450 363

T. Nemkov et al. A

mM and 800 mM, respectively, consistent with previous studies using other storage solutions (e.g. SAGM).16,31 Quantitative hypoxanthine levels also allow discrimination between three metabolic phases of the RBC storage lesion, as previously determined by receiver operating characteristic (ROC) curves for RBC stored in SAGM.28 Here we show similar results for RBC stored in AS-3 (Figure 1B) on the basis of absolute quantitative measurements, confirming and improving on the previous relative quantitative measurements (i.e., arbitrary units).28 Deamination of RBC purines (Figure 1C) by purine deaminases, such as AMPD, was initially described in the 1930s.43 More recent studies focused on the role of oxidative stress in activating AMPD3, the RBC-specific isoform.44 Consistent with observations in humans, hypoxanthine levels were reported to increase in stored C57BL/6J mouse RBC samples, without separation of cells and supernatants;45 here we confirmed and extended this observation by separately analyzing stored mouse RBC and supernatants (Figure 2A). Interestingly, in contrast to the human samples (Figure 1A), supernatants of mouse RBC stored in AS-3 contained less hypoxanthine at the end of storage as compared to RBC cytosol (Figure 1B). Despite increased understanding of the storage lesion,2 it remains controversial whether the (metabolic) storage lesion in vitro holds any clinical relevance.3 PTR is – according to regulations of the US Food and Drug Administration – a standard measure of RBC storage quality with potential clinical relevance, in that the capacity to circulate in the recipient’s bloodstream is a necessary (but not sufficient) requirement for RBC to function in vivo. We, therefore, performed 24 h PTR studies in mice and humans, and compared these results with hypoxanthine levels (Figure 2B). To this end, stored RBC from 79 mice of 364

Figure 1. Hypoxanthine is a metabolic marker of the red blood cell storage lesion. (A) Intracellular (left) and supernatant (right) levels of hypoxanthine increase in packed RBC stored in the presence of AS-3. All data points shown on the x axes were tested and interpolated with third order polynomial curves (not assuming linear evolution of hypoxanthine accumulation during storage) and median ± ranges (n=4) are shown (dark blue lines and light blue areas, respectively). (B) ROC curves identify hypoxanthine as a biomarker of the three metabolic stages of stored RBC with extreme sensitivity and specificity (as indicated by high true positive and low false positive rates), confirming previous observations in SAGM.25 A simplified overview of purine catabolism and oxidation intermediates is outlined. Structures are provided for adenine (top right) and hypoxanthine (bottom right).

14 different strains were assayed for hypoxanthine levels and PTR studies were completed following transfusion into C57BL6 recipients. As a result, a significant negative correlation was found (median of all strains P<0.0001; r = -0.87 Spearman; r = -0.88 for 8 out of 14 strains) (Figure 2C). Similarly, pre-transfusion hypoxanthine levels in stored human RBC from healthy volunteers (n=52, transfused RBC from week 1-6) negatively correlated with PTR in the autologous transfusion setting (P=0.0018, r = -0.43 Spearman) (Figure 2D).

Oxygen saturation affects purine deamination in vivo and ex vivo Refrigerated RBC storage promotes oxidative stress,6 which is mitigated, in part, by decreasing SO2 during hypoxic storage.9 Strikingly, SO2 levels in 977 freshly donated units assayed within 8 h of donation and standard processing were widely distributed on storage day 1 (Online Supplementary Figure S1), ranging from 5% to 95%. This indicates that SO2 at donation is currently an uncontrolled variable in the donor population with the potential to affect RBC purine deamination. We, therefore, performed a series of experiments to determine whether exposure to hypoxia or hyperoxia in vitro or in vivo affected purine deamination and hypoxanthine accumulation in human and mouse RBC. Notably, exposure of healthy volunteers (and good acclimatizers) to high altitude hypoxia (>5000 m for up to 7 days) (Figure 3A) led to significant decreases in RBC hypoxanthine levels, even by 3 h (ALT1am, P=0.02) and 8 h (ALT1pm, P=0.0002) after ascent, which were even greater after 7 days at high altitude (P=8.9x10-5) (Figure 3B). Similarly, exposure of C57BL/6J mice (n=6) to hypobaric hypoxia (8% O2 for up to 8 h) led to significant (P<0.01) decreases in RBC hypohaematologica | 2018; 103(2)

Hypoxanthine in stored RBC

Figure 2. Hypoxanthine negatively correlates with post-trasfusion recovery of using mouse and human red blood cells. (A) Hypoxanthine accumulation is observed in stored C57BL/6J mouse RBC. (B) Transfusion into GFP-RBC mouse recipients (sorting of fluorescence negative RBC) or 51Cr labeling of human RBC was performed to determine 24 h PTR in mice (n=79) and human volunteers (n=52), whereas paired samples were used to determine hypoxanthine levels. (C, D) Negative correlations were observed for mouse (C) and human RBC hypoxanthine levels and 24 h PTR. Linear and quadratic Spearman correlations as well as their levels of significance are shown for mouse (median across strains) and (D) human data.

haematologica | 2018; 103(2)

Figure 3. Hypoxanthine levels decrease in human and mouse red blood cells exposed to hypoxia in vivo and in vitro. (A) RBC were collected from 21 healthy volunteers at sea level (SL) and within 3 or >8 h after exposure to high altitude hypoxia on day 1 (ALT1 am and pm, respectively), and at 7 days (ALT7) after exposure to high altitude (>5000 m) hypoxia. (B) Hypoxanthine levels decreased significantly in human RBC within hours of exposure to high altitude hypoxia (x axis labels consistent with description of panel A). (C) C57BL/6J mice (n=6) were exposed to normoxia or 8% oxygen for 3 h, resulting in decreases in RBC levels of hypoxanthine, a phenomenon accompanied by decreased IMP and increased AMP/IMP ratios (Online Supplementary Figure S1). (D) RBC were collected from C57BL/6J mice prior to in vitro storage in AS-3 for up to 2 weeks under normoxic or hypoxic conditions, resulting in decreased hypoxanthine accumulation.

365

T. Nemkov et al.

xanthine levels (Figure 3C), while boosting AMP/IMP ratios (Online Supplementary Figure S2). In addition, when mouse RBC (n=5) were stored for up to 2 weeks under normoxic or hypoxic (8% O2) conditions, lower hypoxanthine levels were seen with hypoxia (Figure 3D). Analogously, human RBC (n=4) were stored under controlled SO2 conditions, ranging from hyperoxia (SO2 >95%), to normoxia (no SO2 control), to hypoxia (SO2 = 20%, 10%, 5% or <3%) (Figure 4A). Although all conditions induced storagedependent increases in cellular and supernatant hypoxanthine levels, hypoxic RBC (independently of the degree of hypoxia) resulted in significantly (P<0.01) lower levels of intracellular hypoxanthine from storage day 14 onwards (Figure 4B – left panel). Supernatant hypoxanthine levels

were significantly lower in all hypoxic RBC after storage day 14 (Figure 4B – right panel). No significant effect of hyperoxia (SO2 >95%) was observed, except for higher than control (P<0.05) supernatant hypoxanthine levels at storage day 7, consistent with previous observations that SO2 can increase to >95% by storage week 3 in normoxic (uncontrolled SO2) RBC with day 0 SO2 >60% (though units in this study were agitated biweekly after sampling).9

Purine salvage and deamination in human and mouse red blood cells as a function of oxidative stress In the previous section we showed that oxygen saturation affects hypoxanthine accumulation in human and mouse RBC in vitro and in vivo. Mechanistically, hypoxia-

Figure 4. Effects of oxygene saturation on hypoxanthine accumulation during refrigerated storage. (A) Human RBC were donated by healthy volunteers (n=4) prior to refrigerated storage in AS-3 under control (normoxic), hyperoxic (SO2>95%), or hypoxic (SO2 = 20%, 10%, 5% or <3%) conditions for up to 42 days. (B) Hypoxanthine concentration was determined in cells and supernatants, with significant decreases in the presence of hypoxia. All data points shown on the x axes were tested and interpolated with third order polynomial curves (not assuming linear evolution of hypoxanthine accumulation during storage) and medians ± ranges (n=4) are shown (dark blue lines and light blue areas, respectively).

Figure 5. Purine salvage and deamination reactions. (A) Purine salvage and deamination reactions are catalyzed by adenylosuccinate synthase (ADSS - 1) and adenylosuccinate lyase (ASL – 2), and by adenosine monophosphate deaminase (AMPD3 - 3), respectively. (B) Deep proteomic analyses identified ADSS, ASL and AMPD3 in human RBC, in contrast to prior studies indicating the absence of ADSS in mature human RBC.42

366

haematologica | 2018; 103(2)

Hypoxanthine in stored RBC

dependent decreases in purine deamination may be explained either by decreases in AMPD3 activity or increases in salvage reactions, including those catalyzed by adenylosuccinate synthetase (ADSS) and adenylosuccinate lyase (ASL) (Figure 5A). Deep proteomic analyses of RBC from healthy donor volunteers identified AMPD3, ASL (also confirmed by Western blot in mouse RBC) (Online Supplementary Figure S3), and traces of ADSS in human RBC (sequence coverage shown in Figure 5B), despite classic literature indicating the absence of ADSS in mature human RBC.46 Mascot scores for peptide hits for each one of these proteins are reported in Online Supplementary Table S1. AMPD3 activity is stimulated by oxidative stress,44 an observation supported herein by metabolomics evidence demonstrating increased inosine monophosphate (IMP) and hypoxanthine, and decreased AMP and AMP/IMP ratios, in human RBC exposed to hydrogen peroxide-generating xanthine dehydrogenase/oxidase (XDH) in the presence of hypoxanthine (at levels similar to those observed in RBC supernatants at the end of storage; i.e., the high mM to low mM range) (Figure 6A). G6PD is the rate-limiting enzyme of the pentose phosphate pathway (PPP), which is essential for RBC redox homeostasis because it generates reducing cofactors (i.e., NADPH) to recycle oxidized glutathione back to its reduced form. Thus, we hypothesized that refrigerated storage of RBC from G6PD-deficient subjects increases oxidative stress and would result in increased hypoxan-

thine levels. In our hands, refrigerated storage of RBC from G6PD-deficient volunteers (n=6; Mediterranean variant, <10% residual activity, non-hemolytic) demonstrated significant increases in hypoxanthine levels in comparison to controls, beginning after storage day 14 (Figure 6B).

Hypoxia prevents hypoxanthine accumulation by deregulating adenosine monophosphate deamination rather than by promoting salvage reactions Based on our observations, oxidative stress promoted purine deamination and hypoxanthine accumulation in RBC after: (i) incubation with xanthine dehydrogenase in the presence of hypoxanthine, and (ii) refrigerated storage, particularly in the case of G6PD-deficient donors. Conversely, hypoxia in vivo and in vitro has a beneficial role in mice and humans, by preventing purine deamination and hypoxanthine accumulation. Enzymes involved in salvage reactions (ADSS and ASL) and purine deamination (AMPD3) have been identified in mature RBC via deep proteomic measurements, suggesting that either pathway may be susceptible to oxidative stress and/or hypoxia. To expand on these steady state observations to test actual fluxes through these pathways, human RBC were incubated with 13C1-aspartate for 6 h to determine the rate of â&#x20AC;&#x153;heavyâ&#x20AC;? fumarate accumulation through salvage reactions (Figure 7A). Accumulation of isotopologue +1 of fumarate from aspartate may be alternatively explained by fumarate hydratase and malate dehydrogenase activity on

Figure 6. Oxidative injury or impairment of the antioxidant capacity in glucose-6-phosphate dehydrogenase-deficient red blood cells enhances purine deamination. (A) Exposure of human RBC to prooxidant treatment with hypoxanthine plus xanthine dehydrogenase (XDH) for up to 6 h enhances AMP deamination to IMP accompanied by accumulation of hypoxanthine. (B) Human RBC from G6PD-deficient donors (Mediterranean variant, <10% residual activity of the enzyme) were stored for up to 42 days, showing significantly higher levels of hypoxanthine throughout the whole storage period (analyses were performed on whole transfusates: cells + supernatants) (median + ranges for RBC from G6PD donors are plotted as a solid red line and light red area, respectively). In (B), all data points shown on the x axis were tested and interpolated with third order polynomial curves (not assuming linear evolution of hypoxanthine accumulation during storage).

haematologica | 2018; 103(2)

367

T. Nemkov et al.

oxaloacetate derived from transamination of aspartate, reactions possible in mature erythrocytes.47 Although heavy aspartate consumption was faster in hypoxic RBC, the rates of 13C-fumarate generation were not sufficiently higher than in normoxic controls to explain the observed increases in AMP/IMP ratios in hypoxic RBC (e.g., only ~26-30% of fumarate was labeled at 6 h after incubation with heavy aspartate in both normoxic and hypoxic groups). In parallel, human RBC were incubated for 6 h with 13C5-adenosine to determine the rate of purine deamination (Figure 7B), demonstrating significantly decreased rates of IMP generation in hypoxic RBC. Analogously, refrigerated storage of mouse RBC under hypoxic conditions prevented purine deamination, as demonstrated by 13 C5-adenosine tracing experiments (Figure 7C). Finally,

incubating mouse RBC with deoxycoformycin, an adenosine and AMPD inhibitor, mimicked the hypoxic phenotype in terms of purine deamination and increased AMP/IMP ratios (Figure 7D). Thus, these results suggest that hypoxanthine accumulation, in this context, is primarily due to purine deamination.

Discussion Recently, we and others identified hypoxanthine19,28 as a potential metabolic marker of the RBC storage lesion with potential clinical implications. Indeed, circulating hypoxanthine can be readily converted to xanthine and urate by reactions that generate hydrogen peroxide.48 For exam-

Figure 7. Tracing experiments reveal hypoxic inhibition of purine deamination, rather than hypoxic increases in purine salvage. (A, B) Human RBC were incubated with (A) 13C1-aspartate to determine the rate of fumarate generation (salvage) and with (B) 13C5-adenosine to determine the rate of AMP deamination. (C) 13C5-adenosine tracing experiments demonstrated that in vitro storage of mouse RBC under hypoxic conditions prevents purine deamination. (D) Incubation of mouse RBC with deoxycoformycin, an adenosine and AMP deaminase inhibitor, increases AMP/IMP ratios.

368

haematologica | 2018; 103(2)

Hypoxanthine in stored RBC

ple, ischemic accumulation of plasma hypoxanthine can fuel pro-oxidant reactions following reperfusion.49 For this reason, rejuvenated RBC need to be washed prior to transfusion to avoid excess infusion of inosine and its metabolic product, hypoxanthine.50 By quantifying RBC and supernatant hypoxanthine levels during storage in AS-3, we confirmed and extended previous observations made in RBC stored in SAGM16,19 concerning the role of this metabolite as a reliable marker of the storage lesion. Thus, we correlated pre-transfusion in vitro hypoxanthine levels in mouse and human RBC with PTR determinations in vivo; the latter is one of the key US Food and Drug Administration’s criteria for approving novel RBC storage methods.24 Because RBC must circulate to perform their therapeutic function, we believe that – provided the evidence in mice and preliminary findings in humans reported here are further prospectively validated - this observation may support the clinical relevance of hypoxanthine as a potential predictor of transfusion outcomes. In addition, despite intrinsic metabolic differences in purine catabolism between rodents and humans (e.g., the former express functional uricase51 which may contribute to explaining the minor inter-species difference in hypoxanthine levels in cells and supernatants observed here), hypoxanthine levels in vitro correlated negatively with PTR in vivo in both mice and - to a lesser, albeit significant, extent - humans, further documenting the relevance of animal models in transfusion medicine research.52 During the last few decades, retrospective clinical studies,53 coupled with improved mechanistic understanding of the RBC storage lesion,2 have prompted the field of transfusion medicine to question the safety and efficacy of stored RBC. Recent prospective clinical trial evidence reassured the field about the non-inferiority of the current standard of care when compared to transfusion of fresh units.20 Nonetheless, transfusion of RBC stored for more than 5 weeks significantly increases circulating non-transferrin-bound iron levels,21 potentially increasing the risk of complications in certain categories of recipients.22 Omics markers of the storage lesion were recently identified,28,29 with the goal of benchmarking novel strategies or additive solutions to improve RBC storage quality. However, the disconnect between the well-established metabolic storage abnormalities and the reassuring clinical trial evidence prompted the field to question the relevance of metabolic markers of RBC storage age with respect to transfusion outcomes, a critical issue extensively discussed by many key opinion leaders.54 For example, over the past five decades, biochemical studies clearly documented that impairment of RBC energy and redox homeostasis (especially regarding levels of ATP and reduced glutathione) negatively affects RBC recovery in vivo,55 along with oxygen transport and delivery. Interestingly RBC levels of ATP and glutathione (as well as levels of at least 24 metabolites correlating with those of RBC ATP) are heritable traits,56 like hemolysis in vitro, although hemolysis does not correlate with pre-transfusion levels of these metabolites.57 Thus, it has been argued that these metabolites do not accurately predict transfusion outcomes likely because, as historically appreciated for 2,3-DPG,58,59 their levels are restored by 50% within the first 4 h after transfusion and return to normal by 72 h. However, these rates may not be sufficient to restore optimal tissue oxygenation in critically ill recipients who require massive transfusion, even though tissue oxygenation and systemic acidosis are effihaematologica | 2018; 103(2)

ciently restored by current resuscitation strategies. Mechanistically (Figure 8), we propose that refrigerated RBC storage promotes, among many cascades of events,2–5 the activation of AMPD3 (the RBC-specific isoform of AMPD, and the only one identified here using deep proteomics). AMPD3 can be activated by increases in intracellular ROS44 and calcium,60,61 along with decreases in intracellular pH;62,63 in contrast, increases in 2,3-DPG downregulate AMPD3 activity.63 Notably, refrigerated RBC storage, especially after storage day 14, leads to increases in ROS and intracellular calcium, and decreases in pH and 2,3DPG,2,6 all of which can combine to activate AMPD3, as our steady state and metabolic flux results suggest. This concept is supported by the positive effects of hypoxic storage, including decreased purine deamination leading to decreased hypoxanthine levels. Interestingly, AMPD3 activation shortens RBC lifespan64 by converting AMP into IMP, thereby contributing to adenine nucleotide dysregulation in RBC from normal individuals and those with sickle cell anemia. Conversely, the human enzymopathy of AMPD3 deficiency does not produce any relevant phenotype except for inducing increased levels of RBC ATP and ADP, thereby promoting a hypometabolic state with decreased hemoglobin oxygen affinity.65,66 Being at the crossroads between energy generation and redox metabolism, AMPD3 represents an ideal target for modulating RBC metabolism during refrigerated storage. Here we show that oxygen saturation and oxidative stress during refrigerated storage modulate purine deamination in human and mouse RBC in vivo and in vitro. This is relevant because of the unexpected widespread distribution of oxygen saturation of freshly donated human RBC within 8 h after donation and routine normoxic processing, suggesting that oxygen levels may represent a relevant variable that can be controlled to enhance the uniformity of stored blood.35 In addition, RBC from human G6PD-deficient donors - characterized by increased basal and storage-dependent oxidative stress67 - produce higher levels of hypoxanthine during refrigerated storage as compared to

Figure 8. Proposed mechanism of the effect of hypoxia on the purine salvage pathway. RBC storage or oxidative stress promotes activation of RBC AMPD3, which in turn catalyzes purine deamination. This phenomenon is in part counteracted by RBC exposure to hypoxia in vivo or ex vivo, which phenocopies the pharmacological inhibition of purine deaminases.

369

T. Nemkov et al.

controls. G6PD activity in stored human RBC was reported to decline in SAGM additives,67,68 though no notable changes in G6PD activity69 or PPP fluxes, as determined by [1,2,3-13C3]glucose tracing experiments,9 were noted in AS-3 stored RBC. In addition, although human G6PD activity is linked to gender (i.e., the G6PD gene is on chromosome X) and decreases with age,70 correlations between donor age or gender with transfusion outcomes are controversial.71,72 Notably, diamide treatment of G6PD-deficient RBC induced increases in AMP, IMP and hypoxanthine,73 consistent with our results with human G6PD-deficient RBC.67 In human G6PD-deficient RBC, increases in AMP were explained by the inability to reduce oxidized glutathione using PPP-derived NADPH, which, in turn, required increases in ATP-consuming de novo GSH synthesis.73 Tracing experiments with 13C1-aspartate and 13C5-adenosine show that hypoxic storage of RBC induces decreased AMPD3 activity, rather than increased purine salvage. The deep proteomic analyses identified traces of ADSS and ASL, and residual activity of the salvage pathway in mature RBC, thereby extending prior findings.43,46 Pharmacological inhibition of purine deaminase activity in normoxic RBC using deoxycoformycin mimicked the benefits of hypoxic storage, suggesting that manipulating the purinergic signaling axis using novel storage additives may improve RBC storage quality by enhancing energy production and redox homeostasis,74 similar to the metabolic adaptations of RBC to high altitude hypoxia.75 Finally, it is worth noting that, historically, circulating hypoxanthine levels (not RBC levels) have been suggested as a marker of hypoxia.49 Plasma hypoxanthine, under hypoxic conditions (e.g. high altitude or ischemia), may affect the severity of the reperfusion injury during its conversion to urate, a reaction that simultaneously generates pro-oxidant hydrogen peroxide.49 Similarly, accumulation of dicarboxylates and excess activation of salvage reactions in response to constrained oxygen availability were identified as mediators of oxidative reperfusion injury in mitochondria-proficient cells.76 The apparent disconnect between the present findings and the literature is reconciled by the observation that hypoxanthine levels in mitochondria-free RBC are not necessarily related to plasma levels of this metabolite (especially in the platelet and white blood cell-filtered environment of a packed RBC

References 1. Ellingson KD, Sapiano MR, Haass KA, et al. Continued decline in blood collection and transfusion in the United States-2015. Transfusion. 2017;57(Suppl 2):1588–1598. 2. D’Alessandro A, Kriebardis AG, Rinalducci S, et al. An update on red blood cell storage lesions, as gleaned through biochemistry and omics technologies. Transfusion. 2015;55(1):205–219. 3. Zimring JC. Widening our gaze of red blood storage haze: a role for metabolomics. Transfusion. 2015;55(6):1139–1142. 4. Hod EA, Zhang N, Sokol SA, et al. Transfusion of red blood cells after pro-

370

5. 6.

unit); for example, this is observed in underwater mammals, such as dolphins, which are constantly exposed to prolonged periods of hypoxia.77 Thus, hypoxanthine levels in isolated RBC (as in stored RBC units), or in circulating RBC after exposure to high altitude, may not directly correlate with plasma hypoxanthine levels.

Conclusion Hypoxanthine is a marker of the RBC storage lesion in mice and humans in vitro; in addition, it correlates with PTR results in vivo in mice and, preliminarily, in humans though additional validation in larger cohorts including poor “recoverers” will be necessary. Storage-dependent increases in cytosolic and supernatant purine deamination during human and mouse RBC storage are ameliorated by hypoxia. In addition, oxidative stress, induced by exogenously added pro-oxidants or in response to PPP shutdown in G6PD-deficient human RBC, promotes purine deamination and increases intracellular levels of hypoxanthine. Pharmacological inhibition and metabolomics experiments with stable isotope tracers suggest that decreased AMPD3 activity provides a potential mechanistic explanation of the benefits (i.e., decreased purine deamination) associated with RBC exposure to hypoxia in vivo and in vitro. Future studies are needed to investigate whether the correlation between pre-transfusion levels of hypoxanthine in vitro and PTR results in vivo could be explained by the role that this metabolite plays as a substrate for generating hydrogen peroxide in the circulatory system of the transfusion recipient. Acknowledgments Research reported in this publication was supported in part by funds from the National Blood Foundation Early Career grant 2016 (ADA), the Boettcher Webb-Waring Biomedical Research Award – Early Career grant (ADA), the University of Colorado Comprehensive Cancer Center Core Support (P30 CA04693417 to KCH and ADA), and grants from the National Institutes of Health: P50GM049222, T32GM008315, and P50GM049222 (KCH, ADA), R01GM113838 (ADo), and R01HL115557 (SLS). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

longed storage produces harmful effects that are mediated by iron and inflammation. Blood. 2010;115(21):4284–4292. Klein HG. The red cell storage lesion(s): of dogs and men. Blood Transfus. 2017;15 (2):107–111. D’Alessandro A, D’Amici GM, Vaglio S, Zolla L. Time-course investigation of SAGMstored leukocyte-filtered red bood cell concentrates: from metabolism to proteomics. Haematologica. 2012;97(1):107–115. Bardyn M, Tissot J-D, Prudent M. Oxidative stress and antioxidant defenses during blood processing and storage of erythrocyte concentrates. Transfus Clin Biol. 2017. pii: S1246-7820(17)30489-5. Wither M, Dzieciatkowska M, Nemkov T,

et al. Hemoglobin oxidation at functional amino acid residues during routine storage of red blood cells. Transfusion. 2016;56(2): 421–426. 9. Reisz JA, Wither MJ, Dzieciatkowska M, et al. Oxidative modifications of glyceraldehyde 3-phosphate dehydrogenase regulate metabolic reprogramming of stored red blood cells. Blood. 2016;128(12):e32-42. 10. Harper VM, Oh JY, Stapley R, et al. Peroxiredoxin-2 recycling is inhibited during erythrocyte storage. Antioxid Redox Signal. 2015;22(4):294–307. 11. Rinalducci S, D’Amici GM, Blasi B, et al. Peroxiredoxin-2 as a candidate biomarker to test oxidative stress levels of stored red blood cells under blood bank conditions.

haematologica | 2018; 103(2)

Hypoxanthine in stored RBC Transfusion. 2011;51(7):1439–1449. 12. Fu X, Felcyn JR, Zimring JC. Bioactive lipids are generated to micromolar levels during RBC storage, even in leukoreduced units. Blood. 2015;126(23):2344–2344. 13. Silliman CC, Moore EE, Kelher MR, et al. Identification of lipids that accumulate during the routine storage of prestorage leukoreduced red blood cells and cause acute lung injury. Transfusion. 2011;51(12):2549–2554. 14. D’Alessandro A, Nemkov T, Yoshida T, et al. Citrate metabolism in red blood cells stored in additive solution-3. Transfusion. 2017;57(2):325–336. 15. Sut C, Tariket S, Chou ML, et al. Duration of red blood cell storage and inflammatory marker generation. Blood Transfus. 2017;15(2):145–152. 16. Bordbar A, Johansson PI, Paglia G, et al. Identified metabolic signature for assessing red blood cell unit quality is associated with endothelial damage markers and clinical outcomes. Transfusion. 2016;56(4): 852–862. 17. D’Alessandro A, Nemkov T, Kelher M, et al. Routine storage of red blood cell (RBC) units in additive solution-3: a comprehensive investigation of the RBC metabolome. Transfusion. 2015;55(6):1155–1168. 18. Roback JD, Josephson CD, Waller EK, et al. Metabolomics of AS-1 RBC storage. Transfus Med Rev. 2014;28(2):41–55. 19. Pertinhez TA, Casali E, Lindner L, et al. Biochemical assessment of red blood cells during storage by 1H nuclear magnetic resonance spectroscopy. Identification of a biomarker of their level of protection against oxidative stress. Blood Transfus. 2014;12 (4):548–556. 20. Belpulsi D, Spitalnik SL, Hod EA. The controversy over the age of blood: what do the clinical trials really teach us? Blood Transfus. 2017;15(2):112–115. 21. Rapido F, Brittenham GM, Bandyopadhyay S, et al. Prolonged red cell storage before transfusion increases extravascular hemolysis. J. Clin. Invest. 2017;127(1):375–382. 22. Goel R, Johnson DJ, Scott AV, et al. Red blood cells stored 35 days or more are associated with adverse outcomes in high-risk patients. Transfusion. 2016;56(7):1690– 1698. 23. Mays JA, Hess JR. Modelling the effects of blood component storage lesions on the quality of haemostatic resuscitation in massive transfusion for trauma. Blood Transfus. 2017;15(2):153–157. 24. Dumont LJ, AuBuchon JP. Evaluation of proposed FDA criteria for the evaluation of radiolabeled red cell recovery trials. Transfusion. 2008;48(6):1053–1060. 25. Spitalnik SL, Triulzi D, Devine DV, et al. 2015 Proceedings of the National Heart, Lung, and Blood Institute’s State of the Science in Transfusion Medicine symposium. Transfusion. 2015;55(9):2282–2290. 26. Hess JR, Rugg N, Knapp AD, et al. Successful storage of RBCs for 10 weeks in a new additive solution. Transfusion. 2000;40(8):1012– 1016. 27. Yoshida T, Shevkoplyas SS. Anaerobic storage of red blood cells. Blood Transfus. 2010;8(4):220–236. 28. Paglia G, D’Alessandro A, Rolfsson Ó, et al. Biomarkers defining the metabolic age of red blood cells during cold storage. Blood. 2016;128(13):e43-50. 29. D’Alessandro A, Nemkov T, Reisz J, et al. Omics markers of the red cell storage lesion and metabolic linkage. Blood Transfus. 2017;15(2):137–144.

haematologica | 2018; 103(2)

30. Bardyn M, Rappaz B, Jaferzadeh K, et al. Red blood cells ageing markers: a multiparametric analysis. Blood Transfus. 2017;15(3):239–248. 31. Casali E, Berni P, Spisni A, Baricchi R, Pertinhez TA. Hypoxanthine: a new paradigm to interpret the origin of transfusion toxicity. Blood Transfus. 2015;14(6):555– 556. 32. Gilson CR, Kraus TS, Hod EA, et al. A novel mouse model of red blood cell storage and posttransfusion in vivo survival. Transfusion. 2009;49(8):1546–1553. 33. Goldman N, Chen M, Fujita T, et al. Adenosine A1 receptors mediate local antinociceptive effects of acupuncture. Nat Neurosci. 2010;13(7):883–888. 34. de Wolski K, Fu X, Dumont LJ, et al. Metabolic pathways that correlate with post-transfusion circulation of stored murine red blood cells. Haematologica. 2016;101(5): 578–586. 35. Yoshida T, Blair A, D’Alessandro A, et al. Enhancing uniformity and overall quality of red cell concentrate with anaerobic storage. Blood Transfus. 2017;15(2):172–181. 36. D’Alessandro A, Nemkov T, Sun K, et al. AltitudeOmics: red blood cell metabolic adaptation to high altitude hypoxia. J. Proteome Res. 2016;15(10):3883–3895. 37. Baskurt OK, Temiz A, Meiselman HJ. Effect of superoxide anions on red blood cell rheologic properties. Free Radic Biol Med. 1998;24(1):102–110. 38. D’Alessandro A, Dzieciatkowska M, Nemkov T, Hansen KC. Red blood cell proteomics update: is there more to discover? Blood Transfus. 2017;15(2):182–187. 39. Dzieciatkowska M, Hill R, Hansen KC. GeLC-MS/MS analysis of complex protein mixtures. Methods Mol Biol. 2014;1156:53– 66. 40. Batth TS, Francavilla C, Olsen JV. Off-line high-pH reversed-phase fractionation for indepth phosphoproteomics. J Proteome Res. 2014;13(12):6176–6186. 41. Nemkov T, Hansen KC, Dumont LJ, D’Alessandro A. Metabolomics in transfusion medicine. Transfusion. 2016;56(4):980– 993. 42. Nemkov T, Hansen KC, D’Alessandro A. A three-minute method for high-throughput quantitative metabolomics and quantitative tracing experiments of central carbon and nitrogen pathways. Rapid Commun Mass Spectrom. 2017;31(8):663–673. 43. Eiler J, Worthington F. The catabolism of the purine nucleotides: I. the relation to glycolysis in the blood of the rabbit. J Biol Chem. 1938;123:655. 44. Tavazzi B, Amorini AM, Fazzina G, et al. Oxidative stress induces impairment of human erythrocyte energy metabolism through the oxygen radical-mediated direct activation of AMP-deaminase. J Biol Chem. 2001;276(51):48083–48092. 45. Zimring JC, Smith N, Stowell SR, et al. Strain-specific red blood cell storage, metabolism, and eicosanoid generation in a mouse model. Transfusion. 2014;54(1):137–148. 46. Lowy BA, Dorfman B-Z. Adenylosuccinase activity in human and rabbit erythrocyte lysates. J Biol Chem. 1970;245(12):3043– 3046. 47. Bordbar A, Yurkovich JT, Paglia G, et al. Elucidating dynamic metabolic physiology through network integration of quantitative time-course metabolomics. Sci Rep. 2017;7:46249. 48. Kelley EE, Khoo NKH, Hundley NJ, et al. Hydrogen peroxide is the major oxidant

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

product of xanthine oxidase. Free Radic Biol Med. 2010;48(4):493–498. Saugstad OD. Hypoxanthine as an indicator of hypoxia: its role in health and disease through free radical production. Pediatr Res. 1988;23(2):143–150. D’Alessandro A, Gray AD, Szczepiorkowski ZM, et al. Red blood cell metabolic responses to refrigerated storage, rejuvenation and frozen storage. Transfusion. 2017;57(4): 1019–1030. Blais EM, Rawls KD, Dougherty BV, et al. Reconciled rat and human metabolic networks for comparative toxicogenomics and biomarker predictions. Nat Commun. 2017;8:14250. Zimring JC, Spitalnik SL. On the appropriate use and interpretation of animal models in transfusion medicine research. Transfusion. 2013;53(10):2334–2339. Koch CG, Li L, Sessler DI, et al. Duration of red-cell storage and complications after cardiac surgery. N Engl J Med. 2008;358(12): 1229–1239. D’Alessandro A, Liumbruno GM. Red blood cell storage and clinical outcomes: new insights. Blood Transfus. 2017;15(2):101– 103. van Wijk R, van Solinge WW. The energyless red blood cell is lost: erythrocyte enzyme abnormalities of glycolysis. Blood. 2005;106(13):4034–4042. van ’t Erve TJ, Wagner BA, Martin SM, et al. The heritability of metabolite concentrations in stored human red blood cells. Transfusion. 2014;54(8):2055–2063. Van ’t Erve TJ, Wagner BA, Martin SM, et al. The heritability of hemolysis in stored human red blood cells. Transfusion. 2015;55(6):1178–1185. Valeri CR, Hirsch NM. Restoration in vivo of erythrocyte adenosine triphosphate, 2,3diphosphoglycerate, potassium ion, and sodium ion concentrations following the transfusion of acid-citrate-dextrose-stored human red blood cells. Transl Res. 1969;73 (5):722–733. Beutler E, Wood L. The in vivo regeneration of red cell 2,3 diphosphoglyceric acid (DPG) after transfusion of stored blood. J Lab Clin. Med. 1969;74(2):300–304. Mahnke DK, Sabina RL. Calcium activates erythrocyte AMP deaminase [isoform E (AMPD3)] through a protein-protein interaction between calmodulin and the N-terminal domain of the AMPD3 polypeptide. Biochemistry (Mosc.). 2005;44(14):5551– 5559. Sabina RL, Wandersee NJ, Hillery CA. Ca2+-CaM activation of AMP deaminase contributes to adenine nucleotide dysregulation and phosphatidylserine externalization in human sickle erythrocytes. Br J Haematol. 2009;144(3):434–445. Mahnke-Zizelman DK, Sabina RL. N-terminal sequence and distal histidine residues are responsible for pH-regulated cytoplasmic membrane binding of human AMP deaminase isoform E. J Biol Chem. 2002;277 (45):42654–42662. Dudley GA, Terjung RL. Influence of acidosis on AMP deaminase activity in contracting fast-twitch muscle. Am J Physiol. 1985;248(1 Pt 1):C43-50. Hortle E, Nijagal B, Bauer DC, et al. Adenosine monophosphate deaminase 3 activation shortens erythrocyte half-life and provides malaria resistance in mice. Blood. 2016;128(9):1290–1301. Daniels IS, Iii WGO, Nath V, Zhao Z, Lee CC. AMP deaminase 3 deficiency enhanced

371

T. Nemkov et al.

66.

67.

68.

69.

372

5 -AMP induction of hypometabolism. PLOS One. 2013;8(9): e75418. O’Brien WG III, Berka V, Tsai A-L, Zhao Z, Lee CC. CD73 and AMPD3 deficiency enhance metabolic performance via erythrocyte ATP that decreases hemoglobin oxygen affinity. Sci. Rep. 2015;5:13147. Tzounakas VL, Kriebardis AG, Georgatzakou HT, et al. Glucose 6-phosphate dehydrogenase deficient subjects may be better “storers” than donors of red blood cells. Free Radic Biol Med. 2016;96:152–165. Peters AL, van Bruggen R, de Korte D, Van Noorden CJ, Vlaar AP. Glucose-6-phosphate dehydrogenase activity decreases during storage of leukoreduced red blood cells. Transfusion. 2016;56(2):427–432. Francis RO, Jhang J, Hendrickson JE, et al. Frequency of glucose-6-phosphate dehydrogenase-deficient red blood cell units in a

70.

71.

72.

73.

metropolitan transfusion service. Transfusion. 2013;53(3):606–611. Rodgers GP, Lichtman HC, Sheff MF. Red blood cell glucose-6-phosphate dehydrogenase activity in aged humans. J Am Geriatr Soc. 1983;31(1):8–11. Chassé M, Tinmouth A, English SW, et al. Association of blood donor age and sex with recipient survival after red blood cell transfusion. JAMA Intern Med. 2016;176(9):1307– 1314. Edgren G, Ullum H, Rostgaard K, et al. Association of donor age and sex with survival of patients receiving transfusions. JAMA Intern Med. 2017;177(6):854-860. Tang H, Ho H, Wu P, et al. Inability to maintain GSH pool in G6PD-deficient red cells causes futile AMPK sctivation and irreversible metabolic disturbance. Antioxid. Redox Signal. 2015;22(9):744–759.

74. Sun K, D’Alessandro A, Xia Y. Purinergic control of red blood cell metabolism: novel strategies to improve red cell storage quality. Blood Transfus. 2017;15(6):535-542. 75. Liu H, Zhang Y, Wu H, et al. Beneficial role of erythrocyte adenosine A2B receptormediated AMP-activated protein kinase activation in high-altitude hypoxia. Circulation. 2016;134(5):405–421. 76. Chouchani ET, Pell VR, Gaude E, et al. Ischaemic accumulation of succinate controls reperfusion injury through mitochondrial ROS. Nature. 2014;515(7527):431– 435. 77. López-Cruz RI, Crocker DE, Gaxiola-Robles R, et al. Plasma hypoxanthine-guanine phosphoribosyl transferase activity in bottlenose dolphins contributes to avoiding accumulation of non-recyclable purines. Front Physiol. 2016;7:213.

haematologica | 2018; 103(2)

Obituary

Obituary Remembering Professor Felice Gavosto (February 16, 1921 – December 11, 2017) Felice Gavosto, Professor Emeritus of Medicine at the University of Turin, Italy and co-founder of the Institute of Cancer Research and Treatment, Candiolo (Turin) passed away peacefully in the morning of December 11, 2017 aged 96. He had a passion for science and his mantra was “patients are better treated in the institutions which privilege research”. Soon after taking his MD degree in Turin he decided “to do science” and worked both in New York at the Memorial Hospital and in Belgium at the University of Brussels. He then returned to Turin where he became professor of internal medicine and developed a profound interest for hematology, well before hematology became an independent discipline. He was one of the pioneers of the studies on cell kinetics in acute leukemia and his results were published in Nature in the early 1960s when Italian science was certainly not used to such levels of recognition. Indeed, between 1953 and 1969 Gavosto published eight papers in Nature, being the first author of seven of them. His papers are still quoted nowadays. He also became a close friend, better as he used to say - a friendly competitor of legends in hematology such as Bayard Clarkson, Sven-Aage Killmann and Theodore Fliedner, and was well acquainted with iconic figures such as Eugene Cronkite and Albert Bruce Sabin. Felice Gavosto represented the gateway to science for many young investigators who later became prominent figures in hematology. He always showed a sheer enthusiasm for new developments in basic and clinical investigations. Notwithstanding his passion for blood diseases, he decided to remain professor of internal medicine, convinced as he was that the clinical approach to patients must be based upon the capacity to fully grasp all different aspects of internal medicine: this is the message that he conveyed to his students while teaching. He had an important public role in promoting the growth of science and especially of hematologic science in Italy. In this respect, he held prominent positions: he was the chairman of a nationwide project focused on the “Control of Malignant Cell Growth” supported by the National Research Council (CNR), served as member of the Italian

haematologica | 2018; 103(2)

Felice Gavosto, February 16, 1921 - December 11, 2017.

Association for Cancer Research (AIRC) for several years and was one of the founders of the Italian Association against Leukemia (AIL). As he strongly believed that it was essential to couple science with the clinics, it was the field of oncology that in the last part of his career stimulated his enthusiastic activity and led him to champion the possibility of building, in Turin, a Cancer Institute developed along the best international lines of research and treatment. He was successful: this Institute - together with the many people who had the privilege to work with him and to grow as scientists following his example of rigorous methodology and intellectual integrity - are his legacy. Gavosto was an eminent personality who had a full, active and exciting life, and will be missed not only by his family but also by his numerous collaborators and mentees. Federico Caligaris Cappio, Massimo Aglietta, Clara Camaschella, Giuseppe Saglio and Robin Foà doi: 10.3324/haematol.2018.188268

373

haematologica Journal of the European Hematology Association Published by the Ferrata Storti Foundation

Ancient Greek

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

Scientific Latin

aÂma [haima] = blood a·matow [haimatos] = of blood lÒgow [logos]= reasoning

Scientific Latin

haematologicus (adjective) = related to blood

Modern English

haematologica (adjective, plural and neuter, used as a noun) = hematological subjects The oldest hematology journal, publishing the newest research results. 2016 JCR impact factor = 7.702

Haematologica, as the journal of the European Hematology Association (EHA), aims not only to serve the scientific community, but also to promote European cultural identify.